1 2017-03-14 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/80030
4 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
6 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
8 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
9 gcc_fallthrough() instead of __attribute__((fallthrough));
11 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13 * doc/gcc.texi: Remove "up" link to (DIR).
14 * doc/gccint.texi: Ditto.
16 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18 * doc/install.texi (Specific) <avr>: Remove reference to
21 2017-03-13 Jeff Law <law@redhat.com>
23 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24 attribute rather than comments.
26 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27 match_scratch operand is highest.
29 2017-03-13 Martin Liska <mliska@suse.cz>
32 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33 is a CHKP clone, use original declaration.
35 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
37 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
38 (arc_conditional_register_usage): Use a different allocation order
39 when optimizing for size.
40 * common/config/arc/arc-common.c (arc_option_optimization_table):
41 Section anchors default on when optimizing for size.
43 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
45 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
47 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
49 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
50 * config/arc/arc.md (cpu_facility): Add cd variant.
51 (*movqi_insn): Add code density variant.
52 (*movhi_insn): Likewise.
53 (*movqi_insn): Likewise.
54 (*addsi3_mixed): Likewise.
55 (subsi3_insn): Likewise.
57 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
59 * config/arc/arc.md (movsi_cond_exec): Update constraint.
61 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
63 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
64 expressions with MINUS and UNARY ops.
66 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
69 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
71 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
72 between vec_select and vector argument.
73 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
74 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
75 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
76 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
77 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
78 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
80 2017-03-13 Richard Biener <rguenther@suse.de>
83 * params.def (vect-max-peeling-for-alignment): Fix typo.
85 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
87 * doc/install.texi (Specific) <mips-*-*>: Remove description of
88 issue that only occurred with binutils below 2.18.
90 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
92 * doc/install.texi (Specific) <cris-axis-elf>: No longer
93 refer to binutils 2.11/2.12 minimum.
95 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
97 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
98 ftp.kernel.org and simplify binutils requirement.
100 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
102 * doc/invoke.texi (Warning Options): Fix spelling of link-time
104 (Optimize Options): Ditto. Also remove redundancy.
106 2017-03-10 David Malcolm <dmalcolm@redhat.com>
109 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
111 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
112 to G_ to avoid double translation.
114 2017-03-10 David Malcolm <dmalcolm@redhat.com>
117 * auto-profile.c (get_combined_location): Convert leading
118 character of diagnostics to lower case and remove trailing period.
119 (read_profile): Likewise for various diagnostics.
120 * config/arm/arm.c (arm_option_override): Remove trailing period
121 from various diagnostics.
122 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
123 (msp430_expand_delay_cycles): Likewise.
125 2017-03-10 David Malcolm <dmalcolm@redhat.com>
128 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
129 full command-line argument, rather than just "str".
130 (aarch64_validate_march): Likewise.
131 (aarch64_validate_mtune): Likewise.
133 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
135 PR rtl-optimization/78911
136 * lra-assigns.c (must_not_spill_p): New function.
139 2017-03-10 Jakub Jelinek <jakub@redhat.com>
141 PR tree-optimization/79981
142 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
143 ATOMIC_COMPARE_EXCHANGE ifn result.
144 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
145 IFN_ATOMIC_COMPARE_EXCHANGE.
147 2017-03-10 David Malcolm <dmalcolm@redhat.com>
150 * opts.c (parse_sanitizer_options): Add missing question mark to
151 "did you mean" message.
153 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
155 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
157 (VMULEUH_UNS): Likewise.
158 (VMULOUB_UNS): Likewise.
159 (VMULOUH_UNS): Likewise.
160 * config/rs6000/rs6000.c (builtin_function_type): Remove
161 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
163 2017-03-10 David Malcolm <dmalcolm@redhat.com>
166 * read-rtl-function.c (function_reader::read_rtx_operand): Update
167 x with result of extra_parsing_for_operand_code_0.
168 (function_reader::extra_parsing_for_operand_code_0): Convert
169 return type from void to rtx, returning x. When reading
170 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
171 larger size containing struct block_symbol.
173 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
176 -mfloat128-hardware without -m64.
178 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
181 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
182 entries to the case statement that marks unsigned arguments to
183 overloaded functions.
185 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
187 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
188 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
190 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
193 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
194 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
196 2017-03-10 Martin Liska <mliska@suse.cz>
200 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
201 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
204 2017-03-10 Olivier Hainque <hainque@adacore.com>
206 * tree-switch-conversion (array_value_type): Start by resetting
207 candidate type to it's main variant.
209 2017-03-10 Jakub Jelinek <jakub@redhat.com>
211 PR rtl-optimization/79909
212 * combine.c (try_combine): Use simplify_replace_rtx on individual
213 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
214 of the whole CALL_INSN_FUNCTION_USAGE.
216 PR tree-optimization/79972
217 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
218 get_range_info on SSA_NAMEs. Formatting fixes.
220 2017-03-10 Richard Biener <rguenther@suse.de>
221 Jakub Jelinek <jakub@redhat.com>
223 PR tree-optimization/77975
224 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
226 (get_val_for): For constant x return it. Formatting fix.
227 (loop_niter_by_eval): Avoid pointless looping if the next iteration
228 would use the same bases as the current one.
230 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
232 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
233 instead of vec_select for V1TImode.
234 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
236 (VSX_LE_128): Add V1TI to this mode iterator.
237 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
238 (*vsx_le_perm_store_<mode>): Likewise.
239 (pre-reload splitter for VSX stores): Likewise.
240 (post-reload splitter for VSX stores): Likewise.
241 (*vsx_xxpermdi2_le_<mode>): Likewise.
242 (*vsx_lxvd2x2_le_<mode>): Likewise.
243 (*vsx_stxvd2x2_le_<mode>): Likewise.
245 2017-03-09 Michael Eager <eager@eagercon.com>
247 Correct failures with --enable-checking=yes,rtl.
249 * config/microblaze/microblaze.c (microblaze_expand_shift):
250 Replace GET_CODE test with CONST_INT_P and INTVAL test with
252 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
253 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
255 2017-03-09 Richard Biener <rguenther@suse.de>
257 PR tree-optimization/79977
258 * graphite-scop-detection.c (scop_detection::merge_sese):
259 Handle the case of extra exits to blocks dominating the entry.
261 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
263 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
266 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
268 PR rtl-optimization/79949
269 * lra-constraints.c (process_alt_operands): Check memory when
270 trying to predict a cycle. Print about the overall increase.
272 2017-03-09 Richard Biener <rguenther@suse.de>
275 * gimple-expr.c (useless_type_conversion_p): Preserve
276 TYPE_SATURATING for fixed-point types.
278 2017-03-09 Richard Biener <rguenther@suse.de>
281 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
282 alignment of BLKmode params.
284 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
287 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
288 (VALL_NO_V2Q): Likewise.
290 * config/aarch64/aarch64-simd.md
291 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
293 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
294 VALL_NO_V2Q mode iterator.
295 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
297 2017-03-09 Martin Liska <mliska@suse.cz>
299 PR tree-optimization/79631
300 * tree-chkp-opt.c (chkp_is_constant_addr): Call
301 tree_int_cst_sign_bit just for INTEGER constants.
303 2017-03-09 Martin Liska <mliska@suse.cz>
307 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
310 2017-03-09 Marek Polacek <polacek@redhat.com>
313 * tree.c (inchash::add_expr): Handle TREE_VEC.
315 2017-03-09 Martin Liska <mliska@suse.cz>
318 (chkp_narrow_size_and_offset): New function.
319 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
320 (void chkp_parse_bit_field_ref): New function.
321 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
322 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
324 2017-03-09 Martin Liska <mliska@suse.cz>
327 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
328 (chkp_find_bounds_1): Remove gcc_unreachable.
330 2017-03-09 Jakub Jelinek <jakub@redhat.com>
333 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
334 BUILT_IN_SYNC*, determine the access type from the size suffix and
335 always build a MEM_REF with that type. Handle forgotten
336 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
339 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
340 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
341 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
342 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
343 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
344 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
345 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
346 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
347 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
348 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
349 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
350 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
351 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
352 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
353 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
354 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
355 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
356 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
357 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
358 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
359 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
360 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
361 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
362 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
363 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
364 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
365 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
366 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
367 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
368 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
369 definitions outside of __OPTIMIZE__ guarded section.
372 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
373 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
374 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
375 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
378 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
380 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
381 ("vfenez<mode>"): Add missing constraints.
383 2017-03-08 Martin Sebor <msebor@redhat.com>
386 * config/nds32/nds32.c (nds32_option_override):
388 2017-03-08 Jakub Jelinek <jakub@redhat.com>
391 * gimplify.c (gimplify_omp_for): Replace index var in outer
392 taskloop statement with an artificial variable and add
393 OMP_CLAUSE_PRIVATE clause for it.
395 2017-03-08 Richard Biener <rguenther@suse.de>
397 PR tree-optimization/79955
398 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
399 for accesses that are completely outside of the variable.
401 2017-03-08 Andrew Haley <aph@redhat.com>
403 PR tree-optimization/79943
404 * tree-ssa-loop-split.c (compute_new_first_bound): When
405 calculating the new upper bound, (END-BEG) should be added, not
408 2017-03-08 Jakub Jelinek <jakub@redhat.com>
410 * config/avr/avr.md (setmemhi): Make sure match_dup
411 operand number comes before match_scratch.
413 2017-03-08 Richard Biener <rguenther@suse.de>
415 PR tree-optimization/79920
416 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
417 with ncopies == 1 to ...
418 (vect_transform_slp_perm_load): ... here. Properly compute
419 all element loads by iterating VF times over the group. Do
420 not handle ncopies (computed in a broken way) in
421 vect_create_mask_and_perm.
423 2017-03-08 Jakub Jelinek <jakub@redhat.com>
426 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
427 is a uniform vector, use uniform_vector_p return value instead of
428 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
430 2017-03-07 Marek Polacek <polacek@redhat.com>
433 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
434 (alloca_call_type): Likewise.
436 2017-03-07 Martin Liska <mliska@suse.cz>
438 * gcov.c (process_args): Put comment to correct location.
440 2017-03-07 Martin Liska <mliska@suse.cz>
443 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
444 Use array_at_struct_end_p instead of DECL_CHAIN (field).
445 (chkp_narrow_bounds_for_field): Likewise.
446 (chkp_parse_array_and_component_ref): Pass one more argument to
449 2017-03-07 Richard Biener <rguenther@suse.de>
451 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
454 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
456 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
457 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
459 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
462 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
463 to end of description.
464 (PARAM_MAX_STORES_TO_MERGE): Likewise.
466 2017-03-07 Jakub Jelinek <jakub@redhat.com>
468 PR rtl-optimization/79901
469 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
471 (*avx512f_<code><mode>3<mask_name>): ... this.
472 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
473 iterator instead of VI8_AVX2_AVX512BW.
475 PR rtl-optimization/79901
476 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
477 min/max expander, expand it using expand_vec_cond_expr.
480 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
483 2017-03-06 Jakub Jelinek <jakub@redhat.com>
486 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
487 to void * for PCH reasons.
488 * dwarf2out.c (output_loc_operands, output_die): Cast
489 v.val_vec.array to unsigned char *.
491 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
494 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
497 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
499 PR rtl-optimization/79571
500 * lra-constraints.c (process_alt_operands): Calculate static
501 reject and subtract it from overall when only addresses will be
504 2017-03-06 Julia Koval <julia.koval@intel.com>
507 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
508 incoming stack boundary to 128 for 64-bit targets.
510 2017-03-06 Richard Biener <rguenther@suse.de>
512 PR tree-optimization/79894
513 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
514 to NULL after folding it.
516 2017-03-06 Richard Biener <rguenther@suse.de>
518 PR tree-optimization/79824
519 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
520 check disabling peeling for gaps.
522 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
524 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
525 attributes): Document gettimeofday.
527 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
529 * config/s390/s390.c (s390_option_override_internal): Set
530 PARAM_MIN_VECT_LOOP_BOUND
532 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
534 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
535 * config/s390/s390.md: Likewise.
537 2017-03-06 Jakub Jelinek <jakub@redhat.com>
540 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
541 (<avx2_avx512>_perm<mode>): Rename to ...
542 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
544 (<avx512>_perm<mode>_mask): Rename to ...
545 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
547 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
548 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
549 instead of VI8F_256_512.
550 (avx512f_perm<mode>): New define_expand.
551 (avx512f_perm<mode>_mask): Likewise.
552 (avx512f_perm<mode>_1<mask_name>): New define_insn.
553 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
555 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
557 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
558 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
560 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
562 2017-03-06 Martin Liska <mliska@suse.cz>
565 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
566 when having a SSA NAME w/o VAR_DECL assigned to it.
568 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
570 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
571 msa_dpsub_<su>_d): Fix MODE for vec_select.
573 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
575 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
577 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
579 2017-03-06 Richard Biener <rguenther@suse.de>
581 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
582 * plugin.c (register_plugin_info): Likewise.
583 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
585 2017-03-05 Jakub Jelinek <jakub@redhat.com>
587 * config/i386/sse.md (sse_storehps, sse_storelps,
588 avx_<castmode><avxsizesuffix>_<castmode>,
589 avx512f_<castmode><avxsizesuffix>_<castmode>,
590 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
591 in condition that at least one operand is not a MEM.
593 2017-03-03 Jakub Jelinek <jakub@redhat.com>
596 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
597 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
599 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
600 gimple_call_nothrow_p flag based on whether original builtin can throw.
601 If it can, emit following stmts on the fallthrough edge.
602 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
603 don't create new bb if inserting just debug stmts on the edge, try to
604 insert them on the fallthru bb or just reset debug stmts.
606 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
609 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
610 restore recog_data (including the operand rtxes inside it) around
611 the call to get_insn_template.
613 2017-03-03 Martin Sebor <msebor@redhat.com>
615 PR tree-optimization/79699
616 * context.c (context::~context): Free MPFR caches to avoid
617 a memory leak on program exit.
619 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
621 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
622 Use wide_int::ulow () instead of .elt (0).
624 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
626 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
627 (*pushxf): Limit oF constraint to 32bit targets and add oC
628 constraint for 64bit targets.
629 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
630 (*pushdf): Change rmF constraint to rmC.
632 2017-03-03 Martin Liska <mliska@suse.cz>
634 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
635 Remove unused variable.
637 2017-03-03 Jakub Jelinek <jakub@redhat.com>
640 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
641 is a memory operand, increase num_memory.
642 (ix86_expand_args_builtin): Likewise.
644 2017-03-03 Jan Hubicka <jh@suse.cz>
647 * ipa-devirt.c (maybe_record_node): Properly handle
648 __cxa_pure_virtual visibility.
650 2017-03-03 Martin Liska <mliska@suse.cz>
652 PR tree-optimization/79803
653 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
655 (pass_loop_prefetch::execute): Disabled optimization if an
656 assumption about L1 cache size is not met.
658 2017-03-03 Martin Liska <mliska@suse.cz>
660 PR rtl-optimization/79574
661 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
662 (hash_scan_set): Likewise.
663 (dump_hash_table): Likewise.
664 (hoist_code): Likewise.
666 2017-03-03 Richard Biener <rguenther@suse.de>
668 * fixed-value.c (fixed_from_string): Restore use of elt (1)
669 in place of uhigh ().
670 (fixed_convert_from_real): Likewise.
672 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
675 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
677 2017-03-03 Richard Biener <rguenther@suse.de>
680 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
681 TYPE_OVERFLOW_UNDEFINED check.
683 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
685 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
687 (vector_ae_<mode>_p): Likewise.
688 (vector_nez_<mode>_p): Likewise.
689 (vector_ne_v2di_p): Likewise.
690 (vector_ae_v2di_p): Likewise.
691 (vector_ne_<mode>_p): Likewise.
692 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
694 (vsx_tsqrt<mode>2_fe): Likewise.
696 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
699 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
701 2017-03-02 Jakub Jelinek <jakub@redhat.com>
703 PR rtl-optimization/79780
704 * cprop.c (one_cprop_pass): When second and further conditional trap
705 in a single basic block is turned into an unconditional trap, turn it
706 into a deleted note to avoid RTL verification failures.
708 2017-03-02 Richard Biener <rguenther@suse.de>
710 * fold-const.c (const_binop): Use ulow () instead of elt (0).
712 2017-03-02 Richard Biener <rguenther@suse.de>
714 PR tree-optimization/79345
716 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
717 param and abort the walk, returning -1 if it is hit.
718 (walk_aliased_vdefs): Take a limit param and pass it on.
719 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
720 defaulting to 0 and return a signed int.
721 * tree-ssa-uninit.c (struct check_defs_data): New struct.
722 (check_defs): New helper.
723 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
724 about uninitialized memory.
725 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
726 bogus uninitialized warning.
727 (fixed_convert_from_real): Likewise.
729 2017-03-02 Bin Cheng <bin.cheng@arm.com>
731 PR tree-optimization/66768
732 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
733 iv_use if base object can't be determined.
735 2017-03-02 Jakub Jelinek <jakub@redhat.com>
737 PR tree-optimization/79345
738 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
739 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
740 (get_pattern_stats): Initialize it.
741 * genemit.c (gen_expand): Verify match_scratch numbers come after
742 match_operand/match_dup numbers.
743 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
744 match_scratch numbers.
745 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
747 * config/s390/s390.md (trunctdsd2): Likewise.
749 2017-03-02 Richard Biener <rguenther@suse.de>
751 * wide-int.h (wide_int_storage::operator=): Implement in terms
754 2017-03-02 Richard Biener <rguenther@suse.de>
756 PR tree-optimization/79777
757 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
758 the to insert expression to sth existing.
760 2017-03-01 Martin Sebor <msebor@redhat.com>
763 * gimple-ssa-sprintf.c
764 (directive::known_width_and_precision): New function.
765 (format_integer): Use it.
766 (get_mpfr_format_length): Consider the full range of precision
767 when computing %g output with the # flag. Set the likely byte
768 count to 3 rather than 1 when precision is indeterminate.
769 (format_floating): Correct the lower bound of precision.
771 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
773 * doc/invoke.texi: Document default code model for 64-bit Linux.
775 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
778 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
779 udiv rather than div since input pattern is unsigned.
781 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
783 * config/i386/i386.c (print_reg): Warn for values of
784 unsupported size in integer register.
786 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
789 * config/rs6000/predicates.md (current_file_function_operand): Do
790 not allow self calls to be local if the function is replaceable.
792 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
795 * config/rs6000/altivec.h (vec_ctz and others): Change the
796 preprocessor macro that controls conditional compilation from
797 _ARCH_PWR9 to __POWER9_VECTOR__.
798 (vec_all_ne): Change parameterization of __altivec_scalar_pred
799 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
800 control (instead of _ARCH_PWR9 control) so that template
801 definition uses power9-specific function.
802 (vec_any_eq): Likewise.
803 (vec_all_ne): Change macro definition to use a power9-specific
804 expansion under #ifdef __POWER9_VECTOR__ control (instead of
806 (vec_any_eq) Likewise.
807 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
808 expansion for CMPNEF to remove support for xvcmpnesp instruction.
