1 2017-03-16 Jeff Law <law@redhat.com>
3 PR tree-optimization/71437
4 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
5 member function. Implementation moved into after_dom_children
6 member function and into the threader's thread_outgoing_edges
8 (dom_opt_dom_walker::after_dom_children): Simplify by moving
9 some code into new thread_outgoing_edges.
10 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
11 definition. Simplify marker handling (do it here). Assume we always
12 have the available expression and the const/copies tables.
13 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
15 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
16 * tree-vrp.c (equiv_stack): No longer file scoped.
17 (vrp_dom_walker): New class.
18 (vrp_dom_walker::before_dom_children): New member function.
19 (vrp_dom_walker::after_dom_children): Likewise.
20 (identify_jump_threads): Setup domwalker. Use it rather than
21 walking edges in a random order by hand. Simplify setup/finalization.
22 (finalize_jump_threads): Remove.
23 (vrp_finalize): Do not call identify_jump_threads here.
24 (execute_vrp): Do it here instead and call thread_through_all_blocks
27 PR tree-optimization/71437
28 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33 (dom_opt_dom_walker::thread_across_edge): Remove
34 handle_dominating_asserts argument. All callers changed.
35 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36 changes. Remove calls to lhs_of_dominating_assert. Other
37 uses of handle_dominating_asserts turn into unconditional code
38 (simplify_control_stmt_condition_1): Likewise.
39 (simplify_control_stmt_condition): Likewise.
40 (thread_through_normal_block, thread_across_edge): Likewise.
41 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
42 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
43 object if it is not an SSA_NAME.
44 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
45 before calling into the VRP specific simplifiers.
46 (identify_jump_threads): Remove handle_dominating_asserts
49 2017-03-16 Jakub Jelinek <jakub@redhat.com>
52 * tree-diagnostic.c (default_tree_printer): No longer static.
53 * tree-diagnostic.h (default_tree_printer): New prototype.
55 2017-03-16 Tamar Christina <tamar.christina@arm.com>
57 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
60 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
62 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
63 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
64 Use h_con constraint for operand 1.
65 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
66 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
68 2017-03-15 Jeff Law <law@redhat.com>
70 PR tree-optimization/71437
71 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
72 (record_temporary_equivalences): Use it.
74 PR tree-optimization/71437
75 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
76 tree-ssa-scopedtables.
77 (lookup_avail_expr, build_and_record_new_cond): Likewise.
78 (record_conditions, record_cond, vuse_eq): Likewise.
79 (record_edge_info): Adjust to API tweak of record_conditions.
80 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
81 (record_temporary_equivalences, optimize_stmt): Likewise.
82 (eliminate_redundant_computations): Likewise.
83 (record_equivalences_from_stmt): Likewise.
84 * tree-ssa-scopedtables.c: Include options.h and params.h.
85 (vuse_eq): New function, moved from tree-ssa-dom.c
86 (build_and_record_new_cond): Likewise.
87 (record_conditions): Likewise. Accept vector of conditions rather
88 than edge_equivalence structure for first argument.
89 for the first argument.
90 (avail_exprs_stack::lookup_avail_expr): New member function, moved
92 (avail_exprs_stack::record_cond): Likewise.
93 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
95 (avail_exprs_stack): Add new member functions lookup_avail_expr
97 (record_conditions): Declare.
99 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
102 * lra-constraints.c (process_alt_operands): Increase reject for
103 reloading an input/output operand.
105 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
108 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
109 insns to convert from signed/unsigned char/short to IEEE 128-bit
111 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
113 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
116 * config/i386/i386.c (ix86_vector_duplicate_value): Create
117 subreg of inner mode for values already in registers.
119 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
121 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
122 iteration reg is used after the loop.
124 2017-03-14 Martin Sebor <msebor@redhat.com>
126 PR tree-optimization/79800
127 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
128 precision in negative-positive range.
129 (format_floating): Call non-const overload with adjusted precision.
131 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
134 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
137 2017-03-14 Martin Sebor <msebor@redhat.com>
140 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
141 * builtins.def (aligned_alloc): Use it.
144 * Makefile.in (GTFILES): Add calls.c.
145 * calls.c: Include "gt-calls.h".
147 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
149 PR rtl-optimization/79728
150 * regs.h (struct target_regs): New field
151 x_contains_allocatable_regs_of_mode.
152 (contains_allocatable_regs_of_mode): New macro.
153 * reginfo.c (init_reg_sets_1): Initialize it, and change
154 contains_reg_of_mode so it includes global regs as well.
155 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
156 rather than contains_regs_of_mode.
158 2017-03-14 Martin Liska <mliska@suse.cz>
160 * doc/invoke.texi: Document options that can't be combined with
161 -fcheck-pointer-bounds.
163 2017-03-14 Martin Liska <mliska@suse.cz>
166 * doc/invoke.texi (-Wchkp): Document the option.
168 2017-03-14 Martin Liska <mliska@suse.cz>
170 * Makefile.in: Install gcov-dump.
172 2017-03-14 Martin Liska <mliska@suse.cz>
174 * multiple_target.c (expand_target_clones): Bail out for
175 an invalid attribute.
177 2017-03-14 Richard Biener <rguenther@suse.de>
179 * alias.c (struct alias_set_entry): Pack properly.
180 * cfgloop.h (struct loop): Likewise.
181 * cse.c (struct set): Likewise.
182 * ipa-utils.c (struct searchc_env): Likewise.
183 * loop-invariant.c (struct invariant): Likewise.
184 * lra-remat.c (struct cand): Likewise.
185 * recog.c (struct change_t): Likewise.
186 * rtl.h (struct address_info): Likewise.
187 * symbol-summary.h (function_summary): Likewise.
188 * tree-loop-distribution.c (struct partition): Likewise.
189 * tree-object-size.c (struct object_size_info): Likewise.
190 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
191 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
192 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
193 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
194 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
195 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
196 (struct _stmt_vec_info): Likewise.
198 2017-03-14 Martin Liska <mliska@suse.cz>
201 * multiple_target.c (create_dispatcher_calls): Check that
202 a target can create a function dispatcher.
204 2017-03-14 Martin Liska <mliska@suse.cz>
207 * multiple_target.c (expand_target_clones): Drop local.local
208 flag for default implementation.
210 2017-03-14 Richard Biener <rguenther@suse.de>
212 PR tree-optimization/80030
213 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
215 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
217 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
218 gcc_fallthrough() instead of __attribute__((fallthrough));
220 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
222 * doc/gcc.texi: Remove "up" link to (DIR).
223 * doc/gccint.texi: Ditto.
225 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
227 * doc/install.texi (Specific) <avr>: Remove reference to
230 2017-03-13 Jeff Law <law@redhat.com>
232 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
233 attribute rather than comments.
235 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
236 match_scratch operand is highest.
238 2017-03-13 Martin Liska <mliska@suse.cz>
241 * ipa-pure-const.c (warn_function_noreturn): If the declarations
242 is a CHKP clone, use original declaration.
244 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
246 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
247 (arc_conditional_register_usage): Use a different allocation order
248 when optimizing for size.
249 * common/config/arc/arc-common.c (arc_option_optimization_table):
250 Section anchors default on when optimizing for size.
252 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
254 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
256 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
258 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
259 * config/arc/arc.md (cpu_facility): Add cd variant.
260 (*movqi_insn): Add code density variant.
261 (*movhi_insn): Likewise.
262 (*movqi_insn): Likewise.
263 (*addsi3_mixed): Likewise.
264 (subsi3_insn): Likewise.
266 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
268 * config/arc/arc.md (movsi_cond_exec): Update constraint.
270 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
272 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
273 expressions with MINUS and UNARY ops.
275 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
278 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
280 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
281 between vec_select and vector argument.
282 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
283 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
284 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
285 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
286 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
287 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
289 2017-03-13 Richard Biener <rguenther@suse.de>
292 * params.def (vect-max-peeling-for-alignment): Fix typo.
294 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
296 * doc/install.texi (Specific) <mips-*-*>: Remove description of
297 issue that only occurred with binutils below 2.18.
299 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
301 * doc/install.texi (Specific) <cris-axis-elf>: No longer
302 refer to binutils 2.11/2.12 minimum.
304 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
306 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
307 ftp.kernel.org and simplify binutils requirement.
309 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
311 * doc/invoke.texi (Warning Options): Fix spelling of link-time
313 (Optimize Options): Ditto. Also remove redundancy.
315 2017-03-10 David Malcolm <dmalcolm@redhat.com>
318 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
320 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
321 to G_ to avoid double translation.
323 2017-03-10 David Malcolm <dmalcolm@redhat.com>
326 * auto-profile.c (get_combined_location): Convert leading
327 character of diagnostics to lower case and remove trailing period.
328 (read_profile): Likewise for various diagnostics.
329 * config/arm/arm.c (arm_option_override): Remove trailing period
330 from various diagnostics.
331 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
332 (msp430_expand_delay_cycles): Likewise.
334 2017-03-10 David Malcolm <dmalcolm@redhat.com>
337 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
338 full command-line argument, rather than just "str".
339 (aarch64_validate_march): Likewise.
340 (aarch64_validate_mtune): Likewise.
342 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
344 PR rtl-optimization/78911
345 * lra-assigns.c (must_not_spill_p): New function.
348 2017-03-10 Jakub Jelinek <jakub@redhat.com>
350 PR tree-optimization/79981
351 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
352 ATOMIC_COMPARE_EXCHANGE ifn result.
353 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
354 IFN_ATOMIC_COMPARE_EXCHANGE.
356 2017-03-10 David Malcolm <dmalcolm@redhat.com>
359 * opts.c (parse_sanitizer_options): Add missing question mark to
360 "did you mean" message.
362 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
364 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
366 (VMULEUH_UNS): Likewise.
367 (VMULOUB_UNS): Likewise.
368 (VMULOUH_UNS): Likewise.
369 * config/rs6000/rs6000.c (builtin_function_type): Remove
370 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
372 2017-03-10 David Malcolm <dmalcolm@redhat.com>
375 * read-rtl-function.c (function_reader::read_rtx_operand): Update
376 x with result of extra_parsing_for_operand_code_0.
377 (function_reader::extra_parsing_for_operand_code_0): Convert
378 return type from void to rtx, returning x. When reading
379 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
380 larger size containing struct block_symbol.
382 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
384 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
385 -mfloat128-hardware without -m64.
387 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
390 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
391 entries to the case statement that marks unsigned arguments to
392 overloaded functions.
394 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
396 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
397 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
399 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
402 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
403 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
405 2017-03-10 Martin Liska <mliska@suse.cz>
409 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
410 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
413 2017-03-10 Olivier Hainque <hainque@adacore.com>
415 * tree-switch-conversion (array_value_type): Start by resetting
416 candidate type to it's main variant.
418 2017-03-10 Jakub Jelinek <jakub@redhat.com>
420 PR rtl-optimization/79909
421 * combine.c (try_combine): Use simplify_replace_rtx on individual
422 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
423 of the whole CALL_INSN_FUNCTION_USAGE.
425 PR tree-optimization/79972
426 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
427 get_range_info on SSA_NAMEs. Formatting fixes.
429 2017-03-10 Richard Biener <rguenther@suse.de>
430 Jakub Jelinek <jakub@redhat.com>
432 PR tree-optimization/77975
433 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
435 (get_val_for): For constant x return it. Formatting fix.
436 (loop_niter_by_eval): Avoid pointless looping if the next iteration
437 would use the same bases as the current one.
439 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
441 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
442 instead of vec_select for V1TImode.
443 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
445 (VSX_LE_128): Add V1TI to this mode iterator.
446 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
447 (*vsx_le_perm_store_<mode>): Likewise.
448 (pre-reload splitter for VSX stores): Likewise.
449 (post-reload splitter for VSX stores): Likewise.
450 (*vsx_xxpermdi2_le_<mode>): Likewise.
451 (*vsx_lxvd2x2_le_<mode>): Likewise.
452 (*vsx_stxvd2x2_le_<mode>): Likewise.
454 2017-03-09 Michael Eager <eager@eagercon.com>
456 Correct failures with --enable-checking=yes,rtl.
458 * config/microblaze/microblaze.c (microblaze_expand_shift):
459 Replace GET_CODE test with CONST_INT_P and INTVAL test with
461 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
462 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
464 2017-03-09 Richard Biener <rguenther@suse.de>
466 PR tree-optimization/79977
467 * graphite-scop-detection.c (scop_detection::merge_sese):
468 Handle the case of extra exits to blocks dominating the entry.
470 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
472 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
475 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
477 PR rtl-optimization/79949
478 * lra-constraints.c (process_alt_operands): Check memory when
479 trying to predict a cycle. Print about the overall increase.
481 2017-03-09 Richard Biener <rguenther@suse.de>
484 * gimple-expr.c (useless_type_conversion_p): Preserve
485 TYPE_SATURATING for fixed-point types.
487 2017-03-09 Richard Biener <rguenther@suse.de>
490 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
491 alignment of BLKmode params.