809 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
810 support for xvcmpnedp instruction.
811 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
812 macro expansion so that Power9 implementation of vec_all_ne does
813 not use the AltiVec predicate framework.
814 (VCMPNEH_P): Likewise.
815 (VCMPNEW_P): Likewise.
816 (VCMPNED_P): Likewise.
817 (VCMPNEFP_P): Likewise.
818 (VCMPNEDP_P): Likewise.
819 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
820 implementation of vec_any_eq to not use AltiVec predicate
822 (VCMPAEH_P): Likewise.
823 (VCMPAEW_P): Likewise.
824 (VCMPAED_P): Likewise.
825 (VCMPAEFP_P): Likewise.
826 (VCMPAEDP_P): Likewise.
827 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
828 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
829 not use the AltiVec predicate framework.
830 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
831 of vec_any_eq to not use AltiVec predicate framework.
832 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
833 support for predefined __POWER9_VECTOR__ macro to indicate that
834 Power9 instruction selection is enabled.
835 (altivec_overloaded_builtins): Remove extraneous
836 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
837 function argument types RS6000_BTI_bool_V16QI and
838 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
839 entry for overloaded function argument types RS6000_BTI_bool_V4SI
840 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
841 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
842 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
843 Power9 for implementations of vec_cmpne. Change the signature for
844 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
845 (representing vec_all_ne) to remove the previously described first
846 argument of type RS6000_BTI_INTSI, as this was an artifact of
847 reliance on the AltiVec predicate framework, which is no longer
848 used in the implementation of these functions. Add
849 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
850 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
851 since, unlike the AltiVec predicate framework implementation, we
852 do not share function descriptors between vec_alle and vec_anyeq.
853 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
854 set of modes that receive special treatment even when
855 TARGET_P9_VECTOR is true. The special treatment emits code that
856 does not depend on Power9 instructions.
857 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
858 define_expand to not rely on AltiVec predicate framework.
859 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
861 (vector_ne_v2di_p): Change this define_expand to not rely on
862 AltiVec predicate framework.
863 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
865 (vector_ne_<mode>_p): Change this define_expand to not rely on
866 AltiVec predicate framework.
867 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
869 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
870 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
872 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
873 define_insn pattern because the xvcmpne<VSs>. instruction is not
875 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
876 instruction is not supported.
878 2017-03-01 Jakub Jelinek <jakub@redhat.com>
880 * config/nvptx/nvptx.c: Include intl.h.
882 2017-03-01 Martin Jambor <mjambor@suse.cz>
885 * ipa-prop.h (ipa_bits): Removed field known.
886 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
887 to pointers. Adjusted their comments to warn about their sharing.
888 (ipcp_transformation_summary): Change bits to a vector of pointers.
889 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
890 (ipa_get_ipa_bits_for_value): Declare.
891 * tree-vrp.h (value_range): Mark as GTY((for_user)).
892 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
893 (ipa_bits_hash_table): Likewise.
894 (ipa_vr_ggc_hash_traits): Likewise.
895 (ipa_vr_hash_table): Likewise.
896 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
897 being pointers and vr_known being removed.
898 (ipa_set_jf_unknown): Likewise.
899 (ipa_get_ipa_bits_for_value): New function.
900 (ipa_set_jfunc_bits): Likewise.
901 (ipa_get_value_range): New overloaded functions.
902 (ipa_set_jfunc_vr): Likewise.
903 (ipa_compute_jump_functions_for_edge): Use the above functions to
904 construct bits and vr parts of jump functions.
905 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
906 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
908 (ipcp_grow_transformations_if_necessary): Also allocate
909 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
911 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
912 them. Fix too long lines.
913 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
914 vr_known being removed.
915 (ipa_read_jump_function): Use new setter functions to construct bits
916 and vr parts of jump functions or set them to NULL.
917 (write_ipcp_transformation_info): Adjust for bits being pointers.
918 (read_ipcp_transformation_info): Likewise.
919 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
921 Include gt-ipa-prop.h.
922 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
924 (ipcp_store_bits_results): Likewise.
925 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
926 Do not write to existing jump functions but use a temporary instead.
928 2017-03-01 Jakub Jelinek <jakub@redhat.com>
931 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
932 attempt to use its first operand as BIT_FIELD_REF base.
934 2017-03-01 Richard Biener <rguenther@suse.de>
937 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
938 interpolating formula in wrapping arithmetic.
939 (chrec_apply): Convert chrec_evaluate return value to wanted type.
941 2017-03-01 Jakub Jelinek <jakub@redhat.com>
943 PR tree-optimization/79734
944 * tree-vect-generic.c (expand_vector_condition): Optimize
945 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
946 Handle VEC_COND_EXPR where comparison has different inner width from
949 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
951 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
952 markup, and similar issues. Remove @opindex entries for things
953 that aren't options. Add missing -mmpy-option entries.
955 2017-02-28 Jakub Jelinek <jakub@redhat.com>
957 PR tree-optimization/79737
958 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
959 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
960 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
961 instead of byte_size. Formatting fix.
962 (shift_bytes_in_array_right): Formatting fix.
964 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
967 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
968 condition on optimize for the leaf function test.
970 2017-02-28 Martin Liska <mliska@suse.cz>
973 * read-rtl-function.c (function_reader::handle_unknown_directive):
974 Bail out when one uses -flto.
976 2017-02-28 Martin Liska <mliska@suse.cz>
978 * common.opt: Replace space with tabular for options of <number>
980 * config/i386/i386.opt: Show <number> value for
981 -mlarge-data-threshold.
982 * opts.c (print_filtered_help): Do not display number in hexadecimal
985 2017-02-28 Martin Liska <mliska@suse.cz>
987 * common.opt: Fix --help=option -Q for options which are of
990 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
992 * config/i386/i386.c (print_reg): Error out for values
993 of 8-bit size in invalid integer register.
995 2017-02-28 Martin Sebor <msebor@redhat.com>
997 PR tree-optimization/79691
998 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1000 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1003 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1004 gcc_unreachable with output_operand_lossage.
1006 2017-02-28 Richard Biener <rguenther@suse.de>
1008 PR tree-optimization/79740
1009 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1011 (visit_nary_op): Insert the nary into the hashtable if we
1012 pattern-matched sth.
1013 * tree-ssa-pre.c (eliminate_insert): Robustify.
1015 2017-02-28 Richard Biener <rguenther@suse.de>
1018 * fold-const.c (decode_field_reference): Reject out-of-bound
1021 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1023 * config/i386/i386.c: Include intl.h.
1024 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1025 instead of just cond ? "..." : "...".
1026 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1027 * coverage.c (read_counts_file): Likewise.
1028 * omp-offload.c: Include intl.h.
1029 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1030 of just cond ? "..." : "...".
1031 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1032 of just cond ? "..." : "...".
1034 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1037 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1039 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1040 'tune for' CPU name.
1041 * config/arm/arm-cpu-data.h: Regenerated.
1043 2017-02-28 Richard Biener <rguenther@suse.de>
1045 PR tree-optimization/79732
1046 * tree-inline.c (expand_call_inline): Do not shadow var.
1048 2017-02-28 Richard Biener <rguenther@suse.de>
1050 PR tree-optimization/79723
1051 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1052 address-space properly.
1054 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1056 * doc/optinfo.texi (Optimization groups): Fix option used for
1058 * doc/invoke.texi (-fopt-info): Document "omp".
1059 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1060 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1061 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1062 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1063 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1065 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1067 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1068 OPTGROUP_OMP with "omp".
1070 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1073 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1074 for arithmetic shift of unsigned V2DI.
1076 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1078 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1079 arc/linux.h headers.
1080 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1081 (LINK_SPEC): Likewise.
1082 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1083 (EXTRA_SPECS): Likewise.
1084 (STARTFILE_SPEC): Likewise.
1085 (ENDFILE_SPEC): Likewise.
1086 (LIB_SPEC): Likewise.
1087 (TARGET_SDATA_DEFAULT): Likewise.
1088 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1089 (MULTILIB_DEFAULTS): Likewise.
1090 (DWARF2_UNWIND_INFO): Likewise.
1091 * config/arc/big.h: New file.
1092 * config/arc/elf.h: Likewise.
1093 * config/arc/linux.h: Likewise.
1094 * config/arc/t-uClibc: Remove.
1096 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1098 PR tree-optimization/77536
1099 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1100 (tree_transform_and_unroll_loop): Use above function to compute the
1101 estimated niter of unrolled loop and use it when scaling profile.
1102 Also use count info rather than frequency if it's non-zero.
1103 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1104 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1105 (vect_transform_loop): Call above function.
1107 2017-02-27 Richard Biener <rguenther@suse.de>
1109 PR tree-optimization/45397
1110 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1111 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1112 (visit_nary_op): Add pattern matching for CSEing sign-changed
1113 or truncated operations with wider ones.
1115 2017-02-27 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/79690
1118 * tree-vect-stmts.c (vectorizable_store): Use vector type
1119 built from the DR with address-space.
1121 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1123 * doc/invoke.texi (Optimize Options): Refine the description
1124 of asan-use-after-return.
1126 2017-02-25 Alan Modra <amodra@gmail.com>
1128 PR rtl-optimization/79584
1129 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1130 base, not ad->base_term, the reg within base. Remove assertion
1131 that ad->base == ad->base_term. Replace gen_int_mode using
1132 bogus mode with const0_rtx.
1134 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1137 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1138 FMA_EXPR like tcc_binary or tcc_unary.
1140 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1143 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1145 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1146 (output_loc_operands): Handle DW_OP_call_ref and
1147 DW_OP_GNU_variable_value.
1148 (struct variable_value_struct): New type.
1149 (struct variable_value_hasher): Likewise.
1150 (variable_value_hash): New variable.
1151 (string_types): Remove.
1152 (copy_loc_descr): New function.
1153 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1154 (prepend_loc_descr_to_each): New function.
1155 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1156 instead of add_loc_descr_to_each if the first argument is single
1157 location list and the second has multiple.
1158 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1159 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1160 when looking for variable value which doesn't have other location info.
1161 (loc_list_from_tree): Formatting fix.
1162 (gen_array_type_die): Simplify DW_AT_string_length handling.
1163 (adjust_string_types): Remove.
1164 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1165 string_types. Call resolve_variable_values.
1166 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1167 (resolve_addr_in_expr): Likewise. Add A argument.
1168 (copy_deref_exprloc): Remove deref argument. Adjust for the
1169 original expression being DW_OP_GNU_variable_value with optionally
1170 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1171 optionally after it.
1172 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1173 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1174 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1175 (variable_value_hasher::hash, variable_value_hasher::equal): New
1177 (resolve_variable_value_in_expr, resolve_variable_value,
1178 resolve_variable_values, note_variable_value_in_expr,
1179 note_variable_value): New functions.
1180 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1183 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1186 * opts.h (handle_generated_option): Add GENERATED_P argument.
1187 * opts-common.c (handle_option): Adjust function comment.
1188 (handle_generated_option): Add GENERATED_P argument, pass it to
1190 (control_warning_option): Pass false to handle_generated_option
1192 * opts.c (maybe_default_option): Pass true to handle_generated_option
1194 * optc-gen.awk: Likewise.
1196 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1198 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1199 a REG, look at the REG it is a SUBREG of.
1200 (splitter for cmpeqsi_t): Ditto.
1202 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1204 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1205 the special USEs with the pattern of the insn, not the insn itself.
1207 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1210 * doc/invoke.texi: Document -mload-store-pairs.
1212 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1213 Sandra Loosemore <sandra@codesourcery.com>
1215 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1216 argument isn't a CONST_INT.
1217 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1218 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1219 (nios2_validate_compare): Bypass alternate compare logic if *op2
1221 (ldstwm_operation_p): Return false if first_base is not a REG or
1222 if first_offset is not a CONST_INT.
1224 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1226 * config/cris/cris.md: Use correct operand in a define_peephole2.
1228 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1230 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1232 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1234 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1235 this_insn if it is an INSN or JUMP_INSN.
1236 (force_offsettable): Look at base, not at addr.
1237 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1238 on things that aren't necessarily CONST_INTs.
1240 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1242 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1243 -mfpmath=sse is the default also for x86-32 targets with SSE2
1244 instruction set when @option{-ffast-math} is enabled
1246 2017-02-24 Jeff Law <law@redhat.com>
1248 PR rtl-optimizatoin/79286
1249 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1252 2017-02-24 Richard Biener <rguenther@suse.de>
1254 PR tree-optimization/79389
1255 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1258 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1260 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1261 function comment to reflect reality.
1262 (loop_exits_before_overflow): Fix typo in function description.
1264 2017-02-24 Richard Biener <rguenther@suse.de>
1266 PR tree-optimization/79389
1267 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1268 properly that a threading opportunity exists. Detect conditional
1269 copy/constant propagation opportunities.
1271 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1273 * config/visium/visium.md (type): Add trap.
1274 (b): New mode attribute.
1275 (*btst): Rename into...
1276 (*btst<mode>): ...this and adjust.
1277 (*cbranchsi4_btst_insn): Rename into...
1278 (*cbranch<mode>4_btst_insn): ...this and adjust.
1279 (trap): New define_insn.
1281 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1283 PR tree-optimization/79389
1284 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1285 (noce_reversed_cond_code): New function.
1286 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1287 reversed_comparison_code. Formatting fix.
1288 (noce_try_store_flag): Test rev_cond != NULL in addition to
1289 reversed_comparison_code.
1290 (noce_try_store_flag_constants): Likewise.
1291 (noce_try_store_flag_mask): Likewise.
1292 (noce_try_addcc): Use rev_cond if non-NULL instead of
1293 reversed_comparison_code.
1294 (noce_try_cmove_arith): Likewise. Formatting fixes.
1295 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1296 (noce_find_if_block): Initialize rev_cond.
1297 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1298 instead of false as last argument never attempt to reverse it
1301 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1303 PR tree-optimization/79663
1304 * tree-predcom.c (combine_chains): Process refs in reverse order
1305 only for ZERO length chains, and add explaining comment.
1307 2017-02-23 Jeff Law <law@redhat.com>
1309 PR tree-optimization/79578
1310 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1311 in call to operand_equal_p.
1313 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1316 * config/i386/cpuid.h: Fix another undefined behavior.
1318 2017-02-23 Richard Biener <rguenther@suse.de>
1320 PR tree-optimization/79683
1321 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1322 vector types for data-refs.
1324 2017-02-23 Martin Liska <mliska@suse.cz>
1326 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1328 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1331 * internal-fn.c (get_range_pos_neg): Moved to ...
1332 * tree.c (get_range_pos_neg): ... here. No longer static.
1333 * tree.h (get_range_pos_neg): New prototype.
1334 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1335 are known to be in between 0 and signed maximum inclusive, try to
1336 expand both unsigned and signed divmod and use the cheaper one from
1339 2017-02-22 Jeff Law <law@redhat.com>
1341 PR tree-optimization/79578
1342 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1343 to compare base operands.
1345 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1348 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1349 gpc_reg_operand instead of fpr_reg_operand.
1351 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1353 * config/mips/mips.c (mips_return_in_memory): Force FP
1354 vector types to be returned in memory for o32 ABI.
1356 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1358 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1359 instead of DW_TAG_member for static data member declarations and don't
1360 set no_linkage_name for static inline data members.
1361 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1364 2017-02-22 Martin Liska <mliska@suse.cz>
1366 * doc/invoke.texi: Replace inequality signs with square brackets
1369 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1371 PR tree-optimization/68644
1372 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1374 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1377 * lra-constraints.c (simplify_operand_subreg): Handle
1378 WORD_REGISTER_OPERATIONS targets.
1380 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1383 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1384 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1385 elimination by swapping fld*.
1387 2017-02-22 Richard Biener <rguenther@suse.de>
1389 PR tree-optimization/79673
1390 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1391 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1392 irrelevant address-space qualifiers and avoiding a
1393 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1395 2017-02-22 Richard Biener <rguenther@suse.de>
1397 PR tree-optimization/79666
1398 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1399 to not symbolically negate if that may introduce undefined
1402 2017-02-22 Martin Liska <mliska@suse.cz>
1405 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1406 * data-streamer-out.c (streamer_write_gcov_count_stream):
1408 * value-prof.c (stream_out_histogram_value): Make assert more
1409 precise based on type of counter.
1411 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1414 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1415 Use nonimmediate_operand instead of memory_operand for operand 1.
1416 (float-extend standard_x87sse_constant_load splitter): Ditto.
1418 2017-02-21 Jeff Law <law@redhat.com>
1420 PR tree-optimization/79621
1421 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1422 blocks with edges to themselves.
1424 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1427 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1428 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1429 Use gimple_call_builtin_p.
1432 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1433 on temporarily removed DEBUG_INSNs.
1435 PR tree-optimization/79649
1436 * tree-loop-distribution.c (classify_partition): Give up on
1437 non-generic address space loads/stores.
1439 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1441 * doc/loop.texi (Loop manipulation): Remove nonexistent
1442 tree_ssa_loop_version from the documentation.
1443 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1445 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1448 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1449 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1450 * config/rs6000/rs6000.c: Include except.h.
1451 (rs6000_expand_split_stack_prologue): Call
1452 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1454 2017-02-21 Martin Jambor <mjambor@suse.cz>
1457 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1460 2017-02-21 Martin Jambor <mjambor@suse.cz>
1462 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1463 for backward compatibility only.
1464 * doc/invoke.texi (Option Summary): Remove all references to
1467 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1471 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1473 * lra-constraints.c (curr_insn_transform): Handle
1474 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1476 2017-02-21 Martin Liska <mliska@suse.cz>
1478 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1480 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1482 PR translation/79638
1483 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1485 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1488 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1489 (arm_function_ok_for_sibcall): Return false for an indirect call by
1490 descriptor if all the argument registers are used.
1491 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1492 alignment of the function.
1494 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1496 PR tree-optimization/61441
1497 * simplify-rtx.c (simplify_const_unary_operation): For
1498 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1499 the sNaN unmodified.
1501 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1503 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1504 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1505 instead of SYSTEM_HEADER_DIR.
1507 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1508 Martin Liška <mliska@suse.cz>
1510 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1511 Fix typos and grammar, use active voice, and clarify.
1513 2017-02-20 Marek Polacek <polacek@redhat.com>
1516 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1519 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1521 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1524 * config/i386/i386.c (ix86_expand_builtin): Handle
1525 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1526 ix86_builtins_isa[fcode].isa as a requirement of those
1527 flags and any other flag in the bitmask.
1528 (ix86_init_mmx_sse_builtins): Use 0 instead of
1529 ~OPTION_MASK_ISA_64BIT as mask.
1530 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1531 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1532 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1533 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1535 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1538 * lra-constraints.c (split_reg): Check requested split mode
1539 is supported by the register.
1541 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1543 * lra-constraints.c (simplify_operand_subreg): Remove early
1546 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1549 * lra-constraints.c (curr_insn_transform): Tighten condition
1550 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1552 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1555 * lra-constraints.c (curr_insn_transform): Handle
1556 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1558 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1561 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1563 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1565 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1568 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1571 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1573 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1576 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1578 * final.c (last_columnnum, override_columnnum): New variables.
1579 (final_start_function): Set last_columnnum, pass it to begin_prologue
1580 hook and pass 0 to dwarf2out_begin_prologue.
1581 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1582 to source_line debug hook.
1583 (notice_source_line): Compute last_columnnum and for debug_column_info
1584 return true on column changes.
1585 * debug.h (struct gcc_debug_hooks): Add column argument to
1586 source_line and begin_prologue hooks.
1587 (debug_nothing_int_charstar_int_bool): Remove prototype.
1588 (debug_nothing_int_int_charstar,
1589 debug_nothing_int_int_charstar_int_bool): New prototypes.
1590 (dwarf2out_begin_prologue): Add column argument.
1591 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1592 begin_prologue hooks.
1593 (debug_nothing_int_charstar_int_bool): Remove.
1594 (debug_nothing_int_int_charstar,
1595 debug_nothing_int_int_charstar_int_bool): New functions.
1596 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1597 through to dwarf2out_source_line.
1598 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1599 (dwarf2out_source_line): Add column argument, emit it if requested.
1600 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1602 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1603 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1604 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1605 through to dwarf2out_begin_prologue.
1606 (vmsdbgout_source_line): Add column argument, pass it through to
1607 dwarf2out_source_line.
1608 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1609 dbxout_source_line caller.
1610 (dbxout_source_line): Add column argument.
1612 * common.opt (gno-column-info, gcolumn-info): New options.
1613 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1614 (check_die): Also test for multiple DW_AT_decl_column attributes.
1615 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1616 DW_AT_decl_column if requested.
1617 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1619 (gen_variable_die): Likewise.
1620 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1621 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1624 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1625 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1626 (ix86_handle_option): Handle OPT_m3dnowa.
1627 * doc/invoke.texi (-m3dnowa): Document.
1628 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1629 -m3dnowa instead of -m3dnow -march=athlon.
1632 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1633 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1635 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1638 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1639 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1640 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1641 generator for vsx_xxpermdi_<mode>_be.
1642 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1643 force big-endian semantics.
1644 (vsx_xxpermdi_<mode>_be): New define_expand with same
1645 implementation as previous version of vsx_xxpermdi_<mode>.
1647 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1649 PR tree-optimization/79327
1650 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1651 variable, its initialization and use.
1653 2017-02-17 Julia Koval <julia.koval@intel.com>
1655 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1656 (OPTION_MASK_ISA_PKU_UNSET): New.
1657 (ix86_handle_option): Handle -mrdpid.
1658 * config/i386/cpuid.h (bit_RDPID): New.
1659 * config/i386/driver-i386.c (host_detect_local_cpu):
1660 Detect RDPID feature.
1661 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1662 * config/i386/i386-c.c (ix86_target_macros_internal):
1664 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1665 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1666 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1667 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1668 * config/i386/i386.md (define_insn "rdpid"): New.
1669 * config/i386/i386.opt Add -mrdpid.
1670 * config/i386/immintrin.h (_rdpid_u32): New.
1672 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1674 PR rtl-optimization/79541
1675 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1676 instead of transforming it into USE.
1678 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1680 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1681 If HONOR_SNANS (SFmode) force the input to a register.
1682 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1683 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1684 an frsp or similar insn.
1686 2017-02-17 Martin Liska <mliska@suse.cz>
1688 PR rtl-optimization/79577
1689 * params.def (selsched-max-sched-times): Increase minimum to 1.
1691 2017-02-17 Martin Liska <mliska@suse.cz>
1693 PR rtl-optimization/79574
1694 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1696 2017-02-17 Martin Liska <mliska@suse.cz>
1698 PR tree-optimization/79529
1699 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1700 ssa_defined_default_def_p to handle cases which are implicitly
1702 * tree-ssa.c (ssa_defined_default_def_p): New function.
1703 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1704 which are implicitly defined.
1705 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1707 2017-02-17 Richard Biener <rguenther@suse.de>
1710 * params.def (max-ssa-name-query-depth): Limit to 10.
1712 2017-02-17 Richard Biener <rguenther@suse.de>
1714 PR tree-optimization/79552
1715 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1718 2017-02-17 Richard Biener <rguenther@suse.de>
1721 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1723 2017-02-17 Marek Polacek <polacek@redhat.com>
1726 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1727 (fold_negate_expr): New wrapper.
1729 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1731 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1732 Correct terminology and de-emphasize pre-standard behavior.
1734 2017-02-16 Alan Modra <amodra@gmail.com>
1736 PR rtl-optimization/79286
1737 * ira.c (def_dominates_uses): New function.
1738 (update_equiv_regs): Don't create an equivalence for insns that
1739 may trap where the register def does not dominate the use.
1741 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1743 PR rtl-optimization/78127
1744 * lra.c (lra): Call lra_eliminate before finish the loop after
1747 2017-02-16 Richard Biener <rguenther@suse.de>
1749 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1751 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1752 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1753 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1754 (isl_val_int_from_wi): New function.
1755 (extract_affine_gmp): Rename to ...
1756 (extract_affine_wi): ... this, take a widest_int.
1757 (extract_affine_int): Just wrap extract_affine_wi.
1758 (add_param_constraints): Use isl_val_int_from_wi.
1759 (add_loop_constraints): Likewise, and extract_affine_wi.
1761 2017-02-15 Jeff Law <law@redhat.com>
1764 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1765 ira_init_register_move_cost_if_necessary.
1767 2017-02-15 Martin Sebor <msebor@redhat.com>
1770 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1771 removed in a prior commit.
1773 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1775 PR tree-optimization/79347
1776 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1777 counters during peeling.
1779 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1781 * Makefile.in (site.exp): Remove "set ISLVER".
1783 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1786 * real.c (real_from_integer): Call real_convert even for decimal.
1788 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1791 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1793 2017-02-14 Andrew Pinski <apinski@cavium.com>
1795 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1796 cores and change the partno/implementer to be correct.
1797 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1798 the 'B" as the implementer.
1799 * config/aarch64/aarch64-tune.md: Regenerate.
1801 2017-02-14 Carl Love <cel@us.ibm.com>
1803 * config/rs6000/rs6000.c: Add case statement entry to make the
1804 xvcvuxdsp built-in argument unsigned.
1805 * config/rs6000/vsx.md: Fix the source and return operand types so they
1806 match the instruction definitions from the ISA document. Fix typo
1807 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1810 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1813 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1814 member early_clobber_alts.
1815 * lra-lives.c (reg_early_clobber_p): New.
1816 (process_bb_lives): Use it.
1817 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1818 (debug_operand_data): Initialize early_clobber_alts.
1819 (setup_operand_alternative): Set up early_clobber_alts.
1820 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1821 alternatives to new_insn_reg.
1822 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1824 (lra_update_insn_regno_info): Pass the new arg.
1826 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1829 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1830 (new_oacc_loop_raw): Don't clear already cleared fields.
1833 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1834 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1835 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1837 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1840 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1842 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1845 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1846 the extra instruction to the right place to store 128-bit constant
1849 2017-02-14 Martin Sebor <msebor@redhat.com>
1852 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1853 warning for strings of unknown length.
1855 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1857 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1859 2017-02-14 Jeff Law <law@redhat.com>
1862 * ira-costs.c (scan_one_insn): Initialize register move costs
1863 for pseudos seen in USE/CLOBBER insns.
1865 PR tree-optimization/79095
1866 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1867 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1868 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1869 if the operands are known to be not equal, then the resulting range
1871 (intersect_ranges): If the new range is ~[0,0] and the old range is
1872 wide, then prefer ~[0,0].
1873 * tree-vrp.c (overflow_comparison_p_1): New function.
1874 (overflow_comparison_p): New function.
1875 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1876 if NAME is used in an overflow test.
1877 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1878 overflow check that can be expressed as an equality test, then adjust
1879 ops to be that equality test.
1881 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1883 * config/s390/s390-builtin-types.def: Remove flags argument.
1884 * config/s390/s390.c (s390_init_builtins): Likewise.
1886 2017-02-14 Martin Liska <mliska@suse.cz>
1888 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1889 vector. Fix trailing white spaces.
1891 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1893 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1896 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1898 PR rtl-optimization/68664
1899 * config/arm/arm.c (arm_sched_can_speculate_insn):
1900 New function. Declare prototype.
1901 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1903 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1905 PR rtl-optimization/68664
1906 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1908 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1910 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1912 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1913 max skip bytes for function, loop and jump.
1915 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1917 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1918 ABS_EXPR for gimple dump.
1920 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1923 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1925 PR tree-optimization/79408
1926 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1927 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1928 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1929 also if rhs1 is INTEGER_CST.
1931 2017-02-14 Richard Biener <rguenther@suse.de>
1934 * tree-into-ssa.c (insert_phi_nodes): When the function can
1935 have abnormal edges rewrite SSA names with broken use-def
1936 dominance out of SSA and register them for PHI insertion.
1938 2017-02-13 Martin Sebor <msebor@redhat.com>
1941 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1942 clearing info.nowrite flag when snprintf size argument is a range.
1944 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1946 * cprop.c (cprop_jump): Add missing space in string literal.
1947 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1948 (get_constraint_for_component_ref): Likewise.
1949 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1950 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1951 * lra-constraints.c (process_alt_operands): Likewise.
1952 * ipa-inline.c (inline_small_functions): Likewise.
1953 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1954 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1955 * trans-mem.c (diagnose_tm_1_op): Likewise.
1956 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1957 (grid_parallel_clauses_gridifiable): Likewise.
1959 * config/nvptx/mkoffload.c (process): Add space in between
1962 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
1963 "MOD4_SSE_REGS" and "ALL_REGS".
1965 * spellcheck.c (test_data): Add , in between "foo" and "food".
1967 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1970 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
1971 boundary crossing check and subsequent code generation agree.
1973 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1975 * config/aarch64/aarch64.c (has_memory_op): Delete.
1976 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
1979 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1981 PR rtl-optimization/79388
1982 PR rtl-optimization/79450
1983 * combine.c (distribute_notes): When removing TEM_INSN for which
1984 corresponding dest has last value recorded, invalidate that last
1987 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1989 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
1990 of explicit '@'. Add missing assembly comment marker on branch costs
1993 2017-02-13 Nathan Sidwell <nathan@acm.org>
1995 * gengtype-lex.l (<in_struct>): Add '/'.
1997 2017-02-13 Martin Liska <mliska@suse.cz>
2000 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2002 2017-02-13 Richard Biener <rguenther@suse.de>
2004 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2006 * configure: Re-generate.
2007 * config.in: Likewise.
2008 * graphite-dependences.c: Simplify as if
2009 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2010 * graphite-isl-ast-to-gimple.c: Likewise.
2011 * graphite-optimize-isl.c: Likewise.
2012 * graphite-poly.c: Likewise.
2013 * graphite-sese-to-poly.c: Likewise.
2014 * graphite.h: Likewise.
2015 * toplev.c: Include isl/version.h and use isl_version () for
2016 printing the ISL version.
2017 * doc/install.texi: Update ISL requirement.
2019 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2021 * doc/standards.texi (Standards): Update reference to
2024 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2026 * doc/extend.texi (Named Address Spaces): sourceware.org now
2028 * doc/install.texi (Binaries): Ditto.
2031 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2033 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2034 standard terminology "stringize"/"stringizing" throughout.
2035 * doc/cppinternals.texi: Likewise.
2037 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2039 * doc/extend.texi: Fix some spelling mistakes and typos.
2040 * doc/invoke.texi: Likewise.
2042 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2045 * params.def (inline-min-speedup) Change from 10 to 8.
2047 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2049 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2052 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2055 * ipa-inline-analysis.c (get_minimal_bb): New function.
2056 (record_modified): Use it.
2057 (remap_edge_change_prob): Handle also ancestor functions.
2059 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2061 * doc/contrib.texi (Contributors): Remove broken link into
2062 the Mauve CVS repository.
2064 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2067 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2068 result computation whenever lhs doesn't have vector mode, not
2069 just when it has BLKmode.
2071 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2073 * doc/makefile.texi (profiledbootstrap): Refer to the
2074 installation instructions only in textual form.
2076 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2079 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2081 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2083 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2084 (Specific): Update mingw-w64 reference.
2086 (Specific): Remove broken link to Renesas RX processor.
2088 2017-02-10 Richard Biener <rguenther@suse.de>
2090 * toplev.c (process_options): Do not mention obsolete graphite
2091 options when printing sorry message about missing graphite support.
2092 Mention -floop-nest-optimize.
2094 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2096 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2097 (vtst_p16): Likewise.
2098 (vtstq_p8): Likewise.
2099 (vtstq_p16): Likewise.
2101 (vtstq_p64): Likewise.
2102 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2103 (vset_lane_p64): New.
2104 (vsetq_lane_p64): New.
2106 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2108 PR tree-optimization/79411
2109 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2110 stmt operands are SSA_NAMEs used in abnormal phis.
2111 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2114 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2117 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2120 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2122 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2124 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2126 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2130 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2131 not arbitrary TREE_CONSTANT.
2134 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2135 "omp declare target link" attribute unless is_global_var.
2136 * omp-offload.c (find_link_var_op): Likewise.
2138 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2139 Chung-Lin Tang <cltang@codesourcery.com>
2141 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2143 (gimplify_adjust_omp_clauses): Don't delete TILE.
2144 (gimplify_omp_for): Deal with TILE.
2145 * internal-fn.c (expand_GOACC_TILE): New function.
2146 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2148 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2149 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2151 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2152 avoid DIV for outermost collapse var.
2153 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2154 Remove out of date comments, fix whitespace.
2155 * omp-general.c (omp_extract_for_data): Deal with tiling.
2156 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2157 adjust OLF_DIM_BASE value.
2158 (struct omp_for_data): Add tiling field.
2159 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2160 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2161 for auto loops. Remove default auto determining, moved to
2162 oacc_loop_fixed_partitions.
2163 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2164 stmts, add e_mask field.
2165 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2166 (oacc_thread_numbers): Use oacc_dim_call.
2167 (oacc_xform_tile): New.
2168 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2169 (finish_oacc_loop): Adjust for ifns vector.
2170 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2171 add case for GOACC_TILE fns.
2172 (oacc_loop_xform_loop): Delete.
2173 (oacc_loop_process): Iterate over call list directly, and add
2174 handling for GOACC_TILE fns.
2175 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2177 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2178 vector partitioning to outer loops. Assign 2 partitions to loops
2179 when available. Add TILE handling.
2180 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2181 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2182 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2183 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2184 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2186 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2188 * configure.ac (ACX_BUGURL): Update.
2189 * configure: Regenerate.
2191 2017-02-09 Richard Biener <rguenther@suse.de>
2193 PR tree-optimization/69823
2194 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2195 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2197 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2199 * config/arc/arc-c.def: Add __NPS400__ definition.
2200 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2201 (TARGET_NPS400): Define.
2203 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2205 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2207 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2209 (arc_cpu_types): Fill the arch_info field with a pointer into the
2210 arc_arch_types table.
2211 (arc_selected_cpu): Declare.
2212 * config/arc/arc.c (arc_selected_cpu): Make global.
2213 (arc_selected_arch): Delete.
2214 (arc_base_cpu): Delete.
2215 (arc_override_options): Remove references to deleted variables,
2216 update access to arch information.
2217 (ARC_OPT): Update access to arch information.
2218 (ARC_OPTX): Likewise.
2219 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2220 (TARGET_ARC600): Update access to arch information.
2221 (TARGET_ARC601): Likewise.
2222 (TARGET_ARC700): Likewise.
2223 (TARGET_EM): Likewise.
2224 (TARGET_HS): Likewise.
2225 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2228 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2231 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2232 condition/operands for integer GE/LE/GEU/LEU operations.
2234 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2236 PR translation/79397
2237 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2240 2017-02-08 Martin Jambor <mjambor@suse.cz>
2243 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2244 whether allocation happened.
2245 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2246 nothing was allocated.
2248 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2250 PR tree-optimization/79408
2251 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2252 constant, but SSA_NAME with a known integer range, use the minimum
2253 of that range instead of op1 to determine if modulo can be replaced
2254 with its first operand.
2256 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2258 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2260 2017-02-08 Richard Biener <rguenther@suse.de>
2262 PR tree-optimization/71824
2263 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2264 Check all loops contained in the merged region.
2266 2017-02-07 Andrew Pinski <apinski@cavium.com>
2268 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2270 2017-02-07 Andrew Pinski <apinski@cavium.com>
2272 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2273 (thunderxt88): Likewise.
2274 (thunderxt81): Disable LSE and change v8.1 to v8.
2275 (thunderxt83): Likewise.
2277 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2278 Richard Biener <rguenther@suse.de>
2281 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2282 type from int to size_t.
2283 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2285 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2287 PR rtl-optimization/79386
2288 * cprop.c (bypass_conditional_jumps): Initialize
2289 bypass_last_basic_block already before splitting bbs after
2290 unconditional traps...
2291 (bypass_conditional_jumps): ... rather than here.
2294 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2295 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2296 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2297 fix -masm=intel patterns.
2299 2017-02-07 Richard Biener <rguenther@suse.de>
2301 PR tree-optimization/79256
2303 * builtins.c (get_object_alignment_2): Use min_align_of_type
2304 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2305 and ADJUST_FIELD_ALIGN.
2307 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2309 * doc/tm.texi: Regenerate.
2310 * stor-layout.c (layout_decl): Adjust.
2311 (update_alignment_for_field): Likewise.
2312 (place_field): Likewise.
2313 (min_align_of_type): Likewise.
2314 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2315 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2316 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2317 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2318 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2319 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2320 * config/i386/i386.c (x86_field_alignment): Likewise.
2321 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2322 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2323 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2324 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2325 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2326 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2330 2017-01-30 Richard Biener <rguenther@suse.de>
2332 PR tree-optimization/79256
2333 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2334 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2337 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2339 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2340 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2341 builtins to SImode and emit a zero-extend, if necessary.
2343 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2345 * docs/invoke.texi (RISC-V Options): Alphabetize.
2347 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2349 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2352 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2354 * config/riscv/riscv.c: New file.
2355 * gcc/common/config/riscv/riscv-common.c: Likewise.
2356 * config.gcc: Likewise.
2357 * config/riscv/constraints.md: Likewise.
2358 * config/riscv/elf.h: Likewise.
2359 * config/riscv/generic.md: Likewise.
2360 * config/riscv/linux.h: Likewise.
2361 * config/riscv/multilib-generator: Likewise.
2362 * config/riscv/peephole.md: Likewise.
2363 * config/riscv/pic.md: Likewise.
2364 * config/riscv/predicates.md: Likewise.
2365 * config/riscv/riscv-builtins.c: Likewise.
2366 * config/riscv/riscv-c.c: Likewise.
2367 * config/riscv/riscv-ftypes.def: Likewise.
2368 * config/riscv/riscv-modes.def: Likewise.
2369 * config/riscv/riscv-opts.h: Likewise.
2370 * config/riscv/riscv-protos.h: Likewise.
2371 * config/riscv/riscv.h: Likewise.
2372 * config/riscv/riscv.md: Likewise.
2373 * config/riscv/riscv.opt: Likewise.