493 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
496 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
497 (VALL_NO_V2Q): Likewise.
499 * config/aarch64/aarch64-simd.md
500 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
502 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
503 VALL_NO_V2Q mode iterator.
504 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
506 2017-03-09 Martin Liska <mliska@suse.cz>
508 PR tree-optimization/79631
509 * tree-chkp-opt.c (chkp_is_constant_addr): Call
510 tree_int_cst_sign_bit just for INTEGER constants.
512 2017-03-09 Martin Liska <mliska@suse.cz>
516 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
519 2017-03-09 Marek Polacek <polacek@redhat.com>
522 * tree.c (inchash::add_expr): Handle TREE_VEC.
524 2017-03-09 Martin Liska <mliska@suse.cz>
527 (chkp_narrow_size_and_offset): New function.
528 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
529 (void chkp_parse_bit_field_ref): New function.
530 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
531 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
533 2017-03-09 Martin Liska <mliska@suse.cz>
536 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
537 (chkp_find_bounds_1): Remove gcc_unreachable.
539 2017-03-09 Jakub Jelinek <jakub@redhat.com>
542 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
543 BUILT_IN_SYNC*, determine the access type from the size suffix and
544 always build a MEM_REF with that type. Handle forgotten
545 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
548 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
549 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
550 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
551 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
552 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
553 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
554 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
555 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
556 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
557 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
558 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
559 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
560 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
561 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
562 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
563 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
564 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
565 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
566 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
567 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
568 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
569 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
570 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
571 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
572 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
573 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
574 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
575 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
576 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
577 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
578 definitions outside of __OPTIMIZE__ guarded section.
581 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
582 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
583 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
584 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
587 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
589 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
590 ("vfenez<mode>"): Add missing constraints.
592 2017-03-08 Martin Sebor <msebor@redhat.com>
595 * config/nds32/nds32.c (nds32_option_override):
597 2017-03-08 Jakub Jelinek <jakub@redhat.com>
600 * gimplify.c (gimplify_omp_for): Replace index var in outer
601 taskloop statement with an artificial variable and add
602 OMP_CLAUSE_PRIVATE clause for it.
604 2017-03-08 Richard Biener <rguenther@suse.de>
606 PR tree-optimization/79955
607 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
608 for accesses that are completely outside of the variable.
610 2017-03-08 Andrew Haley <aph@redhat.com>
612 PR tree-optimization/79943
613 * tree-ssa-loop-split.c (compute_new_first_bound): When
614 calculating the new upper bound, (END-BEG) should be added, not
617 2017-03-08 Jakub Jelinek <jakub@redhat.com>
619 * config/avr/avr.md (setmemhi): Make sure match_dup
620 operand number comes before match_scratch.
622 2017-03-08 Richard Biener <rguenther@suse.de>
624 PR tree-optimization/79920
625 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
626 with ncopies == 1 to ...
627 (vect_transform_slp_perm_load): ... here. Properly compute
628 all element loads by iterating VF times over the group. Do
629 not handle ncopies (computed in a broken way) in
630 vect_create_mask_and_perm.
632 2017-03-08 Jakub Jelinek <jakub@redhat.com>
635 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
636 is a uniform vector, use uniform_vector_p return value instead of
637 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
639 2017-03-07 Marek Polacek <polacek@redhat.com>
642 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
643 (alloca_call_type): Likewise.
645 2017-03-07 Martin Liska <mliska@suse.cz>
647 * gcov.c (process_args): Put comment to correct location.
649 2017-03-07 Martin Liska <mliska@suse.cz>
652 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
653 Use array_at_struct_end_p instead of DECL_CHAIN (field).
654 (chkp_narrow_bounds_for_field): Likewise.
655 (chkp_parse_array_and_component_ref): Pass one more argument to
658 2017-03-07 Richard Biener <rguenther@suse.de>
660 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
663 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
665 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
666 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
668 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
671 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
672 to end of description.
673 (PARAM_MAX_STORES_TO_MERGE): Likewise.
675 2017-03-07 Jakub Jelinek <jakub@redhat.com>
677 PR rtl-optimization/79901
678 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
680 (*avx512f_<code><mode>3<mask_name>): ... this.
681 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
682 iterator instead of VI8_AVX2_AVX512BW.
684 PR rtl-optimization/79901
685 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
686 min/max expander, expand it using expand_vec_cond_expr.
689 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
692 2017-03-06 Jakub Jelinek <jakub@redhat.com>
695 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
696 to void * for PCH reasons.
697 * dwarf2out.c (output_loc_operands, output_die): Cast
698 v.val_vec.array to unsigned char *.
700 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
703 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
706 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
708 PR rtl-optimization/79571
709 * lra-constraints.c (process_alt_operands): Calculate static
710 reject and subtract it from overall when only addresses will be
713 2017-03-06 Julia Koval <julia.koval@intel.com>
716 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
717 incoming stack boundary to 128 for 64-bit targets.
719 2017-03-06 Richard Biener <rguenther@suse.de>
721 PR tree-optimization/79894
722 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
723 to NULL after folding it.
725 2017-03-06 Richard Biener <rguenther@suse.de>
727 PR tree-optimization/79824
728 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
729 check disabling peeling for gaps.
731 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
733 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
734 attributes): Document gettimeofday.
736 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
738 * config/s390/s390.c (s390_option_override_internal): Set
739 PARAM_MIN_VECT_LOOP_BOUND
741 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
743 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
744 * config/s390/s390.md: Likewise.
746 2017-03-06 Jakub Jelinek <jakub@redhat.com>
749 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
750 (<avx2_avx512>_perm<mode>): Rename to ...
751 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
753 (<avx512>_perm<mode>_mask): Rename to ...
754 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
756 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
757 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
758 instead of VI8F_256_512.
759 (avx512f_perm<mode>): New define_expand.
760 (avx512f_perm<mode>_mask): Likewise.
761 (avx512f_perm<mode>_1<mask_name>): New define_insn.
762 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
764 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
766 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
767 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
769 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
771 2017-03-06 Martin Liska <mliska@suse.cz>
774 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
775 when having a SSA NAME w/o VAR_DECL assigned to it.
777 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
779 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
780 msa_dpsub_<su>_d): Fix MODE for vec_select.
782 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
784 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
786 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
788 2017-03-06 Richard Biener <rguenther@suse.de>
790 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
791 * plugin.c (register_plugin_info): Likewise.
792 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
794 2017-03-05 Jakub Jelinek <jakub@redhat.com>
796 * config/i386/sse.md (sse_storehps, sse_storelps,
797 avx_<castmode><avxsizesuffix>_<castmode>,
798 avx512f_<castmode><avxsizesuffix>_<castmode>,
799 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
800 in condition that at least one operand is not a MEM.
802 2017-03-03 Jakub Jelinek <jakub@redhat.com>
805 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
806 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
808 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
809 gimple_call_nothrow_p flag based on whether original builtin can throw.
810 If it can, emit following stmts on the fallthrough edge.
811 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
812 don't create new bb if inserting just debug stmts on the edge, try to
813 insert them on the fallthru bb or just reset debug stmts.
815 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
818 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
819 restore recog_data (including the operand rtxes inside it) around
820 the call to get_insn_template.
822 2017-03-03 Martin Sebor <msebor@redhat.com>
824 PR tree-optimization/79699
825 * context.c (context::~context): Free MPFR caches to avoid
826 a memory leak on program exit.
828 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
830 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
831 Use wide_int::ulow () instead of .elt (0).
833 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
835 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
836 (*pushxf): Limit oF constraint to 32bit targets and add oC
837 constraint for 64bit targets.
838 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
839 (*pushdf): Change rmF constraint to rmC.
841 2017-03-03 Martin Liska <mliska@suse.cz>
843 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
844 Remove unused variable.
846 2017-03-03 Jakub Jelinek <jakub@redhat.com>
849 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
850 is a memory operand, increase num_memory.
851 (ix86_expand_args_builtin): Likewise.
853 2017-03-03 Jan Hubicka <jh@suse.cz>
856 * ipa-devirt.c (maybe_record_node): Properly handle
857 __cxa_pure_virtual visibility.
859 2017-03-03 Martin Liska <mliska@suse.cz>
861 PR tree-optimization/79803
862 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
864 (pass_loop_prefetch::execute): Disabled optimization if an
865 assumption about L1 cache size is not met.
867 2017-03-03 Martin Liska <mliska@suse.cz>
869 PR rtl-optimization/79574
870 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
871 (hash_scan_set): Likewise.
872 (dump_hash_table): Likewise.
873 (hoist_code): Likewise.
875 2017-03-03 Richard Biener <rguenther@suse.de>
877 * fixed-value.c (fixed_from_string): Restore use of elt (1)
878 in place of uhigh ().
879 (fixed_convert_from_real): Likewise.
881 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
884 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
886 2017-03-03 Richard Biener <rguenther@suse.de>
889 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
890 TYPE_OVERFLOW_UNDEFINED check.
892 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
894 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
896 (vector_ae_<mode>_p): Likewise.
897 (vector_nez_<mode>_p): Likewise.
898 (vector_ne_v2di_p): Likewise.
899 (vector_ae_v2di_p): Likewise.
900 (vector_ne_<mode>_p): Likewise.
901 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
903 (vsx_tsqrt<mode>2_fe): Likewise.
905 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
908 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
910 2017-03-02 Jakub Jelinek <jakub@redhat.com>
912 PR rtl-optimization/79780
913 * cprop.c (one_cprop_pass): When second and further conditional trap
914 in a single basic block is turned into an unconditional trap, turn it
915 into a deleted note to avoid RTL verification failures.
917 2017-03-02 Richard Biener <rguenther@suse.de>
919 * fold-const.c (const_binop): Use ulow () instead of elt (0).
921 2017-03-02 Richard Biener <rguenther@suse.de>
923 PR tree-optimization/79345
925 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
926 param and abort the walk, returning -1 if it is hit.
927 (walk_aliased_vdefs): Take a limit param and pass it on.
928 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
929 defaulting to 0 and return a signed int.
930 * tree-ssa-uninit.c (struct check_defs_data): New struct.
931 (check_defs): New helper.
932 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
933 about uninitialized memory.
934 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
935 bogus uninitialized warning.
936 (fixed_convert_from_real): Likewise.
938 2017-03-02 Bin Cheng <bin.cheng@arm.com>
940 PR tree-optimization/66768
941 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
942 iv_use if base object can't be determined.
944 2017-03-02 Jakub Jelinek <jakub@redhat.com>
946 PR tree-optimization/79345
947 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
948 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
949 (get_pattern_stats): Initialize it.
950 * genemit.c (gen_expand): Verify match_scratch numbers come after
951 match_operand/match_dup numbers.
952 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
953 match_scratch numbers.
954 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
956 * config/s390/s390.md (trunctdsd2): Likewise.
958 2017-03-02 Richard Biener <rguenther@suse.de>
960 * wide-int.h (wide_int_storage::operator=): Implement in terms
963 2017-03-02 Richard Biener <rguenther@suse.de>
965 PR tree-optimization/79777
966 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
967 the to insert expression to sth existing.
969 2017-03-01 Martin Sebor <msebor@redhat.com>
972 * gimple-ssa-sprintf.c
973 (directive::known_width_and_precision): New function.
974 (format_integer): Use it.
975 (get_mpfr_format_length): Consider the full range of precision
976 when computing %g output with the # flag. Set the likely byte
977 count to 3 rather than 1 when precision is indeterminate.
978 (format_floating): Correct the lower bound of precision.
980 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
982 * doc/invoke.texi: Document default code model for 64-bit Linux.
984 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
987 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
988 udiv rather than div since input pattern is unsigned.
990 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
992 * config/i386/i386.c (print_reg): Warn for values of
993 unsupported size in integer register.
995 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
998 * config/rs6000/predicates.md (current_file_function_operand): Do
999 not allow self calls to be local if the function is replaceable.
1001 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
1004 * config/rs6000/altivec.h (vec_ctz and others): Change the
1005 preprocessor macro that controls conditional compilation from
1006 _ARCH_PWR9 to __POWER9_VECTOR__.
1007 (vec_all_ne): Change parameterization of __altivec_scalar_pred
1008 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
1009 control (instead of _ARCH_PWR9 control) so that template
1010 definition uses power9-specific function.
1011 (vec_any_eq): Likewise.
1012 (vec_all_ne): Change macro definition to use a power9-specific
1013 expansion under #ifdef __POWER9_VECTOR__ control (instead of
1014 _ARCH_PWR9 control).
1015 (vec_any_eq) Likewise.
1016 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
1017 expansion for CMPNEF to remove support for xvcmpnesp instruction.
1018 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
1019 support for xvcmpnedp instruction.
1020 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
1021 macro expansion so that Power9 implementation of vec_all_ne does
1022 not use the AltiVec predicate framework.