2374 * config/riscv/sync.md: Likewise.
2375 * config/riscv/t-elf-multilib: Likewise.
2376 * config/riscv/t-linux: Likewise.
2377 * config/riscv/t-linux-multilib: Likewise.
2378 * config/riscv/t-riscv: Likewise.
2379 * configure.ac: Likewise.
2380 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2381 Waterman as RISC-V maintainers.
2382 * doc/install.texi: Add RISC-V entries.
2383 * doc/invoke.texi: Add RISC-V options section.
2384 * doc/md.texi: Add RISC-V constraints section.
2385 * configure: Regenerated.
2387 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2390 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2391 false values to be constant vectors with all 0 or all 1 bits set.
2392 (vcondu<mode><mode>): Likewise.
2393 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2395 (fpmask_comparison_operator): Update comment.
2396 (vecint_comparison_operator): New predicate.
2397 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2398 vector conditionals when the true and false values are constant
2399 vectors with all 0 bits or all 1 bits set.
2401 2017-02-06 Martin Sebor <msebor@redhat.com>
2403 PR tree-optimization/79376
2404 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2406 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2408 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2409 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2410 to simplify split condition.
2412 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2414 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2415 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2418 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2420 PR rtl-optimization/68664
2421 * target.def (can_speculate_insn): New hook.
2422 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2423 * doc/tm.texi: Regenerate.
2424 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2425 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2426 (rs6000_sched_can_speculate_insn): New function.
2428 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2430 PR tree-optimization/79284
2431 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2432 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2433 vectorizable_mask_load_store, vectorizable_operation,
2434 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2435 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2436 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2437 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2438 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2439 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2440 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2441 is_gimple_assign (stmt). Replace another such test with
2442 is_gimple_assign (stmt).
2444 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2447 * config/avr/avr.c (rtl-iter.h): Include it.
2448 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2449 (avr_legitimate_combined_insn): ...and implementation.
2451 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2453 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2454 * config/s390/s390.c (s390_const_operand_ok)
2455 (s390_canonicalize_comparison, s390_extract_part)
2456 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2457 (s390_contiguous_bitmask_p, s390_rtx_costs)
2458 (legitimize_pic_address): Likewise.
2459 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2460 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2461 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2462 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2463 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2465 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2467 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2468 REGNO($0) == REGNO($1).
2470 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2472 * config/s390/linux.h(SIZE_TYPE): Add comment.
2474 2017-02-06 Julian Brown <julian@codesourcery.com>
2475 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2476 Virendra Pathak <virendra.pathak@broadcom.com>
2478 * config/aarch64/aarch64-cores.def: Change the scheduler
2480 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2481 * config/aarch64/thunderx2t99.md: New file.
2483 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2485 * doc/standards.texi (Go Language): Update link to language
2488 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2490 * tree-eh.c (lower_resx): Sanitize profile.
2491 (cleanup_empty_eh_move_lp): Likewise.
2493 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2496 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2498 * cfgloopmanip.h (loop_version): Update prototype.
2499 * modulo-sched.c (sms_schedule): Update call of loop_version.
2500 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2501 * tree-parloops.c (gen_parallel_loop): Likewise.
2502 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2503 * tree-ssa-loop-split.c (split_loop): Likewise.
2504 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2505 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2507 2017-02-05 Martin Liska <mliska@suse.cz>
2510 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2512 (print_operand_address): Initialize a struct to zero.
2514 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2516 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2517 garbage collector only in textual form.
2519 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2521 * doc/extend.texi (x86 specific memory model extensions for
2522 transactional memory): Simplify a phrase.
2524 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2527 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2528 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2529 (atomic_storedi_1): Likewise.
2531 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2533 PR tree-optimization/79338
2534 * tree-parloops.c (gather_scalar_reductions): Don't call
2535 vect_analyze_loop_form for loop->inner before destroying loop's
2538 2017-02-03 Martin Sebor <msebor@redhat.com>
2540 PR tree-optimization/79327
2541 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2542 when precision has resulted in leading zeros.
2543 (format_integer): Adjust the likely counter to assume an unknown
2544 argument that may be zero is non-zero.
2546 2017-02-03 Jason Merrill <jason@redhat.com>
2549 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2550 avoid copying non-taken branch.
2552 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2554 PR tree-optimization/79340
2555 * tree-vect-loop.c (vectorizable_reduction): Release
2556 vec_defs elements after safe_splicing them into other vectors.
2559 PR tree-optimization/79327
2560 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2561 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2563 (format_integer): Use wide_int_to_tree instead of build_int_cst
2564 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2565 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2566 of shortest and longest sequence.
2568 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2570 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2571 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2573 2017-02-03 Walter Lee <walt@tilera.com>
2576 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2577 after initial stackframe link reg save.
2578 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2580 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2583 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2584 wu for stxssp alternative.
2586 2017-02-03 Martin Sebor <msebor@redhat.com>
2588 PR tree-optimization/79352
2589 * gimple-fold.c (get_range_strlen): Add argument.
2590 (get_range_strlen): Change return type to bool.
2591 (get_maxval_strlen): Pass in a dummy argument.
2592 * gimple-fold.h (get_range_strlen): Change return type to bool.
2593 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2594 * tree.h (array_at_struct_end_p): Add argument.
2595 * tree.c (array_at_struct_end_p): Handle it.
2597 2017-02-03 Martin Liska <mliska@suse.cz>
2600 * multiple_target.c (create_dispatcher_calls): Redirect edge
2601 from a caller of a dispatcher.
2602 (expand_target_clones): Make the clones local.
2603 (ipa_target_clone): Do both target clones and resolvers.
2604 (ipa_dispatcher_calls): Remove the pass.
2605 (pass_dispatcher_calls::gate): Likewise.
2606 (make_pass_dispatcher_calls): Likewise.
2607 * passes.def (pass_target_clone): Put as very first IPA early
2610 2017-02-03 Martin Liska <mliska@suse.cz>
2612 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2613 in case of a function with ifunc attribute.
2615 2017-02-03 Martin Liska <mliska@suse.cz>
2617 * cgraph.c (cgraph_node::dump): Dump function version info.
2618 * symtab.c (symtab_node::dump_base): Add missing new line.
2620 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2622 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2623 (ifcombine_ifandif): Use it.
2625 2017-02-03 Martin Liska <mliska@suse.cz>
2627 * doc/invoke.texi: Document default value for
2628 use-after-scope-direct-emission-threshold.
2630 2017-02-03 Martin Liska <mliska@suse.cz>
2632 PR tree-optimization/79339
2633 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2634 (format_floating): Likewise.
2636 2017-02-03 Martin Liska <mliska@suse.cz>
2639 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2641 (ipa_node_params_t::remove): Likewise.
2642 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2643 initialization from removed ipa_node_params_t::insert.
2644 (ipa_node_params::~ipa_node_params): Move from removed
2645 ipa_node_params_t::release.
2646 * symbol-summary.h (symbol_summary::m_released): New member.
2647 Do not release a summary twice. Do not allow to call finalizer
2648 for types of a summary that live in GGC memory.
2650 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2652 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2655 2017-02-02 Martin Sebor <msebor@redhat.com>
2658 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2659 (format_string): Tighten up the range of output for non-constant
2660 strings and correct the expected range for wide non-constant strings.
2662 2017-02-02 Martin Sebor <msebor@redhat.com>
2664 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2667 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2669 (-fdump-tree-@var): Add to index and document how to come up
2670 with pass-specific option and dump file names.
2671 (-fdump-passes): Clarify where to look for output.
2673 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2676 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2677 statistics of the analyzed path; allow threading for speed when
2678 any of BBs along the path are optimized for speed.
2680 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2683 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2684 settings of the virtual registers.
2687 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2689 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2690 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2693 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2695 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2696 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2698 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2700 * config/s390/s390.md: Add missing comments with the expanded
2702 * config/s390/vector.md: Likewise.
2703 * config/s390/vx-builtins.md: Likewise.
2705 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2708 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2709 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2710 conditions on a single line.
2712 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2714 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2715 __S390_VX__ to __VX__.
2717 2017-02-01 Andrew Pinski <apinski@cavium.com>
2719 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2720 stmt_info to record_stmt_cost.
2721 (vect_get_known_peeling_cost): Pass stmt_info if known to
2723 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2724 cpu_vector_cost field into
2725 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2726 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2727 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2728 splitting of scalar_stmt_cost and vec_stmt_cost.
2729 (thunderx_vector_cost): Likewise.
2730 (cortexa57_vector_cost): LIkewise.
2731 (exynosm1_vector_cost): Likewise.
2732 (xgene1_vector_cost): Likewise.
2733 (thunderx2t99_vector_cost): Improve after the splitting of the two
2735 (aarch64_builtin_vectorization_cost): Update for the splitting of
2736 scalar_stmt_cost and vec_stmt_cost.
2738 2017-02-01 Torvald Riegel <triegel@redhat.com>
2739 Richard Henderson <rth@redhat.com>
2741 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2742 conditional on existance of a fast atomic load.
2743 * optabs-query.c (can_atomic_load_p): New function.
2744 * optabs-query.h (can_atomic_load_p): Declare it.
2745 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2746 no fast atomic load is available for the particular size of access.
2747 (expand_atomic_compare_and_swap): Likewise.
2748 (expand_atomic_load): Likewise.
2749 (expand_atomic_store): Likewise.
2750 (expand_atomic_fetch_op): Likewise.
2751 * testsuite/lib/target-supports.exp
2752 (check_effective_target_sync_int_128): Remove x86 because it provides
2753 no fast atomic load.
2754 (check_effective_target_sync_int_128_runtime): Likewise.
2756 2017-02-01 Richard Biener <rguenther@suse.de>
2758 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2759 (graphite_transform_loops): Provide opt-info for optimized nests.
2760 * tree-parloop.c (parallelize_loops): Provide opt-info for
2763 2017-02-01 Richard Biener <rguenther@suse.de>
2766 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2769 2017-02-01 Richard Biener <rguenther@suse.de>
2771 PR tree-optimization/71824
2772 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2773 Verify the loops are valid in the merged SESE region.
2774 (scop_detection::can_represent_loop_1): Check analyzing the
2775 evolution of the number of iterations in the region succeeds.
2777 2017-01-31 Ian Lance Taylor <iant@golang.org>
2779 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2780 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2782 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2784 PR preprocessor/79210
2785 * input.c (get_substring_ranges_for_loc): Replace line_width
2786 assertion with error-handling.
2788 2017-01-31 Richard Biener <rguenther@suse.de>
2790 PR tree-optimization/77318
2791 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2792 (create_pw_aff_from_tree): Take loop parameter.
2793 (add_condition_to_pbb): Pass loop of the condition to
2794 create_pw_aff_from_tree.
2796 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2798 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2799 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2801 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2805 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2807 (convert_int_to_float128): Likewise.
2808 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2809 (convert_int_to_float128): Likewise.
2810 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2811 (UNSPEC_IEEE128_CONVERT): Likewise.
2812 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2813 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2814 Use local variables for IBM extended format.
2815 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2816 (fix_trunc<mode>si2_fprs): Likewise.
2817 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2818 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2819 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2820 to know that we can now have integers of all sizes in vector
2822 (fix<uns>_<mode>di2_hw): Likewise.
2823 (float<uns>_<mode>si2_hw): Likewise.
2824 (fix_<mode>si2_hw): Likewise.
2825 (fixuns_<mode>si2_hw): Likewise.
2826 (float<uns>_<mode>di2_hw): Likewise.
2827 (float_<mode>di2_hw): Likewise.
2828 (float_<mode>si2_hw): Likewise.
2829 (floatuns_<mode>di2_hw): Likewise.
2830 (floatuns_<mode>si2_hw): Likewise.
2831 (xscvqp<su>wz_<mode>): Delete, no longer used.
2832 (xscvqp<su>dz_<mode>): Likewise.
2833 (xscv<su>dqp_<mode>): Likewise.
2834 (ieee128_mfvsrd_64bit): Likewise.
2835 (ieee128_mfvsrd_32bit): Likewise.
2836 (ieee128_mfvsrwz): Likewise.
2837 (ieee128_mtvsrw): Likewise.
2838 (ieee128_mtvsrd_64bit): Likewise.
2839 (ieee128_mtvsrd_32bit): Likewise.
2841 2017-01-31 Martin Liska <mliska@suse.cz>
2844 * ipa-prop.c (ipa_free_all_node_params): Call release method
2845 instead of ~sumbol_summary to not to trigger double times
2848 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2850 PR tree-optimization/71691
2851 * bitmap.h (class auto_bitmap): New.
2852 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2853 is_maybe_undefined instead of ssa_undefined_value_p.
2855 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2857 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2858 __S390_ARCH_LEVEL__ to __ARCH__.
2860 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2862 PR tree-optimization/79267
2863 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2864 if should_remove_lhs_p is true.
2866 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2869 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2870 (add_alignment_attribute): New.
2871 (base_type_die): Add alignment attribute.
2872 (subrange_type_die): Likewise.
2873 (modified_type_die): Likewise.
2874 (gen_array_type_die): Likewise.
2875 (gen_descr_array_type_die: Likewise.
2876 (gen_enumeration_type_die): Likewise.
2877 (gen_subprogram_die): Likewise.
2878 (gen_variable_die): Likewise.
2879 (gen_field_die): Likewise.
2880 (gen_ptr_to_mbr_type_die): Likewise.
2881 (gen_struct_or_union_type_die): Likewise.
2882 (gen_subroutine_type_die): Likewise.
2883 (gen_typedef_die): Likewise.
2884 (base_type_cmp): Compare alignment attribute.
2886 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2889 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2890 (setb_unsigned) New pattern for setb with CCUNS.
2891 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2892 subfc./subfe sequence to avoid overflow problems. Generate a
2893 shorter sequence with cmpld/setb for power9.
2894 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2895 for generating subfc. instruction.
2896 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2897 now uses this instruction.
2899 2017-01-30 Ian Lance Taylor <iant@google.com>
2902 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2903 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2905 2017-01-30 Martin Sebor <msebor@redhat.com>
2907 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2908 Move constant to the right of a relational operator.
2909 (get_mpfr_format_length, format_character, format_string): Ditto.
2910 (should_warn_p, maybe_warn): Same.
2912 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2914 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2917 * asan.c (get_translation_unit_decl): Remove function.
2918 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2920 2017-01-30 Martin Liska <mliska@suse.cz>
2922 PR gcov-profile/79259
2923 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2926 2017-01-30 Martin Liska <mliska@suse.cz>
2929 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2930 Initialize variables with NULL value.
2932 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2935 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2937 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2939 2017-01-30 Richard Biener <rguenther@suse.de>
2941 PR tree-optimization/79276
2942 * tree-vrp.c (process_assert_insertions): Properly adjust common
2943 when removing a duplicate.
2945 * gcc.dg/torture/pr79276.c: New testcase.
2947 2017-01-30 Richard Biener <rguenther@suse.de>
2949 PR tree-optimization/79256
2950 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2951 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2953 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2955 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
2958 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
2959 ("*r<noxa>sbg_<mode>_sll_bitmask")
2960 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
2961 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
2962 Use contiguous_bitmask_nowrap_operand.
2964 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2967 * config/rs6000/altivec.h (vec_xl): Revise #define.
2968 (vec_xst): Likewise.
2970 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
2972 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
2974 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
2976 PR rtl-optimization/79194
2977 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
2978 traps before call to bypass_conditional_jumps.
2980 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2982 PR tree-optimization/71374
2983 * lra-constraints.c (check_conflict_input_operands): New.
2984 (match_reload): Use it.
2986 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2989 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
2990 account to calculate conflict_set.
2992 2017-01-27 Bin Cheng <bin.cheng@arm.com>
2994 PR rtl-optimization/78559
2995 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
2996 other_insn in combine.
2998 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3000 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3001 uint16_type_node for BT_UINT16.
3003 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3005 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3006 "RTL Tests" to menu.
3007 (GIMPLE Tests): New node.
3008 (RTL Tests): New node.
3010 2017-01-27 Richard Biener <rguenther@suse.de>
3012 PR tree-optimization/79245
3013 * tree-loop-distribution.c (distribute_loop): Apply cost
3014 modeling also to detected patterns.
3016 2017-01-27 Richard Biener <rguenther@suse.de>
3018 PR tree-optimization/71433
3019 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3020 (compare_assert_loc): New function.
3021 (process_assert_insertions): Sort and optimize assert locations
3022 to remove duplicates and push down identical assertions on
3023 edges to their destination block.
3025 2017-01-27 Richard Biener <rguenther@suse.de>
3027 PR tree-optimization/79244
3028 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3029 out SSA names even if abnormal.
3031 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3033 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3034 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3035 instead of MPFR_RNDN.
3037 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3040 * arm.c (arm_option_override): Don't call build_target_option_node
3041 until after doing all option overrides.
3042 (arm_valid_target_attribute_tree): Likewise.
3044 2017-01-27 Martin Liska <mliska@suse.cz>
3046 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3047 for {cd}tors and C++ {cd}tors.
3049 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3051 * config/s390/s390.md ("*setmem_long_and")
3052 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3054 2017-01-26 Martin Sebor <msebor@redhat.com>
3056 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3059 2017-01-26 Martin Sebor <msebor@redhat.com>
3061 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3062 HAVE_DFmode before using XFmode or DFmode.
3063 (parse_directive): Avoid using the z length modifier to avoid
3064 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3067 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3068 to accept adjustment as an array.
3069 (get_int_range): New function.
3070 (struct directive): Make width and prec arrays.
3071 (directive::set_width, directive::set_precision): Call get_int_range.
3072 (format_integer, format_floating): Handle width and precision ranges.
3073 (format_string, parse_directive): Same.
3075 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3078 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3079 comdat_type_p set, just clone them, but keep the children in the
3083 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3084 which have direct callers with -fvar-tracking-assignments enabled
3086 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3087 inside of type units.
3089 2017-01-26 Martin Sebor <msebor@redhat.com>
3092 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3094 (struct format_result): Replace number_chars, number_chars_min,
3095 and number_chars_max with a single member of struct result_range.
3097 (format_result::operator+=): Adjust.
3098 (struct fmtresult): Remove bounded. Handle likely and unlikely
3100 (fmtresult::adjust_for_width_or_precision): New function.
3101 (fmtresult:type_max_digits): New function.
3102 (bytes_remaining): Handle likely and unlikely counters.
3103 (min_bytes_remaining): Remove.
3104 (format_percent): Simplify.
3105 (format_integer, format_floating): Set likely and unlikely counters.
3106 (get_string_length, format_character, format_string): Same.
3107 (format_plain, should_warn_p): New function.
3108 (maybe_warn): Call should_warn_p. Update diagnostic messages
3109 and handle those for all directives, including plain strings.
3110 (format_directive): Handle likely and unlikely counters.
3111 Remove unnecessary quoting from diagnostics. Add an informational
3113 (add_bytes): Remove.
3114 (pass_sprintf_length::compute_format_length): Simplify.
3115 (try_substitute_return_value): Handle likely and unlikely counters.
3117 2017-01-26 Carl Love <cel@us.ibm.com>
3119 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3120 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3122 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3125 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3126 endianess for subregs into account.