1023 (VCMPNEH_P): Likewise.
1024 (VCMPNEW_P): Likewise.
1025 (VCMPNED_P): Likewise.
1026 (VCMPNEFP_P): Likewise.
1027 (VCMPNEDP_P): Likewise.
1028 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
1029 implementation of vec_any_eq to not use AltiVec predicate
1031 (VCMPAEH_P): Likewise.
1032 (VCMPAEW_P): Likewise.
1033 (VCMPAED_P): Likewise.
1034 (VCMPAEFP_P): Likewise.
1035 (VCMPAEDP_P): Likewise.
1036 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
1037 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
1038 not use the AltiVec predicate framework.
1039 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
1040 of vec_any_eq to not use AltiVec predicate framework.
1041 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1042 support for predefined __POWER9_VECTOR__ macro to indicate that
1043 Power9 instruction selection is enabled.
1044 (altivec_overloaded_builtins): Remove extraneous
1045 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
1046 function argument types RS6000_BTI_bool_V16QI and
1047 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
1048 entry for overloaded function argument types RS6000_BTI_bool_V4SI
1049 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
1050 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
1051 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
1052 Power9 for implementations of vec_cmpne. Change the signature for
1053 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
1054 (representing vec_all_ne) to remove the previously described first
1055 argument of type RS6000_BTI_INTSI, as this was an artifact of
1056 reliance on the AltiVec predicate framework, which is no longer
1057 used in the implementation of these functions. Add
1058 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
1059 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
1060 since, unlike the AltiVec predicate framework implementation, we
1061 do not share function descriptors between vec_alle and vec_anyeq.
1062 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
1063 set of modes that receive special treatment even when
1064 TARGET_P9_VECTOR is true. The special treatment emits code that
1065 does not depend on Power9 instructions.
1066 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
1067 define_expand to not rely on AltiVec predicate framework.
1068 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1070 (vector_ne_v2di_p): Change this define_expand to not rely on
1071 AltiVec predicate framework.
1072 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
1074 (vector_ne_<mode>_p): Change this define_expand to not rely on
1075 AltiVec predicate framework.
1076 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1078 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
1079 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
1080 define_insn pattern.
1081 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
1082 define_insn pattern because the xvcmpne<VSs>. instruction is not
1084 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
1085 instruction is not supported.
1087 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1089 * config/nvptx/nvptx.c: Include intl.h.
1091 2017-03-01 Martin Jambor <mjambor@suse.cz>
1094 * ipa-prop.h (ipa_bits): Removed field known.
1095 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
1096 to pointers. Adjusted their comments to warn about their sharing.
1097 (ipcp_transformation_summary): Change bits to a vector of pointers.
1098 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
1099 (ipa_get_ipa_bits_for_value): Declare.
1100 * tree-vrp.h (value_range): Mark as GTY((for_user)).
1101 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
1102 (ipa_bits_hash_table): Likewise.
1103 (ipa_vr_ggc_hash_traits): Likewise.
1104 (ipa_vr_hash_table): Likewise.
1105 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
1106 being pointers and vr_known being removed.
1107 (ipa_set_jf_unknown): Likewise.
1108 (ipa_get_ipa_bits_for_value): New function.
1109 (ipa_set_jfunc_bits): Likewise.
1110 (ipa_get_value_range): New overloaded functions.
1111 (ipa_set_jfunc_vr): Likewise.
1112 (ipa_compute_jump_functions_for_edge): Use the above functions to
1113 construct bits and vr parts of jump functions.
1114 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
1115 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1117 (ipcp_grow_transformations_if_necessary): Also allocate
1118 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1120 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
1121 them. Fix too long lines.
1122 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
1123 vr_known being removed.
1124 (ipa_read_jump_function): Use new setter functions to construct bits
1125 and vr parts of jump functions or set them to NULL.
1126 (write_ipcp_transformation_info): Adjust for bits being pointers.
1127 (read_ipcp_transformation_info): Likewise.
1128 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
1130 Include gt-ipa-prop.h.
1131 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
1133 (ipcp_store_bits_results): Likewise.
1134 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
1135 Do not write to existing jump functions but use a temporary instead.
1137 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1140 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
1141 attempt to use its first operand as BIT_FIELD_REF base.
1143 2017-03-01 Richard Biener <rguenther@suse.de>
1146 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
1147 interpolating formula in wrapping arithmetic.
1148 (chrec_apply): Convert chrec_evaluate return value to wanted type.
1150 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1152 PR tree-optimization/79734
1153 * tree-vect-generic.c (expand_vector_condition): Optimize
1154 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
1155 Handle VEC_COND_EXPR where comparison has different inner width from
1158 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
1160 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
1161 markup, and similar issues. Remove @opindex entries for things
1162 that aren't options. Add missing -mmpy-option entries.
1164 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1166 PR tree-optimization/79737
1167 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1168 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1169 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1170 instead of byte_size. Formatting fix.
1171 (shift_bytes_in_array_right): Formatting fix.
1173 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1176 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1177 condition on optimize for the leaf function test.
1179 2017-02-28 Martin Liska <mliska@suse.cz>
1182 * read-rtl-function.c (function_reader::handle_unknown_directive):
1183 Bail out when one uses -flto.
1185 2017-02-28 Martin Liska <mliska@suse.cz>
1187 * common.opt: Replace space with tabular for options of <number>
1189 * config/i386/i386.opt: Show <number> value for
1190 -mlarge-data-threshold.
1191 * opts.c (print_filtered_help): Do not display number in hexadecimal
1194 2017-02-28 Martin Liska <mliska@suse.cz>
1196 * common.opt: Fix --help=option -Q for options which are of
1199 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1201 * config/i386/i386.c (print_reg): Error out for values
1202 of 8-bit size in invalid integer register.
1204 2017-02-28 Martin Sebor <msebor@redhat.com>
1206 PR tree-optimization/79691
1207 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1209 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1212 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1213 gcc_unreachable with output_operand_lossage.
1215 2017-02-28 Richard Biener <rguenther@suse.de>
1217 PR tree-optimization/79740
1218 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1220 (visit_nary_op): Insert the nary into the hashtable if we
1221 pattern-matched sth.
1222 * tree-ssa-pre.c (eliminate_insert): Robustify.
1224 2017-02-28 Richard Biener <rguenther@suse.de>
1227 * fold-const.c (decode_field_reference): Reject out-of-bound
1230 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1232 * config/i386/i386.c: Include intl.h.
1233 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1234 instead of just cond ? "..." : "...".
1235 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1236 * coverage.c (read_counts_file): Likewise.
1237 * omp-offload.c: Include intl.h.
1238 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1239 of just cond ? "..." : "...".
1240 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1241 of just cond ? "..." : "...".
1243 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1246 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1248 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1249 'tune for' CPU name.
1250 * config/arm/arm-cpu-data.h: Regenerated.
1252 2017-02-28 Richard Biener <rguenther@suse.de>
1254 PR tree-optimization/79732
1255 * tree-inline.c (expand_call_inline): Do not shadow var.
1257 2017-02-28 Richard Biener <rguenther@suse.de>
1259 PR tree-optimization/79723
1260 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1261 address-space properly.
1263 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1265 * doc/optinfo.texi (Optimization groups): Fix option used for
1267 * doc/invoke.texi (-fopt-info): Document "omp".
1268 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1269 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1270 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1271 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1272 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1274 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1276 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1277 OPTGROUP_OMP with "omp".
1279 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1282 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1283 for arithmetic shift of unsigned V2DI.
1285 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1287 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1288 arc/linux.h headers.
1289 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1290 (LINK_SPEC): Likewise.
1291 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1292 (EXTRA_SPECS): Likewise.
1293 (STARTFILE_SPEC): Likewise.
1294 (ENDFILE_SPEC): Likewise.
1295 (LIB_SPEC): Likewise.
1296 (TARGET_SDATA_DEFAULT): Likewise.
1297 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1298 (MULTILIB_DEFAULTS): Likewise.
1299 (DWARF2_UNWIND_INFO): Likewise.
1300 * config/arc/big.h: New file.
1301 * config/arc/elf.h: Likewise.
1302 * config/arc/linux.h: Likewise.
1303 * config/arc/t-uClibc: Remove.
1305 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1307 PR tree-optimization/77536
1308 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1309 (tree_transform_and_unroll_loop): Use above function to compute the
1310 estimated niter of unrolled loop and use it when scaling profile.
1311 Also use count info rather than frequency if it's non-zero.
1312 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1313 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1314 (vect_transform_loop): Call above function.
1316 2017-02-27 Richard Biener <rguenther@suse.de>
1318 PR tree-optimization/45397
1319 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1320 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1321 (visit_nary_op): Add pattern matching for CSEing sign-changed
1322 or truncated operations with wider ones.
1324 2017-02-27 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/79690
1327 * tree-vect-stmts.c (vectorizable_store): Use vector type
1328 built from the DR with address-space.
1330 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1332 * doc/invoke.texi (Optimize Options): Refine the description
1333 of asan-use-after-return.
1335 2017-02-25 Alan Modra <amodra@gmail.com>
1337 PR rtl-optimization/79584
1338 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1339 base, not ad->base_term, the reg within base. Remove assertion
1340 that ad->base == ad->base_term. Replace gen_int_mode using
1341 bogus mode with const0_rtx.
1343 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1346 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1347 FMA_EXPR like tcc_binary or tcc_unary.
1349 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1352 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1354 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1355 (output_loc_operands): Handle DW_OP_call_ref and
1356 DW_OP_GNU_variable_value.
1357 (struct variable_value_struct): New type.
1358 (struct variable_value_hasher): Likewise.
1359 (variable_value_hash): New variable.
1360 (string_types): Remove.
1361 (copy_loc_descr): New function.
1362 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1363 (prepend_loc_descr_to_each): New function.
1364 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1365 instead of add_loc_descr_to_each if the first argument is single
1366 location list and the second has multiple.
1367 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1368 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1369 when looking for variable value which doesn't have other location info.
1370 (loc_list_from_tree): Formatting fix.
1371 (gen_array_type_die): Simplify DW_AT_string_length handling.
1372 (adjust_string_types): Remove.
1373 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1374 string_types. Call resolve_variable_values.
1375 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1376 (resolve_addr_in_expr): Likewise. Add A argument.
1377 (copy_deref_exprloc): Remove deref argument. Adjust for the
1378 original expression being DW_OP_GNU_variable_value with optionally
1379 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1380 optionally after it.
1381 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1382 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1383 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1384 (variable_value_hasher::hash, variable_value_hasher::equal): New
1386 (resolve_variable_value_in_expr, resolve_variable_value,
1387 resolve_variable_values, note_variable_value_in_expr,
1388 note_variable_value): New functions.
1389 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1392 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1395 * opts.h (handle_generated_option): Add GENERATED_P argument.
1396 * opts-common.c (handle_option): Adjust function comment.
1397 (handle_generated_option): Add GENERATED_P argument, pass it to
1399 (control_warning_option): Pass false to handle_generated_option
1401 * opts.c (maybe_default_option): Pass true to handle_generated_option
1403 * optc-gen.awk: Likewise.
1405 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1407 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1408 a REG, look at the REG it is a SUBREG of.
1409 (splitter for cmpeqsi_t): Ditto.
1411 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1413 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1414 the special USEs with the pattern of the insn, not the insn itself.
1416 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1419 * doc/invoke.texi: Document -mload-store-pairs.
1421 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1422 Sandra Loosemore <sandra@codesourcery.com>
1424 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1425 argument isn't a CONST_INT.
1426 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1427 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1428 (nios2_validate_compare): Bypass alternate compare logic if *op2
1430 (ldstwm_operation_p): Return false if first_base is not a REG or
1431 if first_offset is not a CONST_INT.
1433 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1435 * config/cris/cris.md: Use correct operand in a define_peephole2.
1437 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1439 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1443 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1444 this_insn if it is an INSN or JUMP_INSN.
1445 (force_offsettable): Look at base, not at addr.
1446 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1447 on things that aren't necessarily CONST_INTs.
1449 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1451 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1452 -mfpmath=sse is the default also for x86-32 targets with SSE2
1453 instruction set when @option{-ffast-math} is enabled
1455 2017-02-24 Jeff Law <law@redhat.com>
1457 PR rtl-optimizatoin/79286
1458 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1461 2017-02-24 Richard Biener <rguenther@suse.de>
1463 PR tree-optimization/79389
1464 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1467 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1469 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1470 function comment to reflect reality.
1471 (loop_exits_before_overflow): Fix typo in function description.
1473 2017-02-24 Richard Biener <rguenther@suse.de>
1475 PR tree-optimization/79389
1476 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1477 properly that a threading opportunity exists. Detect conditional
1478 copy/constant propagation opportunities.
1480 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1482 * config/visium/visium.md (type): Add trap.
1483 (b): New mode attribute.
1484 (*btst): Rename into...
1485 (*btst<mode>): ...this and adjust.
1486 (*cbranchsi4_btst_insn): Rename into...
1487 (*cbranch<mode>4_btst_insn): ...this and adjust.
1488 (trap): New define_insn.
1490 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1492 PR tree-optimization/79389
1493 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1494 (noce_reversed_cond_code): New function.
1495 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1496 reversed_comparison_code. Formatting fix.
1497 (noce_try_store_flag): Test rev_cond != NULL in addition to
1498 reversed_comparison_code.
1499 (noce_try_store_flag_constants): Likewise.
1500 (noce_try_store_flag_mask): Likewise.
1501 (noce_try_addcc): Use rev_cond if non-NULL instead of
1502 reversed_comparison_code.
1503 (noce_try_cmove_arith): Likewise. Formatting fixes.
1504 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1505 (noce_find_if_block): Initialize rev_cond.
1506 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1507 instead of false as last argument never attempt to reverse it
1510 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1512 PR tree-optimization/79663
1513 * tree-predcom.c (combine_chains): Process refs in reverse order
1514 only for ZERO length chains, and add explaining comment.
1516 2017-02-23 Jeff Law <law@redhat.com>
1518 PR tree-optimization/79578
1519 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1520 in call to operand_equal_p.
1522 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1525 * config/i386/cpuid.h: Fix another undefined behavior.
1527 2017-02-23 Richard Biener <rguenther@suse.de>
1529 PR tree-optimization/79683
1530 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1531 vector types for data-refs.
1533 2017-02-23 Martin Liska <mliska@suse.cz>
1535 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1537 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1540 * internal-fn.c (get_range_pos_neg): Moved to ...
1541 * tree.c (get_range_pos_neg): ... here. No longer static.
1542 * tree.h (get_range_pos_neg): New prototype.
1543 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1544 are known to be in between 0 and signed maximum inclusive, try to
1545 expand both unsigned and signed divmod and use the cheaper one from
1548 2017-02-22 Jeff Law <law@redhat.com>
1550 PR tree-optimization/79578
1551 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1552 to compare base operands.
1554 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1557 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1558 gpc_reg_operand instead of fpr_reg_operand.
1560 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1562 * config/mips/mips.c (mips_return_in_memory): Force FP
1563 vector types to be returned in memory for o32 ABI.
1565 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1567 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1568 instead of DW_TAG_member for static data member declarations and don't
1569 set no_linkage_name for static inline data members.
1570 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1573 2017-02-22 Martin Liska <mliska@suse.cz>
1575 * doc/invoke.texi: Replace inequality signs with square brackets
1578 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1580 PR tree-optimization/68644
1581 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1583 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1586 * lra-constraints.c (simplify_operand_subreg): Handle
1587 WORD_REGISTER_OPERATIONS targets.
1589 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1592 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1593 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1594 elimination by swapping fld*.
1596 2017-02-22 Richard Biener <rguenther@suse.de>
1598 PR tree-optimization/79673
1599 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1600 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1601 irrelevant address-space qualifiers and avoiding a
1602 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1604 2017-02-22 Richard Biener <rguenther@suse.de>
1606 PR tree-optimization/79666
1607 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1608 to not symbolically negate if that may introduce undefined
1611 2017-02-22 Martin Liska <mliska@suse.cz>
1614 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1615 * data-streamer-out.c (streamer_write_gcov_count_stream):
1617 * value-prof.c (stream_out_histogram_value): Make assert more
1618 precise based on type of counter.
1620 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1623 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1624 Use nonimmediate_operand instead of memory_operand for operand 1.
1625 (float-extend standard_x87sse_constant_load splitter): Ditto.
1627 2017-02-21 Jeff Law <law@redhat.com>
1629 PR tree-optimization/79621
1630 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1631 blocks with edges to themselves.
1633 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1636 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1637 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1638 Use gimple_call_builtin_p.
1641 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1642 on temporarily removed DEBUG_INSNs.
1644 PR tree-optimization/79649
1645 * tree-loop-distribution.c (classify_partition): Give up on
1646 non-generic address space loads/stores.
1648 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1650 * doc/loop.texi (Loop manipulation): Remove nonexistent
1651 tree_ssa_loop_version from the documentation.
1652 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1654 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1657 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1658 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1659 * config/rs6000/rs6000.c: Include except.h.
1660 (rs6000_expand_split_stack_prologue): Call
1661 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1663 2017-02-21 Martin Jambor <mjambor@suse.cz>
1666 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1669 2017-02-21 Martin Jambor <mjambor@suse.cz>
1671 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1672 for backward compatibility only.
1673 * doc/invoke.texi (Option Summary): Remove all references to
1676 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1680 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1682 * lra-constraints.c (curr_insn_transform): Handle
1683 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1685 2017-02-21 Martin Liska <mliska@suse.cz>
1687 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1689 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1691 PR translation/79638
1692 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1694 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1697 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1698 (arm_function_ok_for_sibcall): Return false for an indirect call by
1699 descriptor if all the argument registers are used.
1700 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1701 alignment of the function.
1703 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1705 PR tree-optimization/61441
1706 * simplify-rtx.c (simplify_const_unary_operation): For
1707 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1708 the sNaN unmodified.
1710 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1712 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1713 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1714 instead of SYSTEM_HEADER_DIR.
1716 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1717 Martin Liška <mliska@suse.cz>
1719 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1720 Fix typos and grammar, use active voice, and clarify.
1722 2017-02-20 Marek Polacek <polacek@redhat.com>
1725 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1728 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1730 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1733 * config/i386/i386.c (ix86_expand_builtin): Handle
1734 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1735 ix86_builtins_isa[fcode].isa as a requirement of those
1736 flags and any other flag in the bitmask.
1737 (ix86_init_mmx_sse_builtins): Use 0 instead of
1738 ~OPTION_MASK_ISA_64BIT as mask.
1739 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1740 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1741 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1742 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1744 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1747 * lra-constraints.c (split_reg): Check requested split mode
1748 is supported by the register.
1750 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1752 * lra-constraints.c (simplify_operand_subreg): Remove early
1755 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1758 * lra-constraints.c (curr_insn_transform): Tighten condition
1759 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1761 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1764 * lra-constraints.c (curr_insn_transform): Handle
1765 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1767 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1770 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1772 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1774 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1777 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1780 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1782 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1785 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1787 * final.c (last_columnnum, override_columnnum): New variables.
1788 (final_start_function): Set last_columnnum, pass it to begin_prologue
1789 hook and pass 0 to dwarf2out_begin_prologue.
1790 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1791 to source_line debug hook.
1792 (notice_source_line): Compute last_columnnum and for debug_column_info
1793 return true on column changes.
1794 * debug.h (struct gcc_debug_hooks): Add column argument to
1795 source_line and begin_prologue hooks.
1796 (debug_nothing_int_charstar_int_bool): Remove prototype.
1797 (debug_nothing_int_int_charstar,
1798 debug_nothing_int_int_charstar_int_bool): New prototypes.
1799 (dwarf2out_begin_prologue): Add column argument.
1800 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1801 begin_prologue hooks.
1802 (debug_nothing_int_charstar_int_bool): Remove.
1803 (debug_nothing_int_int_charstar,
1804 debug_nothing_int_int_charstar_int_bool): New functions.
1805 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1806 through to dwarf2out_source_line.
1807 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1808 (dwarf2out_source_line): Add column argument, emit it if requested.
1809 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1811 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1812 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1813 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1814 through to dwarf2out_begin_prologue.
1815 (vmsdbgout_source_line): Add column argument, pass it through to
1816 dwarf2out_source_line.
1817 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1818 dbxout_source_line caller.
1819 (dbxout_source_line): Add column argument.
1821 * common.opt (gno-column-info, gcolumn-info): New options.
1822 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1823 (check_die): Also test for multiple DW_AT_decl_column attributes.
1824 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1825 DW_AT_decl_column if requested.
1826 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1828 (gen_variable_die): Likewise.
1829 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1830 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1833 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1834 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1835 (ix86_handle_option): Handle OPT_m3dnowa.
1836 * doc/invoke.texi (-m3dnowa): Document.
1837 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1838 -m3dnowa instead of -m3dnow -march=athlon.
1841 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1842 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1844 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1847 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1848 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1849 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1850 generator for vsx_xxpermdi_<mode>_be.
1851 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1852 force big-endian semantics.
1853 (vsx_xxpermdi_<mode>_be): New define_expand with same
1854 implementation as previous version of vsx_xxpermdi_<mode>.
1856 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1858 PR tree-optimization/79327
1859 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1860 variable, its initialization and use.
1862 2017-02-17 Julia Koval <julia.koval@intel.com>
1864 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1865 (OPTION_MASK_ISA_PKU_UNSET): New.
1866 (ix86_handle_option): Handle -mrdpid.
1867 * config/i386/cpuid.h (bit_RDPID): New.
1868 * config/i386/driver-i386.c (host_detect_local_cpu):
1869 Detect RDPID feature.
1870 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1871 * config/i386/i386-c.c (ix86_target_macros_internal):
1873 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1874 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1875 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1876 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1877 * config/i386/i386.md (define_insn "rdpid"): New.
1878 * config/i386/i386.opt Add -mrdpid.
1879 * config/i386/immintrin.h (_rdpid_u32): New.
1881 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1883 PR rtl-optimization/79541
1884 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1885 instead of transforming it into USE.
1887 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1889 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1890 If HONOR_SNANS (SFmode) force the input to a register.
1891 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1892 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1893 an frsp or similar insn.
1895 2017-02-17 Martin Liska <mliska@suse.cz>
1897 PR rtl-optimization/79577
1898 * params.def (selsched-max-sched-times): Increase minimum to 1.
1900 2017-02-17 Martin Liska <mliska@suse.cz>
1902 PR rtl-optimization/79574
1903 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1905 2017-02-17 Martin Liska <mliska@suse.cz>
1907 PR tree-optimization/79529
1908 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1909 ssa_defined_default_def_p to handle cases which are implicitly
1911 * tree-ssa.c (ssa_defined_default_def_p): New function.
1912 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1913 which are implicitly defined.
1914 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1916 2017-02-17 Richard Biener <rguenther@suse.de>
1919 * params.def (max-ssa-name-query-depth): Limit to 10.
1921 2017-02-17 Richard Biener <rguenther@suse.de>
1923 PR tree-optimization/79552
1924 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1927 2017-02-17 Richard Biener <rguenther@suse.de>
1930 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1932 2017-02-17 Marek Polacek <polacek@redhat.com>
1935 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1936 (fold_negate_expr): New wrapper.
1938 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1940 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1941 Correct terminology and de-emphasize pre-standard behavior.
1943 2017-02-16 Alan Modra <amodra@gmail.com>
1945 PR rtl-optimization/79286
1946 * ira.c (def_dominates_uses): New function.
1947 (update_equiv_regs): Don't create an equivalence for insns that
1948 may trap where the register def does not dominate the use.
1950 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1952 PR rtl-optimization/78127
1953 * lra.c (lra): Call lra_eliminate before finish the loop after
1956 2017-02-16 Richard Biener <rguenther@suse.de>
1958 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1960 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1961 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1962 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1963 (isl_val_int_from_wi): New function.
1964 (extract_affine_gmp): Rename to ...
1965 (extract_affine_wi): ... this, take a widest_int.
1966 (extract_affine_int): Just wrap extract_affine_wi.
1967 (add_param_constraints): Use isl_val_int_from_wi.
1968 (add_loop_constraints): Likewise, and extract_affine_wi.
1970 2017-02-15 Jeff Law <law@redhat.com>
1973 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1974 ira_init_register_move_cost_if_necessary.
1976 2017-02-15 Martin Sebor <msebor@redhat.com>
1979 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1980 removed in a prior commit.
1982 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1984 PR tree-optimization/79347
1985 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1986 counters during peeling.
1988 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1990 * Makefile.in (site.exp): Remove "set ISLVER".
1992 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1995 * real.c (real_from_integer): Call real_convert even for decimal.
1997 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
2000 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
2002 2017-02-14 Andrew Pinski <apinski@cavium.com>
2004 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
2005 cores and change the partno/implementer to be correct.
2006 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
2007 the 'B" as the implementer.
2008 * config/aarch64/aarch64-tune.md: Regenerate.
2010 2017-02-14 Carl Love <cel@us.ibm.com>
2012 * config/rs6000/rs6000.c: Add case statement entry to make the
2013 xvcvuxdsp built-in argument unsigned.
2014 * config/rs6000/vsx.md: Fix the source and return operand types so they
2015 match the instruction definitions from the ISA document. Fix typo
2016 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
2019 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
2022 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
2023 member early_clobber_alts.
2024 * lra-lives.c (reg_early_clobber_p): New.
2025 (process_bb_lives): Use it.
2026 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
2027 (debug_operand_data): Initialize early_clobber_alts.
2028 (setup_operand_alternative): Set up early_clobber_alts.