3127 * lra-constraints.c (lra_constraints): Do risky transformations
3128 always on the first iteration.
3129 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3130 last_call_used_reg_set.
3131 (process_bb_lives): Define and use last_call_used_reg_set.
3132 * lra.c (lra): Always continue after lra_constraints on the first
3135 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3137 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3139 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3141 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3143 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3144 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3145 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3146 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3147 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3148 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3149 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3150 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3151 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3153 2017-01-26 Marek Polacek <polacek@redhat.com>
3156 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3157 for the third operand.
3159 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3162 * omp-low.c (struct omp_context): Add simt_stmt field.
3163 (scan_omp_for): Return omp_context *.
3164 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3165 context to the _simt_ SIMD stmt.
3166 (lower_omp_for): For combined SIMD with sibling _simt_
3167 SIMD, make sure to use the same decls in _looptemp_
3168 clauses as in the sibling.
3170 2017-01-26 David Sherwood <david.sherwood@arm.com>
3173 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3176 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3179 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3180 emit fld b; fld a; if possible.
3182 * brig-builtins.def: Update copyright years.
3183 * config/arm/arm_acle_builtins.def: Update copyright years.
3185 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3188 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3189 constraint instead of o for the stxsd instruction.
3191 2017-01-25 Carl Love <cel@us.ibm.com>
3193 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3194 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3196 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3198 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3200 2017-01-25 Richard Biener <rguenther@suse.de>
3202 PR tree-optimization/69264
3203 * target.def (vector_alignment_reachable): Improve documentation.
3204 * doc/tm.texi: Regenerate.
3205 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3207 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3208 earlier changes with respect to TYPE_USER_ALIGN.
3209 (vector_alignment_reachable_p): Likewise. Improve dumping.
3211 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3214 * config/arm/arm.md (xordi3): Force constant operand into a register
3217 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3219 * doc/invoke.texi (-fstore-merging): Correct default optimization
3220 levels at which it is enabled.
3221 (-O): Move -fstore-merging from list to...
3224 2017-01-25 Richard Biener <rguenther@suse.de>
3227 * omp-expand.c: Include debug.h.
3228 (expand_omp_taskreg): Make sure to generate early debug before
3229 outlining anything from a function.
3230 (expand_omp_target): Likewise.
3231 (grid_expand_target_grid_body): Likewise.
3233 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3236 * asan.c (get_translation_unit_decl): New function.
3237 (asan_add_global): Extract modules file name from globals
3238 TRANSLATION_UNIT_DECL name.
3240 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3243 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3244 for long calls with APCS frame and VFP.
3246 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3248 * cfg.c (original_copy_tables_initialized_p): New function.
3249 * cfg.h (original_copy_tables_initialized_p): New decl.
3250 * cfgrtl.c (relink_block_chain): Guard the call to
3251 free_original_copy_tables with a call to
3252 original_copy_tables_initialized_p.
3253 * cgraph.h (symtab_node::native_rtl_p): New decl.
3254 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3255 (symtab_node::needed_p): Don't assert for early assembly output
3256 for __RTL functions.
3257 (cgraph_node::finalize_function): Set "force_output" for __RTL
3259 (cgraph_node::analyze): Bail out early for __RTL functions.
3260 (analyze_functions): Update assertion to support __RTL functions.
3261 (cgraph_node::expand): Bail out early for __RTL functions.
3262 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3264 * function.h (struct function): Update comment for field
3266 * gimple-expr.c: Include "tree-pass.h".
3267 (gimple_has_body_p): Return false for __RTL functions.
3268 * Makefile.in (OBJS): Add run-rtl-passes.o.
3269 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3271 (gcc::pass_manager::get_clean_slate): New accessor.
3272 * passes.c: Include "insn-addr.h".
3273 (should_skip_pass_p): Add logging. Update logic for running
3274 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3275 property-provider override so it is only done for gimple passes.
3277 (skip_pass): New function.
3278 (execute_one_pass): Call skip_pass when skipping passes.
3279 * read-md.c (md_reader::read_char): Support filtering
3280 the input to a subset of line numbers.
3281 (md_reader::md_reader): Initialize fields
3282 m_first_line and m_last_line.
3283 (md_reader::read_file_fragment): New function.
3284 * read-md.h (md_reader::read_file_fragment): New decl.
3285 (md_reader::m_first_line): New field.
3286 (md_reader::m_last_line): New field.
3287 * read-rtl-function.c (function_reader::create_function): Only
3288 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3289 curr_properties. Set DECL_INITIAL to a dummy block.
3290 (read_rtl_function_body_from_file_range): New function.
3291 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3293 * run-rtl-passes.c: New file.
3294 * run-rtl-passes.h: New file.
3296 2017-01-24 Jeff Law <law@redhat.com>
3298 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3301 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3303 PR tree-optimization/79159
3304 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3305 (record_nonwrapping_iv): Improve boundary using above function if no
3306 value range information.
3308 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3309 Martin Jambor <mjambor@suse.cz>
3311 * brig-builtins.def: New file.
3312 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3313 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3314 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3315 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3316 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3317 * builtin-types.def (BT_INT8): New.
3318 (BT_INT16): Likewise.
3319 (BT_UINT8): Likewise.
3320 (BT_UINT16): Likewise.
3321 (BT_FN_ULONG): Likewise.
3322 (BT_FN_UINT_INT): Likewise.
3323 (BT_FN_UINT_ULONG): Likewise.
3324 (BT_FN_UINT_LONG): Likewise.
3325 (BT_FN_UINT_PTR): Likewise.
3326 (BT_FN_ULONG_PTR): Likewise.
3327 (BT_FN_INT8_FLOAT): Likewise.
3328 (BT_FN_INT16_FLOAT): Likewise.
3329 (BT_FN_UINT32_FLOAT): Likewise.
3330 (BT_FN_UINT16_FLOAT): Likewise.
3331 (BT_FN_UINT8_FLOAT): Likewise.
3332 (BT_FN_UINT64_FLOAT): Likewise.
3333 (BT_FN_UINT16_UINT32): Likewise.
3334 (BT_FN_UINT32_UINT16): Likewise.
3335 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3336 (BT_FN_INT_PTR_INT): Likewise.
3337 (BT_FN_UINT_PTR_UINT): Likewise.
3338 (BT_FN_LONG_PTR_LONG): Likewise.
3339 (BT_FN_ULONG_PTR_ULONG): Likewise.
3340 (BT_FN_VOID_UINT64_UINT64): Likewise.
3341 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3342 (BT_FN_INT8_INT8_INT8): Likewise.
3343 (BT_FN_INT16_INT16_INT16): Likewise.
3344 (BT_FN_INT_INT_INT): Likewise.
3345 (BT_FN_UINT_FLOAT_UINT): Likewise.
3346 (BT_FN_FLOAT_UINT_UINT): Likewise.
3347 (BT_FN_ULONG_UINT_UINT): Likewise.
3348 (BT_FN_ULONG_UINT_PTR): Likewise.
3349 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3350 (BT_FN_UINT_UINT_UINT): Likewise.
3351 (BT_FN_VOID_UINT_PTR): Likewise.
3352 (BT_FN_UINT_UINT_PTR: Likewise.
3353 (BT_FN_UINT32_UINT64_PTR): Likewise.
3354 (BT_FN_INT_INT_UINT_UINT): Likewise.
3355 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3356 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3357 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3358 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3359 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3360 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3361 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3362 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3363 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3364 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3365 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3366 * doc/frontends.texi: List BRIG FE.
3367 * doc/install.texi (Testing): Add BRIG tesring requirements.
3368 * doc/invoke.texi (Overall Options): Mention BRIG.
3369 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3371 2017-01-24 Richard Biener <rguenther@suse.de>
3373 PR translation/79208
3374 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3376 2017-01-24 Martin Jambor <mjambor@suse.cz>
3379 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3380 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3383 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3386 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3387 casts from signed to unsigned really don't have a range.
3389 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3391 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3392 GMP_RNDx for compatiblity.
3394 2017-01-24 Martin Liska <mliska@suse.cz>
3397 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3398 that would prevent us to call alloca with -1 as argument.
3400 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3402 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3403 Avoid -Wformat-security warning.
3405 2017-01-23 Andrew Pinski <apinski@cavium.com>
3407 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3410 2017-01-23 Martin Sebor <msebor@redhat.com>
3413 * gimple-ssa-sprintf.c (warn_level): New global.
3414 (format_integer): Use it here and throughout the rest of the file.
3415 Use the same switch to compute sign as base.
3416 (maybe_warn): New function.
3417 (format_directive): Factor out warnings into maybe_warn.
3418 Add debugging output. Use warn_level.
3419 (add_bytes): Use warn_level.
3420 (pass_sprintf_length::compute_format_length): Add debugging output.
3421 (try_substitute_return_value): Same.
3422 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3425 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3426 (struct fmtresult, format_integer, format_floating): Adjust.
3427 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3428 (get_string_length, format_string,format_directive): Same.
3429 (pass_sprintf_length::compute_format_length): Same.
3430 (try_substitute_return_value): Simplify slightly.
3433 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3434 (fmtresult::operator+=): Outlined.
3435 (struct fmtresult): Add ctors.
3436 (struct conversion_spec): Rename...
3437 (struct directive): ...to this. Add and remove data members.
3438 (directive::set_width, directive::set_precision): New functions.
3439 (format_percent): Use fmtresult ctor.
3440 (get_width_and_precision): Remove.
3441 (format_integer): Make naming changes. Avoid computing width and
3443 (format_floating): Same. Adjust indentation.
3444 (format_character, format_none): New functions.
3445 (format_string): Moved character handling to format_character.
3446 (format_directive): Remove arguments, change return type.
3447 (parse_directive): New function.
3448 (pass_sprintf_length::compute_format_length): Move directive
3449 parsing to parse_directive.
3451 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3453 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3454 (assign_assembler_name_if_needed): ... this.
3455 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3456 (assign_assembler_name_if_needed): ... this.
3457 (free_lang_data_in_cgraph): Adjust callers.
3458 * cgraphunit.c (cgraph_node::analyze): Likewise.
3459 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3462 2017-01-23 Richard Biener <rguenther@suse.de>
3464 PR tree-optimization/79088
3465 PR tree-optimization/79188
3466 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3467 resetting loop bounds after last path deletion. Reset loop
3468 bounds of the target loop, make code match the comments.
3469 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3470 Make sure loops need no fixups.
3472 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3474 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3475 exponent support with double type for first argument.
3476 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3477 type returned by __builtin_vec_extract_sig,
3478 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3479 functions from "vector int" to "vector unsigned int" or from
3480 "vector long long int" to "vector unsigned long long int".
3481 Changed type returned by __builtin_vec_extract_exp,
3482 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3483 functions from "vector int" to "vector unsigned int" or from
3484 "vector long long int" to "vector unsigned long long int".
3485 Changed return type of __builtin_vec_test_data_class,
3486 __builtin_vec_test_data_class_sp, and
3487 __builtin_vec_test_data_class_dp from "vector int" to
3488 "vector bool int" or from "vector long long int" to "vector bool
3489 long long int" and changed second argument type from "unsigned
3490 int" to "int". Added new overloaded function forms "vector float
3491 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3492 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3493 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3494 double, vector unsigned long long int)" and "vector double
3495 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3496 long int)". Changed return type of
3497 __builtin_scalar_test_data_class and
3498 __builtin_scalar_test_data_class_sp and
3499 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3500 int" and changed second argument from "unsigned int" to "int".
3501 Changed type returned by __builtin_scalar_test_neg,
3502 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3503 from "int" to "bool int". Added new overloaded function form
3504 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3505 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3506 exponent double-precision with floating point first argument.
3507 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3508 documentation of scalar_test_data_class, scalar_test_neg,
3509 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3510 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3511 vec_test_data_class built-in functions to reflect refinements in
3512 their type signatures.
3514 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3516 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3518 (aarch64_elf_asm_destructor): Likewise.
3520 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3522 PR rtl-optimization/78634
3523 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3524 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3525 * ifcvt.c (noce_try_cmove): Add missing cost check.
3527 PR rtl-optimization/71724
3528 * combine.c (if_then_else_cond): Look for situations where it is
3529 beneficial to undo the work of one of the recursive calls.
3531 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3533 PR tree-optimization/70754
3534 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3535 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3536 combined stmt before it if not NULL.
3537 (combine_chains): Process refs reversely and compute dominance point
3540 2017-01-23 Martin Liska <mliska@suse.cz>
3542 PR tree-optimization/79196
3543 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3544 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3546 (strlen_optimize_stmt): Call the renamed function.
3548 2017-01-23 Michael Matz <matz@suse.de>
3550 PR tree-optimization/78384
3551 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3553 2017-01-23 Richard Biener <rguenther@suse.de>
3555 PR tree-optimization/79186
3556 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3557 both incoming edges before moving an assert.
3559 2017-01-23 Martin Jambor <mjambor@suse.cz>
3561 * ipa-prop.c (load_from_param_1): Removed.
3562 (load_from_unmodified_param): Bits from load_from_param_1 put back
3564 (load_from_param): Removed.
3565 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3566 with stmt. Reverted back to use of load_from_unmodified_param.
3568 2017-01-23 Martin Jambor <mjambor@suse.cz>
3571 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3572 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3573 field a pointer to garbage collected vector, mark lattices and
3574 ipcp_orig_node with GTY((skip)).
3575 (ipa_get_param_count): Adjust to descriptors being a pointer.
3576 (ipa_get_param): Likewise.
3577 (ipa_get_type): Likewise.
3578 (ipa_get_param_move_cost): Likewise.
3579 (ipa_set_param_used): Likewise.
3580 (ipa_get_controlled_uses): Likewise.
3581 (ipa_set_controlled_uses): Likewise.
3582 (ipa_is_param_used): Likewise.
3583 (ipa_node_params_t): Move into garbage collector. New methods insert
3585 (ipa_node_params_sum): Annotate wth GTY(()).
3586 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3588 (ipa_load_from_parm_agg): Adjust declaration.
3589 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3590 * ipa-profile.c (ipa_profile): Likewise.
3591 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3592 (ipa_populate_param_decls): Make descriptors parameter garbage
3594 (ipa_dump_param): Adjust to descriptors being a pointer.
3595 (ipa_alloc_node_params): Likewise.
3596 (ipa_initialize_node_params): Likewise.
3597 (load_from_param_1): Make descriptors parameter garbage collected.
3598 (load_from_unmodified_param): Likewise.
3599 (load_from_param): Likewise.
3600 (ipa_load_from_parm_agg): Likewise.
3601 (ipa_node_params::~ipa_node_params): Removed.
3602 (ipa_free_all_node_params): Remove call to delete operator.
3603 (ipa_node_params_t::insert): New.
3604 (ipa_node_params_t::remove): Likewise.
3605 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3606 copy known_csts and known_contexts vectors.
3607 (ipa_read_node_info): Adjust to descriptors being a pointer.
3608 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3610 (ipcp_transform_function): Make descriptors variable garbage
3613 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3615 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3616 * config/i386/avx512dqintrin.h: Ditto.
3617 * config/i386/avx512fintrin.h: Ditto.
3618 * gcc/config/i386/i386.c: Handle new builtins.
3619 * config/i386/i386-builtin.def: Add new builtins.
3620 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3621 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3623 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3624 Martin Liska <mliska@suse.cz>
3626 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3627 * asan.c (asan_expand_poison_ifn): Support stores and use
3628 appropriate ASAN report function.
3629 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3630 * internal-fn.def (ASAN_POISON_USE): Declare.
3631 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3632 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3633 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3634 ASAN_POISON calls w/o LHS.
3635 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3636 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3637 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3638 * gimplify.c (asan_poison_variables): Add attribute
3639 use_after_scope_memory to variables that really needs to live
3641 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3642 having the attribute.
3644 2017-01-23 Martin Liska <mliska@suse.cz>
3646 * asan.c (create_asan_shadow_var): New function.
3647 (asan_expand_poison_ifn): Likewise.
3648 * asan.h (asan_expand_poison_ifn): New declaration.
3649 * internal-fn.c (expand_ASAN_POISON): Likewise.
3650 * internal-fn.def (ASAN_POISON): New builtin.
3651 * sanopt.c (pass_sanopt::execute): Expand
3652 asan_expand_poison_ifn.
3653 * tree-inline.c (copy_decl_for_dup_finish): Make function
3655 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3656 * tree-ssa.c (is_asan_mark_p): New function.
3657 (execute_update_addresses_taken): Rewrite local variables
3658 (identified just by use-after-scope as addressable) into SSA.
3660 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3662 * doc/install.texi (Specific): opensource.apple.com uses https
3663 now. Remove trailing slash.
3665 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3667 * README.Portability: Remove note on an Irix compatibility issue.
3669 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3671 * gcov.c (INCLUDE_ALGORITHM): Define.
3672 (INCLUDE_VECTOR): Define.
3673 No longer include <vector> and <algorithm> directly.
3675 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3677 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3679 * doc/invoke.texi (Code Gen Options): Ditto.
3681 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3684 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3686 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3688 rtl-optimization/79125
3689 * cprop.c (local_cprop_pass): Handle cases where we make an
3692 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3696 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3697 read from, for big endian.
3699 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3701 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3702 register pauth builtins for LP64 only.
3704 2017-01-20 Marek Polacek <polacek@redhat.com>
3707 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3710 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3712 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3714 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3715 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3716 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3718 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3721 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3722 in big-endian mode when they are not a single duplicated value.
3724 2017-01-20 Richard Biener <rguenther@suse.de>
3726 * BASE-VER: Bump to 7.0.1.
3728 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3730 * omp-low.c (omplow_simd_context): New struct. Use it...
3731 (lower_rec_simd_input_clauses): ...here and...
3732 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3733 references to idx, lane, max_vf, is_simt.
3735 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3737 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3740 2017-01-20 Martin Jambor <mjambor@suse.cz>
3742 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3743 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3745 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3746 (GTFILES): Rename hsa.c to hsa-common.c.
3747 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3748 * hsa-dump.c: Likewise.
3749 * hsa-gen.c: Likewise.
3750 * hsa-regalloc.c: Likewise.
3751 * ipa-hsa.c: Likewise.
3752 * omp-expand.c: Likewise.
3753 * omp-low.c: Likewise.
3754 * toplev.c: Likewise.
3756 2017-01-20 Marek Polacek <polacek@redhat.com>
3759 * doc/invoke.texi: Document -Wduplicated-branches.
3760 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3761 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3762 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3763 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3764 return 0 only when not OEP_LEXICOGRAPHIC.
3765 (fold_build_cleanup_point_expr): Use the expression
3766 location when building CLEANUP_POINT_EXPR.
3767 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3768 * tree.c (add_expr): Handle error_mark_node.
3770 2017-01-20 Martin Liska <mliska@suse.cz>
3773 * tree-profile.c (init_ic_make_global_vars): Do not call
3775 (gimple_init_gcov_profiler): Likewise.
3777 2017-01-20 Martin Liska <mliska@suse.cz>
3780 * cgraph.h (maybe_create_reference): Remove argument and
3782 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3784 * ipa-cp.c (create_specialized_node): Likewise.