2029 (collect_non_operand_hard_regs): Ditto. Pass early clobber
2030 alternatives to new_insn_reg.
2031 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
2033 (lra_update_insn_regno_info): Pass the new arg.
2035 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2038 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
2039 (new_oacc_loop_raw): Don't clear already cleared fields.
2042 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
2043 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
2044 _mm512_prefetch_i64gather_ps): New inline functions and macros.
2046 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
2049 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
2051 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
2054 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
2055 the extra instruction to the right place to store 128-bit constant
2058 2017-02-14 Martin Sebor <msebor@redhat.com>
2061 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
2062 warning for strings of unknown length.
2064 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
2066 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
2068 2017-02-14 Jeff Law <law@redhat.com>
2071 * ira-costs.c (scan_one_insn): Initialize register move costs
2072 for pseudos seen in USE/CLOBBER insns.
2074 PR tree-optimization/79095
2075 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
2076 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
2077 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
2078 if the operands are known to be not equal, then the resulting range
2080 (intersect_ranges): If the new range is ~[0,0] and the old range is
2081 wide, then prefer ~[0,0].
2082 * tree-vrp.c (overflow_comparison_p_1): New function.
2083 (overflow_comparison_p): New function.
2084 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
2085 if NAME is used in an overflow test.
2086 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
2087 overflow check that can be expressed as an equality test, then adjust
2088 ops to be that equality test.
2090 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2092 * config/s390/s390-builtin-types.def: Remove flags argument.
2093 * config/s390/s390.c (s390_init_builtins): Likewise.
2095 2017-02-14 Martin Liska <mliska@suse.cz>
2097 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
2098 vector. Fix trailing white spaces.
2100 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
2102 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
2105 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2107 PR rtl-optimization/68664
2108 * config/arm/arm.c (arm_sched_can_speculate_insn):
2109 New function. Declare prototype.
2110 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2112 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2114 PR rtl-optimization/68664
2115 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
2117 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2119 2017-02-14 Amit Pawar <amit.pawar@amd.com>
2121 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
2122 max skip bytes for function, loop and jump.
2124 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2126 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
2127 ABS_EXPR for gimple dump.
2129 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2132 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
2134 PR tree-optimization/79408
2135 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
2136 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
2137 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
2138 also if rhs1 is INTEGER_CST.
2140 2017-02-14 Richard Biener <rguenther@suse.de>
2143 * tree-into-ssa.c (insert_phi_nodes): When the function can
2144 have abnormal edges rewrite SSA names with broken use-def
2145 dominance out of SSA and register them for PHI insertion.
2147 2017-02-13 Martin Sebor <msebor@redhat.com>
2150 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
2151 clearing info.nowrite flag when snprintf size argument is a range.
2153 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2155 * cprop.c (cprop_jump): Add missing space in string literal.
2156 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
2157 (get_constraint_for_component_ref): Likewise.
2158 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
2159 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
2160 * lra-constraints.c (process_alt_operands): Likewise.
2161 * ipa-inline.c (inline_small_functions): Likewise.
2162 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
2163 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2164 * trans-mem.c (diagnose_tm_1_op): Likewise.
2165 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
2166 (grid_parallel_clauses_gridifiable): Likewise.
2168 * config/nvptx/mkoffload.c (process): Add space in between
2171 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2172 "MOD4_SSE_REGS" and "ALL_REGS".
2174 * spellcheck.c (test_data): Add , in between "foo" and "food".
2176 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2179 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2180 boundary crossing check and subsequent code generation agree.
2182 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2184 * config/aarch64/aarch64.c (has_memory_op): Delete.
2185 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2188 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2190 PR rtl-optimization/79388
2191 PR rtl-optimization/79450
2192 * combine.c (distribute_notes): When removing TEM_INSN for which
2193 corresponding dest has last value recorded, invalidate that last
2196 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2198 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2199 of explicit '@'. Add missing assembly comment marker on branch costs
2202 2017-02-13 Nathan Sidwell <nathan@acm.org>
2204 * gengtype-lex.l (<in_struct>): Add '/'.
2206 2017-02-13 Martin Liska <mliska@suse.cz>
2209 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2211 2017-02-13 Richard Biener <rguenther@suse.de>
2213 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2215 * configure: Re-generate.
2216 * config.in: Likewise.
2217 * graphite-dependences.c: Simplify as if
2218 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2219 * graphite-isl-ast-to-gimple.c: Likewise.
2220 * graphite-optimize-isl.c: Likewise.
2221 * graphite-poly.c: Likewise.
2222 * graphite-sese-to-poly.c: Likewise.
2223 * graphite.h: Likewise.
2224 * toplev.c: Include isl/version.h and use isl_version () for
2225 printing the ISL version.
2226 * doc/install.texi: Update ISL requirement.
2228 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2230 * doc/standards.texi (Standards): Update reference to
2233 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2235 * doc/extend.texi (Named Address Spaces): sourceware.org now
2237 * doc/install.texi (Binaries): Ditto.
2240 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2242 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2243 standard terminology "stringize"/"stringizing" throughout.
2244 * doc/cppinternals.texi: Likewise.
2246 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2248 * doc/extend.texi: Fix some spelling mistakes and typos.
2249 * doc/invoke.texi: Likewise.
2251 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2254 * params.def (inline-min-speedup) Change from 10 to 8.
2256 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2258 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2261 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2264 * ipa-inline-analysis.c (get_minimal_bb): New function.
2265 (record_modified): Use it.
2266 (remap_edge_change_prob): Handle also ancestor functions.
2268 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2270 * doc/contrib.texi (Contributors): Remove broken link into
2271 the Mauve CVS repository.
2273 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2276 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2277 result computation whenever lhs doesn't have vector mode, not
2278 just when it has BLKmode.
2280 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2282 * doc/makefile.texi (profiledbootstrap): Refer to the
2283 installation instructions only in textual form.
2285 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2288 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2290 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2292 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2293 (Specific): Update mingw-w64 reference.
2295 (Specific): Remove broken link to Renesas RX processor.
2297 2017-02-10 Richard Biener <rguenther@suse.de>
2299 * toplev.c (process_options): Do not mention obsolete graphite
2300 options when printing sorry message about missing graphite support.
2301 Mention -floop-nest-optimize.
2303 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2305 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2306 (vtst_p16): Likewise.
2307 (vtstq_p8): Likewise.
2308 (vtstq_p16): Likewise.
2310 (vtstq_p64): Likewise.
2311 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2312 (vset_lane_p64): New.
2313 (vsetq_lane_p64): New.
2315 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2317 PR tree-optimization/79411
2318 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2319 stmt operands are SSA_NAMEs used in abnormal phis.
2320 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2323 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2326 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2329 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2331 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2333 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2335 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2339 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2340 not arbitrary TREE_CONSTANT.
2343 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2344 "omp declare target link" attribute unless is_global_var.
2345 * omp-offload.c (find_link_var_op): Likewise.
2347 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2348 Chung-Lin Tang <cltang@codesourcery.com>
2350 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2352 (gimplify_adjust_omp_clauses): Don't delete TILE.
2353 (gimplify_omp_for): Deal with TILE.
2354 * internal-fn.c (expand_GOACC_TILE): New function.
2355 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2357 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2358 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2360 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2361 avoid DIV for outermost collapse var.
2362 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2363 Remove out of date comments, fix whitespace.
2364 * omp-general.c (omp_extract_for_data): Deal with tiling.
2365 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2366 adjust OLF_DIM_BASE value.
2367 (struct omp_for_data): Add tiling field.
2368 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2369 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2370 for auto loops. Remove default auto determining, moved to
2371 oacc_loop_fixed_partitions.
2372 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2373 stmts, add e_mask field.
2374 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2375 (oacc_thread_numbers): Use oacc_dim_call.
2376 (oacc_xform_tile): New.
2377 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2378 (finish_oacc_loop): Adjust for ifns vector.
2379 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2380 add case for GOACC_TILE fns.
2381 (oacc_loop_xform_loop): Delete.
2382 (oacc_loop_process): Iterate over call list directly, and add
2383 handling for GOACC_TILE fns.
2384 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2386 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2387 vector partitioning to outer loops. Assign 2 partitions to loops
2388 when available. Add TILE handling.
2389 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2390 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2391 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2392 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2393 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2395 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2397 * configure.ac (ACX_BUGURL): Update.
2398 * configure: Regenerate.
2400 2017-02-09 Richard Biener <rguenther@suse.de>
2402 PR tree-optimization/69823
2403 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2404 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2406 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2408 * config/arc/arc-c.def: Add __NPS400__ definition.
2409 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2410 (TARGET_NPS400): Define.
2412 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2414 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2416 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2418 (arc_cpu_types): Fill the arch_info field with a pointer into the
2419 arc_arch_types table.
2420 (arc_selected_cpu): Declare.
2421 * config/arc/arc.c (arc_selected_cpu): Make global.
2422 (arc_selected_arch): Delete.
2423 (arc_base_cpu): Delete.
2424 (arc_override_options): Remove references to deleted variables,
2425 update access to arch information.
2426 (ARC_OPT): Update access to arch information.
2427 (ARC_OPTX): Likewise.
2428 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2429 (TARGET_ARC600): Update access to arch information.
2430 (TARGET_ARC601): Likewise.
2431 (TARGET_ARC700): Likewise.
2432 (TARGET_EM): Likewise.
2433 (TARGET_HS): Likewise.
2434 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2437 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2440 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2441 condition/operands for integer GE/LE/GEU/LEU operations.
2443 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2445 PR translation/79397
2446 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2449 2017-02-08 Martin Jambor <mjambor@suse.cz>
2452 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2453 whether allocation happened.
2454 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2455 nothing was allocated.
2457 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2459 PR tree-optimization/79408
2460 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2461 constant, but SSA_NAME with a known integer range, use the minimum
2462 of that range instead of op1 to determine if modulo can be replaced
2463 with its first operand.
2465 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2467 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2469 2017-02-08 Richard Biener <rguenther@suse.de>
2471 PR tree-optimization/71824
2472 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2473 Check all loops contained in the merged region.
2475 2017-02-07 Andrew Pinski <apinski@cavium.com>
2477 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2479 2017-02-07 Andrew Pinski <apinski@cavium.com>
2481 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2482 (thunderxt88): Likewise.
2483 (thunderxt81): Disable LSE and change v8.1 to v8.
2484 (thunderxt83): Likewise.
2486 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2487 Richard Biener <rguenther@suse.de>
2490 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2491 type from int to size_t.
2492 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2494 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2496 PR rtl-optimization/79386
2497 * cprop.c (bypass_conditional_jumps): Initialize
2498 bypass_last_basic_block already before splitting bbs after
2499 unconditional traps...
2500 (bypass_conditional_jumps): ... rather than here.
2503 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2504 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2505 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2506 fix -masm=intel patterns.
2508 2017-02-07 Richard Biener <rguenther@suse.de>
2510 PR tree-optimization/79256
2512 * builtins.c (get_object_alignment_2): Use min_align_of_type
2513 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2514 and ADJUST_FIELD_ALIGN.
2516 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2518 * doc/tm.texi: Regenerate.
2519 * stor-layout.c (layout_decl): Adjust.
2520 (update_alignment_for_field): Likewise.
2521 (place_field): Likewise.
2522 (min_align_of_type): Likewise.
2523 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2524 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2525 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2526 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2527 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2528 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2529 * config/i386/i386.c (x86_field_alignment): Likewise.
2530 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2531 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2532 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2533 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2534 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2535 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2539 2017-01-30 Richard Biener <rguenther@suse.de>
2541 PR tree-optimization/79256
2542 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2543 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2546 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2548 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2549 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2550 builtins to SImode and emit a zero-extend, if necessary.
2552 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2554 * docs/invoke.texi (RISC-V Options): Alphabetize.
2556 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2558 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2561 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2563 * config/riscv/riscv.c: New file.
2564 * gcc/common/config/riscv/riscv-common.c: Likewise.
2565 * config.gcc: Likewise.
2566 * config/riscv/constraints.md: Likewise.
2567 * config/riscv/elf.h: Likewise.
2568 * config/riscv/generic.md: Likewise.
2569 * config/riscv/linux.h: Likewise.
2570 * config/riscv/multilib-generator: Likewise.
2571 * config/riscv/peephole.md: Likewise.
2572 * config/riscv/pic.md: Likewise.
2573 * config/riscv/predicates.md: Likewise.
2574 * config/riscv/riscv-builtins.c: Likewise.
2575 * config/riscv/riscv-c.c: Likewise.
2576 * config/riscv/riscv-ftypes.def: Likewise.
2577 * config/riscv/riscv-modes.def: Likewise.
2578 * config/riscv/riscv-opts.h: Likewise.
2579 * config/riscv/riscv-protos.h: Likewise.
2580 * config/riscv/riscv.h: Likewise.
2581 * config/riscv/riscv.md: Likewise.