3785 * symtab.c (symtab_node::maybe_create_reference): Handle
3786 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3788 2017-01-20 Martin Liska <mliska@suse.cz>
3790 * read-rtl-function.c (function_reader::create_function): Use
3791 build_decl instread of build_decl_stat.
3793 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3795 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3796 * config/i386/avx512dqintrin.h: Ditto.
3797 * config/i386/avx512fintrin.h: Ditto.
3798 * config/i386/i386-builtin-types.def: Add new types.
3799 * gcc/config/i386/i386.c: Handle new types.
3800 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3801 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3802 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3803 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3804 (__builtin_ia32_kshiftridi): New.
3805 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3807 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3811 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3812 define to rs6000_init_stack_protect_guard.
3813 (rs6000_init_stack_protect_guard): New function.
3815 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3816 Yunqiang Su <yunqiang.su@imgtec.com>
3818 * config.gcc (supported_defaults): Add madd4.
3819 (with_madd4): Add validation.
3820 (all_defaults): Add madd4.
3821 * config/mips/mips.opt (mmadd4): New option.
3822 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3824 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3826 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3827 (ISA_HAS_FUSED_MADD4): Likewise.
3828 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3829 * gcc/doc/install.texi (--with-madd4): Document the new option.
3831 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3833 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3834 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3835 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3836 (aarch64_init_pauth_hint_builtins): New.
3837 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3838 (aarch64_expand_builtin): Expand new builtins.
3840 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3842 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3843 * combine-stack-adj.c (no_unhandled_cfa): Handle
3844 REG_CFA_TOGGLE_RA_MANGLE.
3845 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3846 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3847 info for return address signing.
3848 (aarch64_expand_epilogue): Likewise.
3850 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3852 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3853 * config/aarch64/aarch64-protos.h
3854 (aarch64_return_address_signing_enabled): New declaration.
3855 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3857 (aarch64_expand_prologue): Sign return address before it's pushed onto
3859 (aarch64_expand_epilogue): Authenticate return address fetched from
3861 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3862 (aarch64_attributes): New function attributes for "sign-return-address".
3863 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3864 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3865 ("*do_return"): Generate combined instructions according to key index.
3866 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3867 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3869 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3870 * config/aarch64/aarch64.opt (msign-return-address=): New.
3871 * doc/extend.texi (AArch64 Function Attributes): Documents
3872 "sign-return-address=".
3873 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3875 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3877 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3878 overall option summary.
3880 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3882 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3883 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3884 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3885 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3887 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3889 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3890 -mpower9-minmax by default for -mcpu=power9.
3891 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3892 128-bit floating point.
3894 2017-01-20 Alan Modra <amodra@gmail.com>
3896 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3897 optimizing for size.
3899 2017-01-20 Alan Modra <amodra@gmail.com>
3902 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3903 for strcmp and strncmp from corresponding builtin decl.
3905 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3907 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3908 instead of i386/rtems-64.h.
3909 * config/i386/rtems-64.h: Remove.
3911 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3915 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3917 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3919 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3921 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3922 Change int to HOST_WIDE_INT.
3923 * config/aarch64/aarch64-protos.h
3924 (aarch64_simd_gen_const_vector_dup): Likewise.
3925 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3927 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3929 * langhooks-def.h (lhd_type_for_size): New decl.
3930 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3931 * langhooks.c (lhd_type_for_size): New function, taken from
3934 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3936 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3937 define_bypass for CR latency.
3938 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3939 (power9-alu2): Add define_bypass for CR latency.
3941 (power9-mul): Update insn latency.
3942 (power9-mul-compare): Update insn latency, bypass latency and remove
3945 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3947 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3949 * config/aarch64/aarch64.md
3950 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3951 aarch64_nopcrelative_literal_loads.
3952 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3954 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3956 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
3958 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
3960 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
3963 * config.gcc (supported_defaults): Add lxc1-sxc1.
3964 (with_lxc1_sxc1): Add validation.
3965 (all_defaults): Add lxc1-sxc1.
3966 * config/mips/mips.opt (mlxc1-sxc1): New option.
3967 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3969 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3970 __mips_no_lxc1_sxc1.
3971 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
3972 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
3973 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
3975 2017-01-19 Richard Biener <rguenther@suse.de>
3977 PR tree-optimization/72488
3978 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
3979 sure to restore SSA info.
3980 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
3982 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
3984 PR rtl-optimization/79121
3985 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
3986 of the inner type when shifting an extended value.
3988 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3991 * symtab.c (symtab_node::equal_address_to): Fix comparing of
3992 interposable aliases.
3994 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
3997 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
3998 Use the evmergelohi instruction.
3999 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4000 (mov_sitf_e500_subreg8_2_be): Likewise.
4001 (mov_sitf_e500_subreg12_2_le): Likewise.
4002 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4003 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4004 (mov_sitf_e500_subreg8_2_le): Likewise.
4005 (mov_sitf_e500_subreg12_2_be): Likewise.
4007 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4009 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4010 attribute from vecsimple to vecperm.
4011 (altivec_vbpermq2): Likewise.
4013 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4016 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4018 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4019 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4020 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4021 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4022 case where N arg is SIZE_MAX.
4023 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4024 (cmpstrsi): Add pattern.
4026 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4028 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4029 __builtin_vec_revb builtins.
4030 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4031 built-in functions to support generation of the ISA 3.0 XXBR<x>
4032 vector byte reverse instructions.
4033 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4034 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4035 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4036 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4037 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4038 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4039 (P9V_BUILTIN_VEC_REVB): Likewise.
4040 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4041 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4042 (p9_xxbrq_v16qi): Likewise.
4043 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4044 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4045 (p9_xxbrh_v8hi): Likewise.
4046 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4047 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4048 vec_revb built-in functions.
4050 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4052 PR rtl-optimization/78952
4053 * config/i386/i386.md (any_extract): New code iterator.
4054 (*insvqi_2): Use any_extract for source operand.
4055 (*insvqi_3): Use any_shiftrt for source operand.
4057 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4059 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4061 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4063 2017-01-18 Matthias Klose <doko@ubuntu.com>
4065 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4067 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4069 * config/rs6000/altivec.h (vec_bperm): Change #define.
4070 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4071 (altivec_vbpermq2): New define_insn.
4072 (altivec_vbpermd): Likewise.
4073 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4075 (VBPERMD): Likewise.
4076 (VBPERM): New polymorphic function interface.
4077 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4078 Add entries for P9V_BUILTIN_VEC_VBPERM.
4079 * doc/extend.texi: Add interfaces for vec_bperm.
4081 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4083 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4084 first letter of error messages.
4085 (s390_resolve_overloaded_builtin): Likewise.
4086 * config/s390/s390.c (s390_expand_builtin): Likewise.
4087 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4088 (s390_valid_target_attribute_inner_p): Likewise.
4089 * config/s390/s390.md ("tabort"): Likewise.
4091 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4093 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4094 (ISA_AVOID_DIV_HILO): New macro.
4095 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4096 (ISA_HAS_DDIV): Likewise.
4098 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4100 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4102 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4104 * doc/invoke.texi (fabi-version): Spelling fix.
4106 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4109 * doc/invoke.texi (fabi-version): Mention mangling fix for
4112 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4115 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4117 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4120 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4123 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4125 (stack_protect_test): Ditto.
4126 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4127 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4129 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4130 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4131 -mstack-protector-guard-offset=.
4132 (RS/6000 and PowerPC Options): Ditto.
4134 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4136 * config/i386/i386.h (MASK_CLASS_P): New define.
4137 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4138 there are no registers from different register sets also when
4139 mask registers are used. Update function comment.
4140 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4141 to (*k/*r) and (*k/*km) alternatives.
4143 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4145 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4146 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4147 (EH_RETURN_HANDLER_RTX): New define.
4148 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4149 Force frame pointer in EH return functions.
4150 (aarch64_expand_epilogue): Add barrier for eh_return.
4151 (aarch64_final_eh_return_addr): Remove.
4152 (aarch64_eh_return_handler_rtx): New function.
4153 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4155 (aarch64_eh_return_handler_rtx): New prototype.
4157 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4159 * config/rs6000/altivec.h (vec_rlmi): New #define.
4160 (vec_vrlnm): Likewise.
4161 (vec_rlnm): Likewise.
4162 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4163 (UNSPEC_VRLNM): Likewise.
4164 (VIlong): New mode iterator.
4165 (altivec_vrl<VI_char>mi): New define_insn.
4166 (altivec_vrl<VI_char>nm): Likewise.
4167 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4170 (RLNM): New polymorphic function entry.
4171 (VRLWMI): New monomorphic function entry.
4173 (RLMI): New polymorphic function entry.
4174 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4175 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4176 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4179 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4182 * dwarf2out.c (field_byte_offset): Restore the
4183 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4184 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4185 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4188 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4191 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4193 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4196 * dwarf2out.c (add_data_member_location_attribute): For constant
4197 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4198 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4199 DW_AT_byte_size attributes.
4201 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4203 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4204 after forcing to constant memory when the code model is medium.
4206 2017-01-17 Julia Koval <julia.koval@intel.com>
4209 * config/i386/avx512fintrin.h
4210 (_mm512_i32gather_ps): Change __addr type to void const*.
4211 (_mm512_mask_i32gather_ps): Ditto.
4212 (_mm512_i32gather_pd): Ditto.
4213 (_mm512_mask_i32gather_pd): Ditto.
4214 (_mm512_i64gather_ps): Ditto.
4215 (_mm512_mask_i64gather_ps): Ditto.
4216 (_mm512_i64gather_pd): Ditto.
4217 (_mm512_mask_i64gather_pd): Ditto.
4218 (_mm512_i32gather_epi32): Ditto.
4219 (_mm512_mask_i32gather_epi32): Ditto.
4220 (_mm512_i32gather_epi64): Ditto.
4221 (_mm512_mask_i32gather_epi64): Ditto.
4222 (_mm512_i64gather_epi32): Ditto.
4223 (_mm512_mask_i64gather_epi32): Ditto.
4224 (_mm512_i64gather_epi64): Ditto.
4225 (_mm512_mask_i64gather_epi64): Ditto.
4226 (_mm512_i32scatter_ps): Change __addr type to void*.
4227 (_mm512_mask_i32scatter_ps): Ditto.
4228 (_mm512_i32scatter_pd): Ditto.
4229 (_mm512_mask_i32scatter_pd): Ditto.
4230 (_mm512_i64scatter_ps): Ditto.
4231 (_mm512_mask_i64scatter_ps): Ditto.
4232 (_mm512_i64scatter_pd): Ditto.
4233 (_mm512_mask_i64scatter_pd): Ditto.
4234 (_mm512_i32scatter_epi32): Ditto.
4235 (_mm512_mask_i32scatter_epi32): Ditto.
4236 (_mm512_i32scatter_epi64): Ditto.
4237 (_mm512_mask_i32scatter_epi64): Ditto.
4238 (_mm512_i64scatter_epi32): Ditto.
4239 (_mm512_mask_i64scatter_epi32): Ditto.
4240 (_mm512_i64scatter_epi64): Ditto.
4241 (_mm512_mask_i64scatter_epi64): Ditto.
4242 * config/i386/avx512pfintrin.h
4243 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4244 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4245 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4246 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4247 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4248 (_mm512_prefetch_i32scatter_ps): Ditto.
4249 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4250 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4251 (_mm512_prefetch_i64scatter_pd): Ditto.
4252 (_mm512_prefetch_i64scatter_ps): Ditto.
4253 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4254 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4255 * config/i386/avx512vlintrin.h
4256 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4257 (_mm_mmask_i32gather_ps): Ditto.
4258 (_mm256_mmask_i32gather_pd): Ditto.
4259 (_mm_mmask_i32gather_pd): Ditto.
4260 (_mm256_mmask_i64gather_ps): Ditto.
4261 (_mm_mmask_i64gather_ps): Ditto.
4262 (_mm256_mmask_i64gather_pd): Ditto.
4263 (_mm_mmask_i64gather_pd): Ditto.
4264 (_mm256_mmask_i32gather_epi32): Ditto.
4265 (_mm_mmask_i32gather_epi32): Ditto.
4266 (_mm256_mmask_i32gather_epi64): Ditto.
4267 (_mm_mmask_i32gather_epi64): Ditto.
4268 (_mm256_mmask_i64gather_epi32): Ditto.
4269 (_mm_mmask_i64gather_epi32): Ditto.
4270 (_mm256_mmask_i64gather_epi64): Ditto.
4271 (_mm_mmask_i64gather_epi64): Ditto.
4272 (_mm256_i32scatter_ps): Change __addr type to void*.
4273 (_mm256_mask_i32scatter_ps): Ditto.
4274 (_mm_i32scatter_ps): Ditto.
4275 (_mm_mask_i32scatter_ps): Ditto.
4276 (_mm256_i32scatter_pd): Ditto.
4277 (_mm256_mask_i32scatter_pd): Ditto.
4278 (_mm_i32scatter_pd): Ditto.
4279 (_mm_mask_i32scatter_pd): Ditto.
4280 (_mm256_i64scatter_ps): Ditto.
4281 (_mm256_mask_i64scatter_ps): Ditto.
4282 (_mm_i64scatter_ps): Ditto.
4283 (_mm_mask_i64scatter_ps): Ditto.
4284 (_mm256_i64scatter_pd): Ditto.
4285 (_mm256_mask_i64scatter_pd): Ditto.
4286 (_mm_i64scatter_pd): Ditto.
4287 (_mm_mask_i64scatter_pd): Ditto.
4288 (_mm256_i32scatter_epi32): Ditto.
4289 (_mm256_mask_i32scatter_epi32): Ditto.
4290 (_mm_i32scatter_epi32): Ditto.
4291 (_mm_mask_i32scatter_epi32): Ditto.
4292 (_mm256_i32scatter_epi64): Ditto.
4293 (_mm256_mask_i32scatter_epi64): Ditto.
4294 (_mm_i32scatter_epi64): Ditto.
4295 (_mm_mask_i32scatter_epi64): Ditto.
4296 (_mm256_i64scatter_epi32): Ditto.
4297 (_mm256_mask_i64scatter_epi32): Ditto.
4298 (_mm_i64scatter_epi32): Ditto.
4299 (_mm_mask_i64scatter_epi32): Ditto.
4300 (_mm256_i64scatter_epi64): Ditto.
4301 (_mm256_mask_i64scatter_epi64): Ditto.
4302 (_mm_i64scatter_epi64): Ditto.
4303 (_mm_mask_i64scatter_epi64): Ditto.
4304 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4305 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4306 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4307 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4308 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4309 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4310 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4311 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4312 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4313 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4314 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4315 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4316 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4317 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4318 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4319 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4320 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4321 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4322 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4323 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4324 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4325 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4326 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4327 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4328 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4329 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4330 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4331 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4332 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4333 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4334 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4335 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4336 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4337 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4338 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4339 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4340 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4341 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4342 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4343 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4344 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4345 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4346 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4347 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4348 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4349 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4350 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4351 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4352 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4353 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4354 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4355 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4356 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4357 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4358 definitions accordingly.
4360 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4361 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4364 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4367 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4370 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4371 hard regs for inner regno.
4373 2017-01-17 Martin Liska <mliska@suse.cz>
4376 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4377 assumption and add comment.
4379 2017-01-17 Nathan Sidwell <nathan@acm.org>
4381 * ipa-visibility.c (localize_node): New function, broken out of ...
4382 (function_and_variable_visibility): ... here. Call it.
4384 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4387 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4388 correctly set frequency of oudgoing edge.
4389 (duplicate_thread_path): Fix profile updating.
4391 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4394 * configure.ac: Add GCC_BASE_VER.
4395 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4396 version from BASE-VER file.
4397 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4398 (gcc.o): Depend on $(BASEVER).
4399 * common.opt (dumpfullversion): New option.
4400 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4401 * doc/invoke.texi: Document -dumpfullversion.
4402 * doc/install.texi: Document --with-gcc-major-version-only.
4403 * configure: Regenerated.
4405 2017-01-17 Richard Biener <rguenther@suse.de>
4407 PR tree-optimization/71433
4408 * tree-vrp.c (register_new_assert_for): Merge same asserts
4409 on all incoming edges.
4410 (process_assert_insertions_for): Handle insertions at the
4413 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4415 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4416 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4418 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4421 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4424 2017-01-17 Alan Modra <amodra@gmail.com>
4427 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4428 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4429 symbolic stack limit when pic.
4431 2017-01-16 Martin Sebor <msebor@redhat.com>
4433 PR tree-optimization/78608
4434 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4436 2017-01-16 Jeff Law <law@redhat.com>
4439 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4440 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4441 for several include directories that may be relative to sysroot.
4442 * config/i386/x-mingw32 (gplus_includedir): Define.
4443 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4444 (native_system_includedir): Likewise.
4445 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4446 override if TARGET_SYSTEM_ROOT is defined.
4447 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4449 PR tree-optimization/79090
4450 PR tree-optimization/33562
4451 PR tree-optimization/61912
4452 PR tree-optimization/77485
4453 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4454 and computed trims into the dump file.
4456 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4458 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4460 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4463 * gimplify.c (gimplify_init_constructor): If want_value and
4464 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4468 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4469 sequence. Formatting fixes.
4470 (doloop_optimize): Formatting fixes.
4473 * gcc.c (debug_level_greater_than_spec_func): New function.
4474 (static_spec_functions): Add debug-level-gt spec function.
4475 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4477 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4478 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4479 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4480 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4481 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4482 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4484 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4486 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4487 QImode fixups to general and mask registers only.
4489 2017-01-16 Carl Love <cel@us.ibm.com>
4491 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4492 for built-in functions
4493 vector signed char vec_nabs (vector signed char)
4494 vector signed short vec_nabs (vector signed short)
4495 vector signed int vec_nabs (vector signed int)
4496 vector signed long long vec_nabs (vector signed long long)
4497 vector float vec_nabs (vector float)
4498 vector double vec_nabs (vector double)
4499 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4501 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4502 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4503 * doc/extend.texi: Update the documentation file for the new built-in
4506 2017-01-16 Martin Sebor <msebor@redhat.com>
4508 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4511 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4513 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4514 UNSPEC_VSX__XXSPLTD to require special splat handling.
4516 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4519 * system.h: Poison strndup.
4521 2017-01-16 Alan Modra <amodra@gmail.com>
4524 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4527 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4529 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4531 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4533 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4534 call recog here. Assert that INSN_CODE (insn) is non-negative.
4536 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4539 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4541 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4542 in the currently scheduled RTL fragment.
4544 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4546 PR rtl-optimization/78751
4547 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4550 2017-01-14 Jeff Law <law@redhat.com>
4552 PR tree-optimization/79090
4553 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4554 variable length stores.
4555 (compute_trims): Delete dead assignment to *trim_tail.
4556 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4559 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4561 PR rtl-optimization/78626
4562 PR rtl-optimization/78727
4563 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4564 of a block, and split such blocks after everything else is finished.
4566 2017-01-14 Alan Modra <amodra@gmail.com>
4569 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4570 target legitimate_combined_insn.