2582 * config/riscv/riscv.opt: Likewise.
2583 * config/riscv/sync.md: Likewise.
2584 * config/riscv/t-elf-multilib: Likewise.
2585 * config/riscv/t-linux: Likewise.
2586 * config/riscv/t-linux-multilib: Likewise.
2587 * config/riscv/t-riscv: Likewise.
2588 * configure.ac: Likewise.
2589 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2590 Waterman as RISC-V maintainers.
2591 * doc/install.texi: Add RISC-V entries.
2592 * doc/invoke.texi: Add RISC-V options section.
2593 * doc/md.texi: Add RISC-V constraints section.
2594 * configure: Regenerated.
2596 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2599 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2600 false values to be constant vectors with all 0 or all 1 bits set.
2601 (vcondu<mode><mode>): Likewise.
2602 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2604 (fpmask_comparison_operator): Update comment.
2605 (vecint_comparison_operator): New predicate.
2606 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2607 vector conditionals when the true and false values are constant
2608 vectors with all 0 bits or all 1 bits set.
2610 2017-02-06 Martin Sebor <msebor@redhat.com>
2612 PR tree-optimization/79376
2613 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2615 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2617 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2618 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2619 to simplify split condition.
2621 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2623 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2624 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2627 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2629 PR rtl-optimization/68664
2630 * target.def (can_speculate_insn): New hook.
2631 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2632 * doc/tm.texi: Regenerate.
2633 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2634 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2635 (rs6000_sched_can_speculate_insn): New function.
2637 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2639 PR tree-optimization/79284
2640 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2641 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2642 vectorizable_mask_load_store, vectorizable_operation,
2643 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2644 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2645 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2646 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2647 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2648 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2649 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2650 is_gimple_assign (stmt). Replace another such test with
2651 is_gimple_assign (stmt).
2653 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2656 * config/avr/avr.c (rtl-iter.h): Include it.
2657 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2658 (avr_legitimate_combined_insn): ...and implementation.
2660 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2662 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2663 * config/s390/s390.c (s390_const_operand_ok)
2664 (s390_canonicalize_comparison, s390_extract_part)
2665 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2666 (s390_contiguous_bitmask_p, s390_rtx_costs)
2667 (legitimize_pic_address): Likewise.
2668 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2669 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2670 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2671 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2672 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2674 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2676 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2677 REGNO($0) == REGNO($1).
2679 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2681 * config/s390/linux.h(SIZE_TYPE): Add comment.
2683 2017-02-06 Julian Brown <julian@codesourcery.com>
2684 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2685 Virendra Pathak <virendra.pathak@broadcom.com>
2687 * config/aarch64/aarch64-cores.def: Change the scheduler
2689 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2690 * config/aarch64/thunderx2t99.md: New file.
2692 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2694 * doc/standards.texi (Go Language): Update link to language
2697 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2699 * tree-eh.c (lower_resx): Sanitize profile.
2700 (cleanup_empty_eh_move_lp): Likewise.
2702 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2705 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2707 * cfgloopmanip.h (loop_version): Update prototype.
2708 * modulo-sched.c (sms_schedule): Update call of loop_version.
2709 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2710 * tree-parloops.c (gen_parallel_loop): Likewise.
2711 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2712 * tree-ssa-loop-split.c (split_loop): Likewise.
2713 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2714 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2716 2017-02-05 Martin Liska <mliska@suse.cz>
2719 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2721 (print_operand_address): Initialize a struct to zero.
2723 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2725 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2726 garbage collector only in textual form.
2728 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2730 * doc/extend.texi (x86 specific memory model extensions for
2731 transactional memory): Simplify a phrase.
2733 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2736 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2737 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2738 (atomic_storedi_1): Likewise.
2740 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2742 PR tree-optimization/79338
2743 * tree-parloops.c (gather_scalar_reductions): Don't call
2744 vect_analyze_loop_form for loop->inner before destroying loop's
2747 2017-02-03 Martin Sebor <msebor@redhat.com>
2749 PR tree-optimization/79327
2750 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2751 when precision has resulted in leading zeros.
2752 (format_integer): Adjust the likely counter to assume an unknown
2753 argument that may be zero is non-zero.
2755 2017-02-03 Jason Merrill <jason@redhat.com>
2758 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2759 avoid copying non-taken branch.
2761 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2763 PR tree-optimization/79340
2764 * tree-vect-loop.c (vectorizable_reduction): Release
2765 vec_defs elements after safe_splicing them into other vectors.
2768 PR tree-optimization/79327
2769 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2770 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2772 (format_integer): Use wide_int_to_tree instead of build_int_cst
2773 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2774 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2775 of shortest and longest sequence.
2777 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2779 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2780 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2782 2017-02-03 Walter Lee <walt@tilera.com>
2785 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2786 after initial stackframe link reg save.
2787 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2789 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2792 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2793 wu for stxssp alternative.
2795 2017-02-03 Martin Sebor <msebor@redhat.com>
2797 PR tree-optimization/79352
2798 * gimple-fold.c (get_range_strlen): Add argument.
2799 (get_range_strlen): Change return type to bool.
2800 (get_maxval_strlen): Pass in a dummy argument.
2801 * gimple-fold.h (get_range_strlen): Change return type to bool.
2802 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2803 * tree.h (array_at_struct_end_p): Add argument.
2804 * tree.c (array_at_struct_end_p): Handle it.
2806 2017-02-03 Martin Liska <mliska@suse.cz>
2809 * multiple_target.c (create_dispatcher_calls): Redirect edge
2810 from a caller of a dispatcher.
2811 (expand_target_clones): Make the clones local.
2812 (ipa_target_clone): Do both target clones and resolvers.
2813 (ipa_dispatcher_calls): Remove the pass.
2814 (pass_dispatcher_calls::gate): Likewise.
2815 (make_pass_dispatcher_calls): Likewise.
2816 * passes.def (pass_target_clone): Put as very first IPA early
2819 2017-02-03 Martin Liska <mliska@suse.cz>
2821 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2822 in case of a function with ifunc attribute.
2824 2017-02-03 Martin Liska <mliska@suse.cz>
2826 * cgraph.c (cgraph_node::dump): Dump function version info.
2827 * symtab.c (symtab_node::dump_base): Add missing new line.
2829 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2831 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2832 (ifcombine_ifandif): Use it.
2834 2017-02-03 Martin Liska <mliska@suse.cz>
2836 * doc/invoke.texi: Document default value for
2837 use-after-scope-direct-emission-threshold.
2839 2017-02-03 Martin Liska <mliska@suse.cz>
2841 PR tree-optimization/79339
2842 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2843 (format_floating): Likewise.
2845 2017-02-03 Martin Liska <mliska@suse.cz>
2848 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2850 (ipa_node_params_t::remove): Likewise.
2851 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2852 initialization from removed ipa_node_params_t::insert.
2853 (ipa_node_params::~ipa_node_params): Move from removed
2854 ipa_node_params_t::release.
2855 * symbol-summary.h (symbol_summary::m_released): New member.
2856 Do not release a summary twice. Do not allow to call finalizer
2857 for types of a summary that live in GGC memory.
2859 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2861 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2864 2017-02-02 Martin Sebor <msebor@redhat.com>
2867 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2868 (format_string): Tighten up the range of output for non-constant
2869 strings and correct the expected range for wide non-constant strings.
2871 2017-02-02 Martin Sebor <msebor@redhat.com>
2873 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2876 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2878 (-fdump-tree-@var): Add to index and document how to come up
2879 with pass-specific option and dump file names.
2880 (-fdump-passes): Clarify where to look for output.
2882 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2885 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2886 statistics of the analyzed path; allow threading for speed when
2887 any of BBs along the path are optimized for speed.
2889 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2892 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2893 settings of the virtual registers.
2896 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2898 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2899 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2902 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2904 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2905 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2907 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2909 * config/s390/s390.md: Add missing comments with the expanded
2911 * config/s390/vector.md: Likewise.
2912 * config/s390/vx-builtins.md: Likewise.
2914 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2917 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2918 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2919 conditions on a single line.
2921 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2923 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2924 __S390_VX__ to __VX__.
2926 2017-02-01 Andrew Pinski <apinski@cavium.com>
2928 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2929 stmt_info to record_stmt_cost.
2930 (vect_get_known_peeling_cost): Pass stmt_info if known to
2932 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2933 cpu_vector_cost field into
2934 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2935 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2936 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2937 splitting of scalar_stmt_cost and vec_stmt_cost.
2938 (thunderx_vector_cost): Likewise.
2939 (cortexa57_vector_cost): LIkewise.
2940 (exynosm1_vector_cost): Likewise.
2941 (xgene1_vector_cost): Likewise.
2942 (thunderx2t99_vector_cost): Improve after the splitting of the two
2944 (aarch64_builtin_vectorization_cost): Update for the splitting of
2945 scalar_stmt_cost and vec_stmt_cost.
2947 2017-02-01 Torvald Riegel <triegel@redhat.com>
2948 Richard Henderson <rth@redhat.com>
2950 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2951 conditional on existance of a fast atomic load.
2952 * optabs-query.c (can_atomic_load_p): New function.
2953 * optabs-query.h (can_atomic_load_p): Declare it.
2954 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2955 no fast atomic load is available for the particular size of access.
2956 (expand_atomic_compare_and_swap): Likewise.
2957 (expand_atomic_load): Likewise.
2958 (expand_atomic_store): Likewise.
2959 (expand_atomic_fetch_op): Likewise.
2960 * testsuite/lib/target-supports.exp
2961 (check_effective_target_sync_int_128): Remove x86 because it provides
2962 no fast atomic load.
2963 (check_effective_target_sync_int_128_runtime): Likewise.
2965 2017-02-01 Richard Biener <rguenther@suse.de>
2967 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2968 (graphite_transform_loops): Provide opt-info for optimized nests.
2969 * tree-parloop.c (parallelize_loops): Provide opt-info for
2972 2017-02-01 Richard Biener <rguenther@suse.de>
2975 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2978 2017-02-01 Richard Biener <rguenther@suse.de>
2980 PR tree-optimization/71824
2981 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2982 Verify the loops are valid in the merged SESE region.
2983 (scop_detection::can_represent_loop_1): Check analyzing the
2984 evolution of the number of iterations in the region succeeds.
2986 2017-01-31 Ian Lance Taylor <iant@golang.org>
2988 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2989 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2991 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2993 PR preprocessor/79210
2994 * input.c (get_substring_ranges_for_loc): Replace line_width
2995 assertion with error-handling.
2997 2017-01-31 Richard Biener <rguenther@suse.de>
2999 PR tree-optimization/77318
3000 * graphite-sese-to-poly.c (extract_affine): Fix assert.
3001 (create_pw_aff_from_tree): Take loop parameter.
3002 (add_condition_to_pbb): Pass loop of the condition to
3003 create_pw_aff_from_tree.
3005 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3007 * config/s390/s390.c (s390_asan_shadow_offset): New function.
3008 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
3010 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3014 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
3016 (convert_int_to_float128): Likewise.
3017 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
3018 (convert_int_to_float128): Likewise.
3019 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
3020 (UNSPEC_IEEE128_CONVERT): Likewise.
3021 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
3022 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
3023 Use local variables for IBM extended format.
3024 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
3025 (fix_trunc<mode>si2_fprs): Likewise.
3026 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
3027 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
3028 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
3029 to know that we can now have integers of all sizes in vector
3031 (fix<uns>_<mode>di2_hw): Likewise.
3032 (float<uns>_<mode>si2_hw): Likewise.
3033 (fix_<mode>si2_hw): Likewise.
3034 (fixuns_<mode>si2_hw): Likewise.
3035 (float<uns>_<mode>di2_hw): Likewise.
3036 (float_<mode>di2_hw): Likewise.
3037 (float_<mode>si2_hw): Likewise.
3038 (floatuns_<mode>di2_hw): Likewise.
3039 (floatuns_<mode>si2_hw): Likewise.
3040 (xscvqp<su>wz_<mode>): Delete, no longer used.
3041 (xscvqp<su>dz_<mode>): Likewise.
3042 (xscv<su>dqp_<mode>): Likewise.
3043 (ieee128_mfvsrd_64bit): Likewise.
3044 (ieee128_mfvsrd_32bit): Likewise.
3045 (ieee128_mfvsrwz): Likewise.
3046 (ieee128_mtvsrw): Likewise.
3047 (ieee128_mtvsrd_64bit): Likewise.
3048 (ieee128_mtvsrd_32bit): Likewise.
3050 2017-01-31 Martin Liska <mliska@suse.cz>
3053 * ipa-prop.c (ipa_free_all_node_params): Call release method
3054 instead of ~sumbol_summary to not to trigger double times
3057 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
3059 PR tree-optimization/71691
3060 * bitmap.h (class auto_bitmap): New.
3061 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
3062 is_maybe_undefined instead of ssa_undefined_value_p.