4571 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4572 (rs6000_legitimate_combined_insn): New function.
4573 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4575 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4576 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4577 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4579 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4581 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4583 2017-01-13 Jeff Law <law@redhat.com>
4585 PR tree-optimization/33562
4586 PR tree-optimization/61912
4587 PR tree-optimization/77485
4588 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4590 (delete_dead_assignment): Likewise.
4591 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4592 statement to delete_dead_call and delete_dead_assignment.
4594 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4597 * substring-locations.c (format_warning_va): Strengthen case 1 so
4598 that both endpoints of the substring must be within the format
4599 range for just the substring to be printed.
4601 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4603 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4604 * config/i386/i386.c (ix86_target_string): Add missing options
4605 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4606 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4607 flags_other and ix86_target_other to flags2_other. Display unknown
4609 (ix86_valid_target_attribute_inner_p): Add missing options and
4610 reorder options by implied ISAs, as in ix86_target_string.
4612 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4614 * hash-table.h (hash_table::too_empty_p): New function.
4615 (hash_table::expand): Use it.
4616 (hash_table::traverse): Likewise.
4617 (hash_table::empty_slot): Use sizeof (value_type) instead of
4618 sizeof (PTR) to convert bytes to elements. Shrink the table
4619 if the current size is excessive for the current number of
4622 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4624 * ira-costs.c (record_reg_classes): Break from the inner loop
4625 early once alt_fail is known to be true. Update outer loop
4626 handling accordingly.
4628 2017-01-13 Jeff Law <law@redhat.com>
4630 * tree-ssa-dse.c (decrement_count): New function.
4631 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4632 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4633 when we know the partially dead statement is a mem* function.
4635 PR tree-optimization/61912
4636 PR tree-optimization/77485
4637 * tree-ssa-dse.c: Include expr.h.
4638 (maybe_trim_constructor_store): New function.
4639 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4641 PR tree-optimization/33562
4642 PR tree-optimization/61912
4643 PR tree-optimization/77485
4644 * doc/invoke.texi: Document new dse-max-object-size param.
4645 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4646 * tree-ssa-dse.c: Include params.h.
4647 (dse_store_status): New enum.
4648 (initialize_ao_ref_for_dse): New, partially extracted from
4650 (valid_ao_ref_for_dse, normalize_ref): New.
4651 (setup_live_bytes_from_ref, compute_trims): Likewise.
4652 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4653 (maybe_trim_partially_dead_store): Likewise.
4654 (maybe_trim_complex_store): Likewise.
4655 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4656 Track what bytes live from the original store. Return tri-state
4657 for dead, partially dead or live.
4658 (dse_dom_walker): Add constructor, destructor and new private members.
4659 (delete_dead_call, delete_dead_assignment): New extracted from
4661 (dse_optimize_stmt): Make a member of dse_dom_walker.
4662 Use initialize_ao_ref_for_dse.
4664 PR tree-optimization/33562
4665 PR tree-optimization/61912
4666 PR tree-optimization/77485
4667 * sbitmap.h (bitmap_count_bits): Prototype.
4668 (bitmap_clear_range, bitmap_set_range): Likewise.
4669 * sbitmap.c (bitmap_clear_range): New function.
4670 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4672 2017-01-13 Martin Liska <mliska@suse.cz>
4675 * function.c (set_cfun): Add new argument force.
4676 * function.h (set_cfun): Likewise.
4677 * ipa-inline-transform.c (inline_call): Use the function when
4678 strict alising from is dropped for function we inline to.
4680 2017-01-13 Richard Biener <rguenther@suse.de>
4682 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4683 for dumping GIMPLE INTEGER_CSTs.
4685 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4687 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4688 to 201112L since C++17.
4690 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4693 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4694 if -fsanitize=kernel-address is present.
4696 2017-01-13 Richard Biener <rguenther@suse.de>
4698 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4699 as _Literal ( type ) number in case usual suffixes do not
4700 preserve all information.
4702 2017-01-13 Richard Biener <rguenther@suse.de>
4704 PR tree-optimization/77283
4705 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4706 and ssa-iterators.h.
4707 (is_feasible_trace): Implement a cost model based on joiner
4710 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4713 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4714 char or short to __float128/_Float128 directly.
4716 2017-01-12 Martin Sebor <msebor@redhat.com>
4718 to -Wformat-overflow.
4719 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4720 (min_bytes_remaining): Same.
4721 (get_string_length): Same.
4722 (format_string): Same.
4723 (format_directive): Same.
4725 (pass_sprintf_length::handle_gimple_call): Same.
4727 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4729 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4730 info.nowrite calls with no lhs that can't throw. Return bool
4731 whether gsi_remove has been called or not.
4732 (pass_sprintf_length::handle_gimple_call): Return bool whether
4733 try_substitute_return_value called gsi_remove. Formatting fix.
4734 (pass_sprintf_length::execute): Don't use gsi_remove if
4735 handle_gimple_call returned true.
4738 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4739 be removed due to side-effects, don't remove following barrier nor
4740 turn the successor edge into fallthru edge.
4742 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4745 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4746 element-reversing loads and stores as not swappable.
4748 2017-01-12 Nathan Sidwell <nathan@acm.org>
4749 Nicolai Stange <nicstange@gmail.com>
4751 * combine.c (try_combine): Don't ignore result of overlap checking
4752 loop. Combine overlap & asm check into single loop.
4754 2017-01-12 Richard Biener <rguenther@suse.de>
4756 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4757 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4759 2017-01-12 Richard Biener <rguenther@suse.de>
4761 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4762 and TS_TARGET_OPTION directly derive from TS_BASE.
4763 * tree-core.h (tree_optimization_option): Derive from tree_base.
4764 (tree_target_option): Likewise.
4766 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4768 * config/i386/i386.c (memory_address_length): Increase len
4769 only when rip_relative_addr_p returns false.
4771 2017-01-11 Julia Koval <julia.koval@intel.com>
4773 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4774 (OPTION_MASK_ISA_SGX_SET): New.
4775 (ix86_handle_option): Handle OPT_msgx.
4776 * config.gcc: Added sgxintrin.h.
4777 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4778 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4779 * config/i386/i386.c (ix86_target_string): Add -msgx.
4781 (ix86_option_override_internal): Handle new options.
4782 (ix86_valid_target_attribute_inner_p): Add sgx.
4783 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4784 * config/i386/i386.opt: Add msgx.
4785 * config/i386/sgxintrin.h: New file.
4786 * config/i386/x86intrin.h: Add sgxintrin.h.
4788 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4791 * fold-const.c (maybe_nonzero_address): Return 1 for function
4793 (tree_single_nonzero_warnv_p): Don't handle function local objects
4797 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4800 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4803 * opts.c: Include "spellcheck.h"
4804 (struct string_fragment): New struct.
4805 (struct edit_distance_traits<const string_fragment &>): New
4807 (get_closest_sanitizer_option): New function.
4808 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4810 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4812 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4814 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4815 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4816 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4817 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4818 for initial die_offset if dwarf_split_debug_info.
4819 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4820 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4822 (output_skeleton_debug_sections): Formatting fix. Use
4823 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4824 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4826 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4828 * config/arm/cortex-a53.md: Add bypasses for
4830 (cortex_a53_r2f): Only use for transfers.
4831 (cortex_a53_f2r): Likewise.
4832 (cortex_a53_r2f_cvt): Add reservation for conversions.
4833 (cortex_a53_f2r_cvt): Likewise.
4835 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4837 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4838 to all inlined functions, change static to extern.
4840 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4843 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4845 (arm_assemble_integer): Likewise.
4847 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4849 * config.gcc: Use new awk script to check CPU, FPU and architecture
4850 parameters for --with-... options.
4851 * config/arm/parsecpu.awk: New file
4852 * config/arm/arm-cpus.in: New file.
4853 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4855 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4857 * config/arm/t-arm: Update dependency rules.
4858 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4859 of processing .def files.
4860 * config/arm/genopt.sh: Deleted.
4861 * config/arm/gentune.sh: Deleted.
4862 * config/arm/arm-cores.def: Deleted.
4863 * config/arm/arm-arches.def: Deleted.
4864 * config/arm/arm-fpus.def: Deleted.
4865 * config/arm/arm-tune.md: Regenerated.
4866 * config/arm/arm-tables.opt: Regenerated.
4867 * config/arm/arm-cpu.h: New generated file.
4868 * config/arm/arm-cpu-data.h: New generated file.
4869 * config/arm/arm-cpu-cdata.h: New generated file.
4871 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4874 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4876 (input_varpool_node): Unpack dynamically_initialized bit.
4878 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4880 PR rtl-optimization/79032
4881 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4882 the alignment of the adjusted memory reference against that of MODE,
4883 instead of the alignment of the original memory reference.
4885 2017-01-11 Martin Jambor <mjambor@suse.cz>
4887 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4889 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4890 decorated functions.
4892 2017-01-11 Richard Biener <rguenther@suse.de>
4894 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4895 set range/nonnull info for PHI results. Do not set it on
4896 stmts marked for removal.
4898 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4900 * expr.c (store_field): In the bitfield case, fetch the return value
4901 from the registers before applying a single big-endian adjustment.
4902 Always do a final load for a BLKmode value not larger than a word.
4904 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4907 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4908 that we correctly handle column numbers greater than
4909 LINE_MAP_MAX_COLUMN_NUMBER.
4911 2017-01-10 Martin Sebor <msebor@redhat.com>
4914 * gimple-ssa-sprintf.c (get_destination_size): Call
4915 {init,fini}object_sizes.
4916 * tree-object-size.c (addr_object_size): Adjust.
4917 (pass_through_call): Adjust.
4918 (pass_object_sizes::execute): Adjust.
4919 * tree-object-size.h (fini_object_sizes): Declare.
4921 2017-01-10 Martin Sebor <msebor@redhat.com>
4923 PR tree-optimization/78775
4924 * builtins.c (get_size_range): Move...
4925 * calls.c: ...to here.
4926 (alloc_max_size): Accept zero argument.
4927 (operand_signed_p): Remove.
4928 (maybe_warn_alloc_args_overflow): Call get_size_range.
4929 * calls.h (get_size_range): Declare.
4931 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4933 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4934 from TI's devices.csv file as of September 2016.
4935 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4937 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4939 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4940 * doc/invoke.texi: Likewise.
4941 * doc/md.texi: Likewise.
4942 * doc/objc.texi: Likewise.
4944 2017-01-10 Joshua Conner <joshconner@google.com>
4946 * config/arm/fuchsia-elf.h: New file.
4947 * config/fuchsia.h: New file.
4948 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4949 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4951 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4953 2016-01-10 Richard Biener <rguenther@suse.de>
4955 PR tree-optimization/79034
4956 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4957 Propagate out degenerate PHIs in the joiner.
4959 2017-01-10 Martin Liska <mliska@suse.cz>
4961 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
4962 (sort_congruence_classes_by_decl_uid): Likewise.
4963 (sort_congruence_class_groups_by_decl_uid): Likewise.
4964 (sem_item_optimizer::merge_classes): Sort class, groups in these
4965 classes and members in the groups by DECL_UID of declarations.
4966 This would make merge operations stable.
4968 2017-01-10 Martin Liska <mliska@suse.cz>
4970 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
4971 usage of m_classes_vec.
4972 (sem_item_optimizer::~sem_item_optimizer): Likewise.
4973 (sem_item_optimizer::get_group_by_hash): Likewise.
4974 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
4975 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
4976 (sem_item_optimizer::verify_classes): Likewise.
4977 (sem_item_optimizer::process_cong_reduction): Likewise.
4978 (sem_item_optimizer::dump_cong_classes): Likewise.
4979 (sem_item_optimizer::merge_classes): Likewise.
4980 * ipa-icf.h (congruence_class_hash): Rename from
4981 congruence_class_group_hash. Remove declaration of m_classes_vec.
4983 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
4985 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
4986 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
4987 * config.gcc: Add avx512vpopcntdqintrin.h.
4988 * config/i386/avx512vpopcntdqintrin.h: New.
4989 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
4990 * config/i386/i386-builtin-types.def: Add new types.
4991 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
4992 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
4993 __builtin_ia32_vpopcountq_v8di_mask): New.
4994 * config/i386/i386-c.c (ix86_target_macros_internal): Define
4995 __AVX512VPOPCNTDQ__.
4996 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
4997 (PTA_AVX512VPOPCNTDQ): Define.
4998 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
4999 TARGET_AVX512VPOPCNTDQ_P): Define.
5000 * config/i386/i386.opt: Add mavx512vpopcntdq.
5001 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5002 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5004 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5007 * predict.def (PRED_CALL): Set to 67.
5009 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5011 * expr.c (store_field): In the bitfield case, if the value comes from
5012 a function call and is of an aggregate type returned in registers, do
5013 not modify the field mode; extract the value in all cases if the mode
5014 is BLKmode and the size is not larger than a word.
5016 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5019 * config/i386/cpuid.h: Fix undefined behavior.
5021 2017-01-04 Jeff Law <law@redhat.com>
5023 PR tree-optimization/79007
5024 PR tree-optimization/67955
5025 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5026 conservative for pt.null when flag_non_call_exceptions is on.
5028 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5030 PR translation/79019
5031 PR translation/79020
5032 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5033 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5034 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5036 * config/avr/avr.opt (maccumulate-args): Likewise.
5037 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5038 * common.opt (freport-bug): Likewise.
5039 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5040 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5041 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5042 translatable string.
5043 * config/i386/i386.c (function_value_32): Likewise.
5044 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5045 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5047 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5048 * common/config/msp430/msp430-common.c (msp430_handle_option):
5050 * symtab.c (symtab_node::verify_base): Likewise.
5051 * opts.c (set_debug_level): Likewise.
5052 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5053 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5054 missing whitespace to translatable strings.
5055 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5056 * config/sh/superh.h: Likewise.
5057 * config/i386/xopintrin.h: Likewise.
5058 * config/i386/znver1.md: Likewise.
5059 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5060 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5061 * double-int.h (struct double_int): Likewise.
5062 * double-int.c (div_and_round_double): Likewise.
5063 * wide-int.cc: Likewise.
5064 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5066 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5067 (crossjumps_occurred): ... this.
5068 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5071 PR tree-optimization/78899
5072 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5073 returning bool return struct loop *, NULL for failure and the new
5075 (versionable_outer_loop_p): Don't version outer loop if it has
5076 dont_vectorized bit set.
5077 (tree_if_conversion): When versioning outer loop, ensure
5078 tree_if_conversion is performed also on the inner loop of the
5079 non-vectorizable outer loop copy.
5080 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5081 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5082 prevent vectorization of it.
5083 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5084 the outer loop vectorization of the non-scalar version is attempted
5085 before vectorization of the inner loop in scalar version. If
5086 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5087 vectorization of its inner loop.
5088 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5089 has 2 inner loops, rename also on edges from bb whose single pred
5090 is outer_loop->header. Fix typo in function comment.
5092 2017-01-09 Martin Sebor <msebor@redhat.com>
5095 * asan.c (asan_emit_stack_protection): Increase local buffer size
5096 to avoid snprintf truncation warning.
5098 2017-01-09 Andrew Pinski <apinski@cavium.com>
5100 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5101 to reference thunderx2t99 for the tuning structure
5102 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5104 (thunderx2t99_extra_costs): This.
5105 * config/aarch64/aarch64-tune.md: Regenerate.
5106 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5107 (vulcan_addrcost_table): This.
5108 (vulcan_regmove_cost): Rename to ...
5109 (thunderx2t99_regmove_cost): This.
5110 (vulcan_vector_cost): Rename to ...
5111 (thunderx2t99_vector_cost): this.
5112 (vulcan_branch_cost): Rename to ...
5113 (thunderx2t99_branch_cost): This.
5114 (vulcan_tunings): Rename to ...
5115 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5116 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5118 2017-01-09 Martin Jambor <mjambor@suse.cz>
5122 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5123 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5124 (propagate_vr_accross_jump_function): Use the above function for all
5125 value range computations for pass-through jump functions and type
5126 converasion from explicit value range values.
5127 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5128 parameters from TYPE_ARG_TYPES.
5129 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5130 (ipa_write_node_info): Stream type of the actual argument.
5131 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5133 2017-01-09 Martin Liska <mliska@suse.cz>
5136 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5137 (lookup_compiler): Do not show error message with have_E.
5139 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5141 PR tree-optimization/78938
5142 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5143 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5144 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5145 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5148 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5150 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5153 2017-01-09 Richard Biener <rguenther@suse.de>
5155 PR tree-optimization/78997
5156 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5157 name condition properly.
5159 2017-01-09 Richard Biener <rguenther@suse.de>
5162 * dwarf2out.c (is_cxx): New overload with context.
5163 (is_naming_typedef_decl): Use it.
5165 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5167 * invoke.texi (Option Summary): Correct spacing in option lists
5168 and add line breaks to fix over-long lines.
5170 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5174 * extend.texi (Common Variable Attributes): Add xref to GCC
5175 Internals manual to explain mode attribute keywords.
5177 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5180 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5181 and Preprocessor Options.
5182 (Options for Linking): Document -pthread here....
5183 (RS/6000 and PowerPC Options): ...not here.
5184 (Solaris 2 Options): ...or here.
5185 * doc/cppopts.texi: Document -pthread.
5187 2017-01-08 Martin Sebor <msebor@redhat.com>
5190 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5191 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5192 New member functions.
5193 (format_directive): Used them.
5195 (pass_sprintf_length::handle_gimple_call): Same.
5196 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5197 to avoid truncation for any argument.
5198 (extract_affine_mul): Same.
5199 * tree.c (get_file_function_name): Same.
5201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5204 * predict.def (PRED_INDIR_CALL): Set to 86.
5206 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5208 PR preprocessor/54124
5209 * doc/cppopts.texi: Reformat -d subtable to list the full name
5210 of the options. Add cross-reference to the docs for the general
5211 compiler -d options.
5212 * doc/invoke.texi (Developer Options): Add cross-reference to the
5213 preprocessor-specific -d option documentation.
5215 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5217 PR preprocessor/13498
5218 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5219 redudant material, and reflect new command-line options.
5220 (System Headers): Likewise.
5222 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5224 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5225 -isystem, and -idirafter. Copy-edit.
5226 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5227 default for -ftrack-macro-expansion. Delete obsolete and
5228 badly-formatted implementation details about -fdebug-cpp output.
5229 * doc/cppwarnopts.texi: Copy-edit.
5231 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5234 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5235 that the transition from a max line width >= 1<<10 to narrower
5236 lines works correctly.
5238 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5240 * doc/options.texi (PerFunction): New.
5241 * opt-functions.awk (switch_flags): Map both Optimization and
5242 PerFunction to CL_OPTIMIZATION.
5243 * opth-gen.awk: Test for PerFunction flag along with
5245 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5246 it only when the latter is present. Skip those that don't in
5247 the hash function generator.
5248 * common.opt (fvar-tracking): Mark as PerFunction instead of
5250 (fvar-tracking-assignments): Likewise.
5251 (fvar-tracking-assignments-toggle): Likewise.
5252 (fvar-tracking-uninit): Likewise.