3064 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3066 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3067 __S390_ARCH_LEVEL__ to __ARCH__.
3069 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3071 PR tree-optimization/79267
3072 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
3073 if should_remove_lhs_p is true.
3075 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
3078 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
3079 (add_alignment_attribute): New.
3080 (base_type_die): Add alignment attribute.
3081 (subrange_type_die): Likewise.
3082 (modified_type_die): Likewise.
3083 (gen_array_type_die): Likewise.
3084 (gen_descr_array_type_die: Likewise.
3085 (gen_enumeration_type_die): Likewise.
3086 (gen_subprogram_die): Likewise.
3087 (gen_variable_die): Likewise.
3088 (gen_field_die): Likewise.
3089 (gen_ptr_to_mbr_type_die): Likewise.
3090 (gen_struct_or_union_type_die): Likewise.
3091 (gen_subroutine_type_die): Likewise.
3092 (gen_typedef_die): Likewise.
3093 (base_type_cmp): Compare alignment attribute.
3095 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3098 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
3099 (setb_unsigned) New pattern for setb with CCUNS.
3100 * config/rs6000/rs6000.c (expand_block_compare): Use a different
3101 subfc./subfe sequence to avoid overflow problems. Generate a
3102 shorter sequence with cmpld/setb for power9.
3103 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
3104 for generating subfc. instruction.
3105 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
3106 now uses this instruction.
3108 2017-01-30 Ian Lance Taylor <iant@google.com>
3111 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
3112 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
3114 2017-01-30 Martin Sebor <msebor@redhat.com>
3116 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
3117 Move constant to the right of a relational operator.
3118 (get_mpfr_format_length, format_character, format_string): Ditto.
3119 (should_warn_p, maybe_warn): Same.
3121 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
3123 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3126 * asan.c (get_translation_unit_decl): Remove function.
3127 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
3129 2017-01-30 Martin Liska <mliska@suse.cz>
3131 PR gcov-profile/79259
3132 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
3135 2017-01-30 Martin Liska <mliska@suse.cz>
3138 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
3139 Initialize variables with NULL value.
3141 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
3144 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
3146 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
3148 2017-01-30 Richard Biener <rguenther@suse.de>
3150 PR tree-optimization/79276
3151 * tree-vrp.c (process_assert_insertions): Properly adjust common
3152 when removing a duplicate.
3154 * gcc.dg/torture/pr79276.c: New testcase.
3156 2017-01-30 Richard Biener <rguenther@suse.de>
3158 PR tree-optimization/79256
3159 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3160 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3162 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
3164 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3167 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3168 ("*r<noxa>sbg_<mode>_sll_bitmask")
3169 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3170 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3171 Use contiguous_bitmask_nowrap_operand.
3173 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3176 * config/rs6000/altivec.h (vec_xl): Revise #define.
3177 (vec_xst): Likewise.
3179 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3181 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3183 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3185 PR rtl-optimization/79194
3186 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3187 traps before call to bypass_conditional_jumps.
3189 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3191 PR tree-optimization/71374
3192 * lra-constraints.c (check_conflict_input_operands): New.
3193 (match_reload): Use it.
3195 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3198 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3199 account to calculate conflict_set.
3201 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3203 PR rtl-optimization/78559
3204 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3205 other_insn in combine.
3207 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3209 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3210 uint16_type_node for BT_UINT16.
3212 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3214 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3215 "RTL Tests" to menu.
3216 (GIMPLE Tests): New node.
3217 (RTL Tests): New node.
3219 2017-01-27 Richard Biener <rguenther@suse.de>
3221 PR tree-optimization/79245
3222 * tree-loop-distribution.c (distribute_loop): Apply cost
3223 modeling also to detected patterns.
3225 2017-01-27 Richard Biener <rguenther@suse.de>
3227 PR tree-optimization/71433
3228 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3229 (compare_assert_loc): New function.
3230 (process_assert_insertions): Sort and optimize assert locations
3231 to remove duplicates and push down identical assertions on
3232 edges to their destination block.
3234 2017-01-27 Richard Biener <rguenther@suse.de>
3236 PR tree-optimization/79244
3237 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3238 out SSA names even if abnormal.
3240 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3242 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3243 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3244 instead of MPFR_RNDN.
3246 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3249 * arm.c (arm_option_override): Don't call build_target_option_node
3250 until after doing all option overrides.
3251 (arm_valid_target_attribute_tree): Likewise.
3253 2017-01-27 Martin Liska <mliska@suse.cz>
3255 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3256 for {cd}tors and C++ {cd}tors.
3258 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3260 * config/s390/s390.md ("*setmem_long_and")
3261 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3263 2017-01-26 Martin Sebor <msebor@redhat.com>
3265 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3268 2017-01-26 Martin Sebor <msebor@redhat.com>
3270 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3271 HAVE_DFmode before using XFmode or DFmode.
3272 (parse_directive): Avoid using the z length modifier to avoid
3273 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3276 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3277 to accept adjustment as an array.
3278 (get_int_range): New function.
3279 (struct directive): Make width and prec arrays.
3280 (directive::set_width, directive::set_precision): Call get_int_range.
3281 (format_integer, format_floating): Handle width and precision ranges.
3282 (format_string, parse_directive): Same.
3284 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3287 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3288 comdat_type_p set, just clone them, but keep the children in the
3292 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3293 which have direct callers with -fvar-tracking-assignments enabled
3295 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3296 inside of type units.
3298 2017-01-26 Martin Sebor <msebor@redhat.com>
3301 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3303 (struct format_result): Replace number_chars, number_chars_min,
3304 and number_chars_max with a single member of struct result_range.
3306 (format_result::operator+=): Adjust.
3307 (struct fmtresult): Remove bounded. Handle likely and unlikely
3309 (fmtresult::adjust_for_width_or_precision): New function.
3310 (fmtresult:type_max_digits): New function.
3311 (bytes_remaining): Handle likely and unlikely counters.
3312 (min_bytes_remaining): Remove.
3313 (format_percent): Simplify.
3314 (format_integer, format_floating): Set likely and unlikely counters.
3315 (get_string_length, format_character, format_string): Same.
3316 (format_plain, should_warn_p): New function.
3317 (maybe_warn): Call should_warn_p. Update diagnostic messages
3318 and handle those for all directives, including plain strings.
3319 (format_directive): Handle likely and unlikely counters.
3320 Remove unnecessary quoting from diagnostics. Add an informational
3322 (add_bytes): Remove.
3323 (pass_sprintf_length::compute_format_length): Simplify.
3324 (try_substitute_return_value): Handle likely and unlikely counters.
3326 2017-01-26 Carl Love <cel@us.ibm.com>
3328 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3329 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3331 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3334 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3335 endianess for subregs into account.
3336 * lra-constraints.c (lra_constraints): Do risky transformations
3337 always on the first iteration.
3338 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3339 last_call_used_reg_set.
3340 (process_bb_lives): Define and use last_call_used_reg_set.
3341 * lra.c (lra): Always continue after lra_constraints on the first
3344 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3346 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3348 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3350 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3352 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3353 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3354 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3355 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3356 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3357 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3358 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3359 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3360 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3362 2017-01-26 Marek Polacek <polacek@redhat.com>
3365 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3366 for the third operand.
3368 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3371 * omp-low.c (struct omp_context): Add simt_stmt field.
3372 (scan_omp_for): Return omp_context *.
3373 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3374 context to the _simt_ SIMD stmt.
3375 (lower_omp_for): For combined SIMD with sibling _simt_
3376 SIMD, make sure to use the same decls in _looptemp_
3377 clauses as in the sibling.
3379 2017-01-26 David Sherwood <david.sherwood@arm.com>
3382 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3385 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3388 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3389 emit fld b; fld a; if possible.
3391 * brig-builtins.def: Update copyright years.
3392 * config/arm/arm_acle_builtins.def: Update copyright years.
3394 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3397 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3398 constraint instead of o for the stxsd instruction.
3400 2017-01-25 Carl Love <cel@us.ibm.com>
3402 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3403 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3405 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3407 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3409 2017-01-25 Richard Biener <rguenther@suse.de>
3411 PR tree-optimization/69264
3412 * target.def (vector_alignment_reachable): Improve documentation.
3413 * doc/tm.texi: Regenerate.
3414 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3416 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3417 earlier changes with respect to TYPE_USER_ALIGN.
3418 (vector_alignment_reachable_p): Likewise. Improve dumping.
3420 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3423 * config/arm/arm.md (xordi3): Force constant operand into a register
3426 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3428 * doc/invoke.texi (-fstore-merging): Correct default optimization
3429 levels at which it is enabled.
3430 (-O): Move -fstore-merging from list to...
3433 2017-01-25 Richard Biener <rguenther@suse.de>
3436 * omp-expand.c: Include debug.h.
3437 (expand_omp_taskreg): Make sure to generate early debug before
3438 outlining anything from a function.
3439 (expand_omp_target): Likewise.
3440 (grid_expand_target_grid_body): Likewise.
3442 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3445 * asan.c (get_translation_unit_decl): New function.
3446 (asan_add_global): Extract modules file name from globals
3447 TRANSLATION_UNIT_DECL name.
3449 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3452 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3453 for long calls with APCS frame and VFP.
3455 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3457 * cfg.c (original_copy_tables_initialized_p): New function.
3458 * cfg.h (original_copy_tables_initialized_p): New decl.
3459 * cfgrtl.c (relink_block_chain): Guard the call to
3460 free_original_copy_tables with a call to
3461 original_copy_tables_initialized_p.
3462 * cgraph.h (symtab_node::native_rtl_p): New decl.
3463 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3464 (symtab_node::needed_p): Don't assert for early assembly output
3465 for __RTL functions.
3466 (cgraph_node::finalize_function): Set "force_output" for __RTL
3468 (cgraph_node::analyze): Bail out early for __RTL functions.
3469 (analyze_functions): Update assertion to support __RTL functions.
3470 (cgraph_node::expand): Bail out early for __RTL functions.
3471 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3473 * function.h (struct function): Update comment for field
3475 * gimple-expr.c: Include "tree-pass.h".
3476 (gimple_has_body_p): Return false for __RTL functions.
3477 * Makefile.in (OBJS): Add run-rtl-passes.o.
3478 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3480 (gcc::pass_manager::get_clean_slate): New accessor.
3481 * passes.c: Include "insn-addr.h".
3482 (should_skip_pass_p): Add logging. Update logic for running
3483 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3484 property-provider override so it is only done for gimple passes.
3486 (skip_pass): New function.
3487 (execute_one_pass): Call skip_pass when skipping passes.
3488 * read-md.c (md_reader::read_char): Support filtering
3489 the input to a subset of line numbers.
3490 (md_reader::md_reader): Initialize fields
3491 m_first_line and m_last_line.
3492 (md_reader::read_file_fragment): New function.
3493 * read-md.h (md_reader::read_file_fragment): New decl.
3494 (md_reader::m_first_line): New field.
3495 (md_reader::m_last_line): New field.
3496 * read-rtl-function.c (function_reader::create_function): Only
3497 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3498 curr_properties. Set DECL_INITIAL to a dummy block.
3499 (read_rtl_function_body_from_file_range): New function.
3500 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3502 * run-rtl-passes.c: New file.
3503 * run-rtl-passes.h: New file.
3505 2017-01-24 Jeff Law <law@redhat.com>
3507 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3510 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3512 PR tree-optimization/79159
3513 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3514 (record_nonwrapping_iv): Improve boundary using above function if no
3515 value range information.
3517 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3518 Martin Jambor <mjambor@suse.cz>
3520 * brig-builtins.def: New file.
3521 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3522 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3523 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3524 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3525 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3526 * builtin-types.def (BT_INT8): New.
3527 (BT_INT16): Likewise.
3528 (BT_UINT8): Likewise.
3529 (BT_UINT16): Likewise.
3530 (BT_FN_ULONG): Likewise.
3531 (BT_FN_UINT_INT): Likewise.
3532 (BT_FN_UINT_ULONG): Likewise.
3533 (BT_FN_UINT_LONG): Likewise.
3534 (BT_FN_UINT_PTR): Likewise.
3535 (BT_FN_ULONG_PTR): Likewise.
3536 (BT_FN_INT8_FLOAT): Likewise.
3537 (BT_FN_INT16_FLOAT): Likewise.
3538 (BT_FN_UINT32_FLOAT): Likewise.
3539 (BT_FN_UINT16_FLOAT): Likewise.
3540 (BT_FN_UINT8_FLOAT): Likewise.
3541 (BT_FN_UINT64_FLOAT): Likewise.
3542 (BT_FN_UINT16_UINT32): Likewise.
3543 (BT_FN_UINT32_UINT16): Likewise.
3544 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3545 (BT_FN_INT_PTR_INT): Likewise.
3546 (BT_FN_UINT_PTR_UINT): Likewise.
3547 (BT_FN_LONG_PTR_LONG): Likewise.