5254 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5256 PR translation/79018
5257 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5260 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5263 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5264 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5265 TARGET_LONG_JUMP_TABLE_OFFSETS.
5266 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5267 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5268 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5269 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5270 * config/m68k/m68k.md (tablejump expander): Likewise.
5271 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5272 TARGET_LONG_JUMP_TABLE_OFFSETS.
5273 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5274 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5276 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5277 David Holsgrove <david.holsgrove@xilinx.com>
5279 * common/config/microblaze/microblaze-common.c
5280 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5281 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5283 * config/microblaze/microblaze.c (microblaze_must_save_register)
5284 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5286 (microblaze_eh_return): New function.
5287 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5288 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5289 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5290 * config/microblaze/microblaze.md (eh_return): New pattern.
5292 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5294 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5295 GCC_DIAGNOSTIC_STRINGIFY): Define.
5297 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5299 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5301 * config/arm/arm.md (<mcrr>): New.
5303 * config/arm/arm.c (arm_arch5te): New.
5304 (arm_option_override): Set arm_arch5te.
5305 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5307 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5308 (arm_mcrr_qualifiers): ... this. New.
5309 (MRRC_QUALIFIERS): Define to...
5310 (arm_mrrc_qualifiers): ... this. New.
5311 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5313 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5314 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5315 (MRRCI, mrrc, MRRC): New.
5316 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5317 VUNSPEC_MRRC2): New.
5319 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5321 * config/arm/arm.md (<mcr>): New.
5323 * config/arm/arm.c (arm_coproc_builtin_available): Add
5324 support for mcr, mrc, mcr2 and mrc2.
5325 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5326 (arm_mcr_qualifiers): ... this. New.
5327 (MRC_QUALIFIERS): Define to ...
5328 (arm_mrc_qualifiers): ... this. New.
5329 (MCR_QUALIFIERS): Define to ...
5330 (arm_mcr_qualifiers): ... this. New.
5331 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5333 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5334 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5335 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5338 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5340 * config/arm/arm.md (*ldc): New.
5344 * config/arm/arm.c (arm_coproc_builtin_available): Add
5345 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5346 (arm_coproc_ldc_stc_legitimate_address): New.
5347 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5348 'qualifier_const_pointer'.
5349 (LDC_QUALIFIERS): Define to...
5350 (arm_ldc_qualifiers): ... this. New.
5351 (STC_QUALIFIERS): Define to...
5352 (arm_stc_qualifiers): ... this. New.
5353 * config/arm/arm-protos.h
5354 (arm_coproc_ldc_stc_legitimate_address): New.
5355 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5356 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5357 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5358 stc2, stcl, stc2l): New.
5359 * config/arm/constraints.md (Uz): New.
5360 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5361 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5362 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5363 VUNSPEC_STC2L): New.
5365 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5367 * config/arm/arm.md (<cdp>): New.
5368 * config/arm/arm.c (neon_const_bounds): Rename this ...
5369 (arm_const_bounds): ... this.
5370 (arm_coproc_builtin_available): New.
5371 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5372 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5373 (CDP_QUALIFIERS): Define to...
5374 (arm_cdp_qualifiers): ... this. New.
5376 (arm_expand_builtin_args): Add case for 6 arguments.
5377 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5378 (arm_const_bounds): ... this.
5379 (arm_coproc_builtin_available): New.
5380 * config/arm/arm_acle.h (__arm_cdp): New.
5382 * config/arm/arm_acle_builtins.def (cdp): New.
5384 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5385 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5387 * config/arm/types.md (coproc): New.
5388 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5389 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5390 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5391 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5393 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5395 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5396 (UBINOP_QUALIFIERS): New.
5398 (acle_builtin_data): New. Change comment.
5399 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5400 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5401 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5402 arm_acle_builtins.def.
5403 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5404 (arm_init_acle_builtins): New.
5405 (CRC32_BUILTIN): Remove.
5406 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5407 crc32cb, crc32ch and crc32cw.
5408 (arm_init_crc32_builtins): Remove.
5409 (arm_init_builtins): Use arm_init_acle_builtins rather
5410 than arm_init_crc32_builtins.
5411 (arm_expand_acle_builtin): New.
5412 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5413 * config/arm/arm_acle_builtins.def: New.
5415 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5417 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5418 (arm_builtin_datum): ... this.
5419 (arm_init_neon_builtin): Rename to ...
5420 (arm_init_builtin): ... this. Add a new parameters PREFIX
5421 and USE_SIG_IN_NAME.
5422 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5423 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5424 'arm_builtin_datum'.
5425 (arm_init_vfp_builtins): Likewise.
5426 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5427 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5428 (arm_expand_neon_args): Rename to ...
5429 (arm_expand_builtin_args): ... this. Rename builtin_arg
5430 enum values and differentiate between ARG_BUILTIN_MEMORY
5431 and ARG_BUILTIN_NEON_MEMORY.
5432 (arm_expand_neon_builtin_1): Rename to ...
5433 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5434 values, arm_expand_builtin_args and add bool parameter NEON.
5435 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5436 (arm_expand_vfp_builtin): Likewise.
5437 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5439 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5442 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5443 * predict.c (tree_estimate_probability_bb): Reverse direction of
5444 polymorphic call predictor.
5446 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5448 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5449 (determine_pass_name_match): ...this new function and...
5450 (should_skip_pass_p): ...this new function.
5452 2017-01-06 Nathan Sidwell <nathan@acm.org>
5454 * ipa-visibility.c (function_and_variable_visibility): Reformat
5455 comments and long lines. Remove extrneous if.
5456 * symtab.c (symtab_node::make_decl_local): Fix code format.
5457 (symtab_node::set_section_for_node): Fix comment typo.
5459 2017-01-06 Martin Liska <mliska@suse.cz>
5462 * lra-constraints.c: Rename invariant to lra_invariant.
5463 * predict.c (set_even_probabilities): Initialize e to NULL.
5465 2017-01-05 Martin Sebor <msebor@redhat.com>
5467 PR tree-optimization/78910
5468 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5469 (format_integer): Correct off-by-one error in the handling
5470 of precision with negative numbers in signed conversions..
5472 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5474 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5476 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5478 PR tree-optimization/71016
5479 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5480 factor_out_conditional_conversion. Formatting fix.
5481 (factor_out_conditional_conversion): Add cond_stmt argument.
5482 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5483 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5486 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5488 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5489 read-rtl-function.o, and selftest-rtl.o.
5490 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5491 (selftest::aarch64_test_loading_full_dump): New function.
5492 (selftest::aarch64_run_selftests): New function.
5493 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5494 selftest::aarch64_run_selftests.
5495 * config/i386/i386.c
5496 (selftest::ix86_test_loading_dump_fragment_1): New function.
5497 (selftest::ix86_test_loading_call_insn): New function.
5498 (selftest::ix86_test_loading_full_dump): New function.
5499 (selftest::ix86_test_loading_unspec): New function.
5500 (selftest::ix86_run_selftests): Call the new functions.
5501 * emit-rtl.c (maybe_set_max_label_num): New function.
5502 * emit-rtl.h (maybe_set_max_label_num): New decl.
5503 * function.c (instantiate_decls): Guard call to
5504 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5505 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5507 * gensupport.c (gen_reader::gen_reader): Pass "false"
5508 for new "compact" param of rtx_reader.
5509 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5510 rather than an empty string for NULL strings.
5511 * read-md.c: Potentially include config.h rather than bconfig.h.
5512 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5513 (have_error): New global, copied from errors.c.
5514 (md_reader::read_name): Rename to...
5515 (md_reader::read_name_1): ...this, adding "out_loc" param,
5516 and converting "missing name or number" to returning false, rather
5518 (md_reader::read_name): Reimplement in terms of read_name_1.
5519 (md_reader::read_name_or_nil): New function.
5520 (md_reader::read_string): Handle "(nil)" by returning NULL.
5521 (md_reader::md_reader): Add new param "compact".
5522 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5523 (md_reader::read_file): New method.
5524 * read-md.h (md_reader::md_reader): Add new param "compact".
5525 (md_reader::read_file): New method.
5526 (md_reader::is_compact): New accessor.
5527 (md_reader::read_name): Convert return type from void to file_location.
5528 (md_reader::read_name_or_nil): New decl.
5529 (md_reader::read_name_1): New decl.
5530 (md_reader::m_compact): New field.
5531 (noop_reader::noop_reader): Pass "false" for new "compact" param
5533 (rtx_reader::rtx_reader): Add new "compact" param.
5534 (rtx_reader::read_rtx_operand): Make virtual and convert return
5535 type from void to rtx.
5536 (rtx_reader::read_until): New decl.
5537 (rtx_reader::handle_any_trailing_information): New virtual function.
5538 (rtx_reader::postprocess): New virtual function.
5539 (rtx_reader::finalize_string): New virtual function.
5540 (rtx_reader::m_in_call_function_usage): New field.
5541 (rtx_reader::m_reuse_rtx_by_id): New field.
5542 * read-rtl-function.c: New file.
5543 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5544 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5545 (selftest::verify_three_block_rtl_cfg): New decl.
5546 * read-rtl-function.h: New file.
5547 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5548 For host, include function.h, memmodel.h, and emit-rtl.h.
5549 (one_time_initialization): New function.
5550 (struct compact_insn_name): New struct.
5551 (compact_insn_names): New array.
5552 (find_code): Handle insn codes in compact dumps.
5553 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5554 (bind_subst_iter_and_attr): Likewise.
5555 (add_condition_to_string): Likewise.
5556 (add_condition_to_rtx): Likewise.
5557 (apply_attribute_uses): Likewise.
5558 (add_current_iterators): Likewise.
5559 (apply_iterators): Likewise.
5560 (initialize_iterators): Guard usage of apply_subst_iterator with
5561 #ifdef GENERATOR_FILE.
5562 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5563 (md_reader::read_mapping): Likewise.
5564 (add_define_attr_for_define_subst): Likewise.
5565 (add_define_subst_attr): Likewise.
5566 (read_subst_mapping): Likewise.
5567 (check_code_iterator): Likewise.
5568 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5570 (one_time_initialization): New function.
5571 (rtx_reader::read_until): New method.
5572 (read_flags): New function.
5573 (parse_reg_note_name): New function.
5574 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5575 Handle reuse_rtx ids.
5576 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5577 Add parsing support for RTL dumps, mirroring the special-cases in
5578 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5579 values, and calling handle_any_trailing_information.
5580 (rtx_reader::read_rtx_operand): Convert return type from void
5581 to rtx, returning return_rtx. Handle case 'e'. Call
5582 finalize_string on XSTR and XTMPL fields.
5583 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5584 "(nil)" values were omitted. Call the postprocess vfunc on the
5586 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5587 class ctor. Initialize m_in_call_function_usage. Call
5588 one_time_initialization.
5589 * rtl-tests.c (selftest::test_uncond_jump): Call
5590 set_new_first_and_last_insn.
5591 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5592 * selftest-rtl.c: New file.
5593 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5594 (selftest::get_insn_by_uid): New decl.
5595 * selftest-run-tests.c (selftest::run_tests): Call
5596 read_rtl_function_c_tests.
5597 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5598 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5601 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5603 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5604 operands in a special way. Assert that pos+len <= mode precision.
5606 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5608 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5609 3 argument Alias with unlimited for the negative form.
5610 (fno-vect-cost-model): Removed.
5612 2017-01-05 Martin Liska <mliska@suse.cz>
5614 * hsa-gen.c (gen_hsa_divmod): New function.
5615 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5617 2017-01-05 Martin Liska <mliska@suse.cz>
5620 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5623 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5625 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5626 small constant length operands.
5628 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5630 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5631 between loop iterations.
5633 2017-01-05 Martin Liska <mliska@suse.cz>
5636 * gimplify.c (gimplify_decl_expr): Compare to
5637 asan_poisoned_variables instread of checking flags.
5638 (gimplify_target_expr): Likewise.
5639 (gimplify_expr): Likewise.
5640 (gimplify_function_tree): Conditionally initialize
5641 asan_poisoned_variables.
5643 2017-01-04 Jeff Law <law@redhat.com>
5645 PR tree-optimizatin/78812
5646 * rtl.h (contains_mem_rtx_p): Prototype.
5647 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5648 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5649 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5650 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5651 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5653 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5655 * input.c (assert_char_at_range): Default-initialize actual_range.
5657 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5659 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5660 to match the caller.
5662 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5664 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5665 insns after final jump in test to emit dummy move.
5667 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5669 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5670 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5672 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5674 * multiple_target.c (create_dispatcher_calls): Init e_next.
5675 * tree-ssa-loop-split.c (split_loop): Init border.
5676 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5679 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5684 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5685 (altivec_register_operand): Do not return true if the operand
5686 contains a SUBREG mixing SImode and SFmode.
5687 (vsx_register_operand): Likewise.
5688 (vsx_reg_sfsubreg_ok): New predicate.
5689 (vfloat_operand): Do not return true if the operand contains a
5690 SUBREG mixing SImode and SFmode.
5691 (vint_operand): Likewise.
5692 (vlogical_operand): Likewise.
5693 (gpc_reg_operand): Likewise.
5694 (int_reg_operand): Likewise.
5695 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5696 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5697 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5699 (rs6000_emit_move_si_sf_subreg): New helper function.
5700 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5701 fixup SUBREGs involving SImode and SFmode.
5702 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5703 numbers for the new peephole2 optimization.
5704 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5705 the GLIBC math library that do AND/IOR/XOR operations on single
5706 precision floating point.
5707 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5708 target macros to say whether we need to avoid SUBREGs mixing
5710 (TARGET_ALLOW_SF_SUBREG): Likewise.
5711 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5712 (UNSPEC_SI_FROM_SF): Likewise.
5713 (iorxor): Change spacing.
5714 (and_ior_xor): New iterator for AND, IOR, and XOR.
5715 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5716 (movdi_from_sf_zero_ext): Likewise.
5717 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5718 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5719 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5720 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5721 (fms<mode>4): Likewise.
5722 (fnma<mode>4): Likewise.
5723 (fnms<mode>4): Likewise.
5724 (nfma<mode>4): Likewise.
5725 (nfms<mode>4): Likewise.
5727 2017-01-04 Marek Polacek <polacek@redhat.com>
5730 * doc/invoke.texi: Document -Wpointer-compare.
5732 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5734 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5737 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5738 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5741 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5744 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5745 documentation of the powerpc_popcntb_ok attribute.
5746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5747 code to issue warning messages if a requested CPU configuration is
5748 not supported by the binary (assembler and loader) toolchain.
5749 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5750 made to define a built-in function that has been disabled.
5751 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5752 made to define a built-in function that has been disabled.
5753 (altivec_init_builtins): Add comment explaining why definition
5754 of the DST built-in functions is not preceded by an assertion
5755 check. Add assertions to prevent ICE if attempts are made to
5756 define an altivec predicate or an abs* built-in function that has
5758 (htm_init_builtins): Add comment explaining why definition of the
5759 htm built-in functions is not preceded by an assertion check.
5761 2017-01-04 Jeff Law <law@redhat.com>
5763 PR tree-optimizatin/67955
5764 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5765 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5766 the points-to solution does not include pt_null. Use DECL_PT_UID
5769 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5771 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5772 Use gen_int_mode instead of gen_lopwart for const_int operands.
5774 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5776 PR tree-optimization/71563
5777 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5778 out of range value - has low bits known to be zero.
5780 2017-01-04 Alan Modra <amodra@gmail.com>
5782 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5783 * configure: Regenerate.
5784 * config.in: Regenerate.
5786 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5789 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5790 a substring of the message, but strcmp with the whole message. Ifdef
5791 ENABLE_NLS, translate the message first using dgettext.
5793 2017-01-03 Jeff Law <law@redhat.com>
5795 PR tree-optimizatin/78856
5796 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5797 (mark_threaded_blocks): Remove code to truncate thread paths that
5798 cross multiple loop headers. Instead invalidate the cached loop
5799 iteration information and handle case of a thread path walking
5800 into an irreducible region.
5802 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5805 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5806 assertions. Add support for doing the signbit if the IEEE 128-bit
5807 floating point value is in a GPR.
5808 * config/rs6000/rs6000.md (Fsignbit): Delete.
5809 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5810 Update the length attribute if the value is in a GPR.
5811 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5812 the sign or zero extension instruction, since the value is always 0/1.
5813 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5816 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5817 extracting SImode to a GPR register so that we can generate a
5818 store, limit the vector to be in a traditional Altivec register
5819 for the vextuwrx instruction.
5821 2017-01-03 Ian Lance Taylor <iant@google.com>
5823 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5825 2017-01-03 Martin Sebor <msebor@redhat.com>
5827 PR tree-optimization/78696
5828 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5829 precision. Use MPFR for %f for greater fidelity. Correct handling
5831 (pass_sprintf_length::compute_format_length): Set width and precision
5832 specified by asrerisk to void_node for vararg functions.
5833 (try_substitute_return_value): Adjust dump output.
5835 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5837 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5839 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5841 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5842 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5843 -mlra/-mno-lra was passed to the compiler.
5845 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5847 PR rtl-optimization/65618
5848 * emit-rtl.c (try_split): Move initialization of "before" and
5849 "after" to just before the call to emit_insn_after_setloc.
5851 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5853 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5855 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5857 * dwarf2out.c (gen_enumeration_type_die): When
5858 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5860 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5862 PR tree-optimization/78965
5863 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5864 Change first argument from const call_info & to call_info &. For %n
5865 set info.nowrite to false.
5868 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5869 possibly throwing calls.
5871 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5872 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5873 and fns handling, rather than in a separate case SSA_NAME.
5875 2017-01-02 Jeff Law <law@redhat.com>
5877 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5878 fixes for first_period and second_period variables.
5880 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5883 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5884 (*insvqi_1): New insn pattern.
5885 (*insvqi_1_mem_rex64): Ditto.
5887 (*insvqi_3): Rename from *insvqi.
5889 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5891 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5893 * doc/cfg.texi (Edges): Remove reference to Java.
5894 (Maintaining the CFG): Ditto.
5896 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5899 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5900 transparent aliases.
5902 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5905 * predict.def (PRED_CALL): Update hitrate.
5906 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5907 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5908 into direct/indirect/polymorphic variants.
5910 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5912 Update copyright years.
5914 * gcc.c (process_command): Update copyright notice dates.
5915 * gcov-dump.c (print_version): Ditto.
5916 * gcov.c (print_version): Ditto.
5917 * gcov-tool.c (print_version): Ditto.
5918 * gengtype.c (create_file): Ditto.
5919 * doc/cpp.texi: Bump @copying's copyright year.
5920 * doc/cppinternals.texi: Ditto.
5921 * doc/gcc.texi: Ditto.
5922 * doc/gccint.texi: Ditto.
5923 * doc/gcov.texi: Ditto.
5924 * doc/install.texi: Ditto.
5925 * doc/invoke.texi: Ditto.
5927 Copyright (C) 2017 Free Software Foundation, Inc.
5929 Copying and distribution of this file, with or without modification,
5930 are permitted in any medium without royalty provided the copyright
5931 notice and this notice are preserved.