3548 (BT_FN_ULONG_PTR_ULONG): Likewise.
3549 (BT_FN_VOID_UINT64_UINT64): Likewise.
3550 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3551 (BT_FN_INT8_INT8_INT8): Likewise.
3552 (BT_FN_INT16_INT16_INT16): Likewise.
3553 (BT_FN_INT_INT_INT): Likewise.
3554 (BT_FN_UINT_FLOAT_UINT): Likewise.
3555 (BT_FN_FLOAT_UINT_UINT): Likewise.
3556 (BT_FN_ULONG_UINT_UINT): Likewise.
3557 (BT_FN_ULONG_UINT_PTR): Likewise.
3558 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3559 (BT_FN_UINT_UINT_UINT): Likewise.
3560 (BT_FN_VOID_UINT_PTR): Likewise.
3561 (BT_FN_UINT_UINT_PTR: Likewise.
3562 (BT_FN_UINT32_UINT64_PTR): Likewise.
3563 (BT_FN_INT_INT_UINT_UINT): Likewise.
3564 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3565 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3566 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3567 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3568 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3569 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3570 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3571 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3572 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3573 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3574 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3575 * doc/frontends.texi: List BRIG FE.
3576 * doc/install.texi (Testing): Add BRIG tesring requirements.
3577 * doc/invoke.texi (Overall Options): Mention BRIG.
3578 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3580 2017-01-24 Richard Biener <rguenther@suse.de>
3582 PR translation/79208
3583 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3585 2017-01-24 Martin Jambor <mjambor@suse.cz>
3588 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3589 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3592 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3595 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3596 casts from signed to unsigned really don't have a range.
3598 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3600 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3601 GMP_RNDx for compatiblity.
3603 2017-01-24 Martin Liska <mliska@suse.cz>
3606 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3607 that would prevent us to call alloca with -1 as argument.
3609 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3611 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3612 Avoid -Wformat-security warning.
3614 2017-01-23 Andrew Pinski <apinski@cavium.com>
3616 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3619 2017-01-23 Martin Sebor <msebor@redhat.com>
3622 * gimple-ssa-sprintf.c (warn_level): New global.
3623 (format_integer): Use it here and throughout the rest of the file.
3624 Use the same switch to compute sign as base.
3625 (maybe_warn): New function.
3626 (format_directive): Factor out warnings into maybe_warn.
3627 Add debugging output. Use warn_level.
3628 (add_bytes): Use warn_level.
3629 (pass_sprintf_length::compute_format_length): Add debugging output.
3630 (try_substitute_return_value): Same.
3631 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3634 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3635 (struct fmtresult, format_integer, format_floating): Adjust.
3636 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3637 (get_string_length, format_string,format_directive): Same.
3638 (pass_sprintf_length::compute_format_length): Same.
3639 (try_substitute_return_value): Simplify slightly.
3642 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3643 (fmtresult::operator+=): Outlined.
3644 (struct fmtresult): Add ctors.
3645 (struct conversion_spec): Rename...
3646 (struct directive): ...to this. Add and remove data members.
3647 (directive::set_width, directive::set_precision): New functions.
3648 (format_percent): Use fmtresult ctor.
3649 (get_width_and_precision): Remove.
3650 (format_integer): Make naming changes. Avoid computing width and
3652 (format_floating): Same. Adjust indentation.
3653 (format_character, format_none): New functions.
3654 (format_string): Moved character handling to format_character.
3655 (format_directive): Remove arguments, change return type.
3656 (parse_directive): New function.
3657 (pass_sprintf_length::compute_format_length): Move directive
3658 parsing to parse_directive.
3660 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3662 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3663 (assign_assembler_name_if_needed): ... this.
3664 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3665 (assign_assembler_name_if_needed): ... this.
3666 (free_lang_data_in_cgraph): Adjust callers.
3667 * cgraphunit.c (cgraph_node::analyze): Likewise.
3668 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3671 2017-01-23 Richard Biener <rguenther@suse.de>
3673 PR tree-optimization/79088
3674 PR tree-optimization/79188
3675 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3676 resetting loop bounds after last path deletion. Reset loop
3677 bounds of the target loop, make code match the comments.
3678 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3679 Make sure loops need no fixups.
3681 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3683 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3684 exponent support with double type for first argument.
3685 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3686 type returned by __builtin_vec_extract_sig,
3687 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3688 functions from "vector int" to "vector unsigned int" or from
3689 "vector long long int" to "vector unsigned long long int".
3690 Changed type returned by __builtin_vec_extract_exp,
3691 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3692 functions from "vector int" to "vector unsigned int" or from
3693 "vector long long int" to "vector unsigned long long int".
3694 Changed return type of __builtin_vec_test_data_class,
3695 __builtin_vec_test_data_class_sp, and
3696 __builtin_vec_test_data_class_dp from "vector int" to
3697 "vector bool int" or from "vector long long int" to "vector bool
3698 long long int" and changed second argument type from "unsigned
3699 int" to "int". Added new overloaded function forms "vector float
3700 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3701 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3702 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3703 double, vector unsigned long long int)" and "vector double
3704 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3705 long int)". Changed return type of
3706 __builtin_scalar_test_data_class and
3707 __builtin_scalar_test_data_class_sp and
3708 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3709 int" and changed second argument from "unsigned int" to "int".
3710 Changed type returned by __builtin_scalar_test_neg,
3711 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3712 from "int" to "bool int". Added new overloaded function form
3713 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3714 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3715 exponent double-precision with floating point first argument.
3716 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3717 documentation of scalar_test_data_class, scalar_test_neg,
3718 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3719 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3720 vec_test_data_class built-in functions to reflect refinements in
3721 their type signatures.
3723 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3725 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3727 (aarch64_elf_asm_destructor): Likewise.
3729 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3731 PR rtl-optimization/78634
3732 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3733 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3734 * ifcvt.c (noce_try_cmove): Add missing cost check.
3736 PR rtl-optimization/71724
3737 * combine.c (if_then_else_cond): Look for situations where it is
3738 beneficial to undo the work of one of the recursive calls.
3740 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3742 PR tree-optimization/70754
3743 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3744 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3745 combined stmt before it if not NULL.
3746 (combine_chains): Process refs reversely and compute dominance point
3749 2017-01-23 Martin Liska <mliska@suse.cz>
3751 PR tree-optimization/79196
3752 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3753 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3755 (strlen_optimize_stmt): Call the renamed function.
3757 2017-01-23 Michael Matz <matz@suse.de>
3759 PR tree-optimization/78384
3760 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3762 2017-01-23 Richard Biener <rguenther@suse.de>
3764 PR tree-optimization/79186
3765 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3766 both incoming edges before moving an assert.
3768 2017-01-23 Martin Jambor <mjambor@suse.cz>
3770 * ipa-prop.c (load_from_param_1): Removed.
3771 (load_from_unmodified_param): Bits from load_from_param_1 put back
3773 (load_from_param): Removed.
3774 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3775 with stmt. Reverted back to use of load_from_unmodified_param.
3777 2017-01-23 Martin Jambor <mjambor@suse.cz>
3780 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3781 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3782 field a pointer to garbage collected vector, mark lattices and
3783 ipcp_orig_node with GTY((skip)).
3784 (ipa_get_param_count): Adjust to descriptors being a pointer.
3785 (ipa_get_param): Likewise.
3786 (ipa_get_type): Likewise.
3787 (ipa_get_param_move_cost): Likewise.
3788 (ipa_set_param_used): Likewise.
3789 (ipa_get_controlled_uses): Likewise.
3790 (ipa_set_controlled_uses): Likewise.
3791 (ipa_is_param_used): Likewise.
3792 (ipa_node_params_t): Move into garbage collector. New methods insert
3794 (ipa_node_params_sum): Annotate wth GTY(()).
3795 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3797 (ipa_load_from_parm_agg): Adjust declaration.
3798 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3799 * ipa-profile.c (ipa_profile): Likewise.
3800 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3801 (ipa_populate_param_decls): Make descriptors parameter garbage
3803 (ipa_dump_param): Adjust to descriptors being a pointer.
3804 (ipa_alloc_node_params): Likewise.
3805 (ipa_initialize_node_params): Likewise.
3806 (load_from_param_1): Make descriptors parameter garbage collected.
3807 (load_from_unmodified_param): Likewise.
3808 (load_from_param): Likewise.
3809 (ipa_load_from_parm_agg): Likewise.
3810 (ipa_node_params::~ipa_node_params): Removed.
3811 (ipa_free_all_node_params): Remove call to delete operator.
3812 (ipa_node_params_t::insert): New.
3813 (ipa_node_params_t::remove): Likewise.
3814 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3815 copy known_csts and known_contexts vectors.
3816 (ipa_read_node_info): Adjust to descriptors being a pointer.
3817 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3819 (ipcp_transform_function): Make descriptors variable garbage
3822 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3824 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3825 * config/i386/avx512dqintrin.h: Ditto.
3826 * config/i386/avx512fintrin.h: Ditto.
3827 * gcc/config/i386/i386.c: Handle new builtins.
3828 * config/i386/i386-builtin.def: Add new builtins.
3829 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3830 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3832 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3833 Martin Liska <mliska@suse.cz>
3835 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3836 * asan.c (asan_expand_poison_ifn): Support stores and use
3837 appropriate ASAN report function.
3838 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3839 * internal-fn.def (ASAN_POISON_USE): Declare.
3840 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3841 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3842 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3843 ASAN_POISON calls w/o LHS.
3844 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3845 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3846 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3847 * gimplify.c (asan_poison_variables): Add attribute
3848 use_after_scope_memory to variables that really needs to live
3850 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3851 having the attribute.
3853 2017-01-23 Martin Liska <mliska@suse.cz>
3855 * asan.c (create_asan_shadow_var): New function.
3856 (asan_expand_poison_ifn): Likewise.
3857 * asan.h (asan_expand_poison_ifn): New declaration.
3858 * internal-fn.c (expand_ASAN_POISON): Likewise.
3859 * internal-fn.def (ASAN_POISON): New builtin.
3860 * sanopt.c (pass_sanopt::execute): Expand
3861 asan_expand_poison_ifn.
3862 * tree-inline.c (copy_decl_for_dup_finish): Make function
3864 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3865 * tree-ssa.c (is_asan_mark_p): New function.
3866 (execute_update_addresses_taken): Rewrite local variables
3867 (identified just by use-after-scope as addressable) into SSA.
3869 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3871 * doc/install.texi (Specific): opensource.apple.com uses https
3872 now. Remove trailing slash.
3874 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3876 * README.Portability: Remove note on an Irix compatibility issue.
3878 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3880 * gcov.c (INCLUDE_ALGORITHM): Define.
3881 (INCLUDE_VECTOR): Define.
3882 No longer include <vector> and <algorithm> directly.
3884 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3886 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3888 * doc/invoke.texi (Code Gen Options): Ditto.
3890 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3893 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3895 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3897 rtl-optimization/79125
3898 * cprop.c (local_cprop_pass): Handle cases where we make an
3901 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3905 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3906 read from, for big endian.
3908 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3910 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3911 register pauth builtins for LP64 only.
3913 2017-01-20 Marek Polacek <polacek@redhat.com>
3916 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3919 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3921 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3923 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3924 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3925 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3927 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3930 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3931 in big-endian mode when they are not a single duplicated value.
3933 2017-01-20 Richard Biener <rguenther@suse.de>
3935 * BASE-VER: Bump to 7.0.1.
3937 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3939 * omp-low.c (omplow_simd_context): New struct. Use it...
3940 (lower_rec_simd_input_clauses): ...here and...
3941 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3942 references to idx, lane, max_vf, is_simt.
3944 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3946 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3949 2017-01-20 Martin Jambor <mjambor@suse.cz>
3951 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3952 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3954 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3955 (GTFILES): Rename hsa.c to hsa-common.c.
3956 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3957 * hsa-dump.c: Likewise.
3958 * hsa-gen.c: Likewise.
3959 * hsa-regalloc.c: Likewise.
3960 * ipa-hsa.c: Likewise.
3961 * omp-expand.c: Likewise.
3962 * omp-low.c: Likewise.
3963 * toplev.c: Likewise.
3965 2017-01-20 Marek Polacek <polacek@redhat.com>
3968 * doc/invoke.texi: Document -Wduplicated-branches.
3969 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3970 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3971 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3972 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3973 return 0 only when not OEP_LEXICOGRAPHIC.
3974 (fold_build_cleanup_point_expr): Use the expression
3975 location when building CLEANUP_POINT_EXPR.
3976 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3977 * tree.c (add_expr): Handle error_mark_node.
3979 2017-01-20 Martin Liska <mliska@suse.cz>
3982 * tree-profile.c (init_ic_make_global_vars): Do not call
3984 (gimple_init_gcov_profiler): Likewise.
3986 2017-01-20 Martin Liska <mliska@suse.cz>
3989 * cgraph.h (maybe_create_reference): Remove argument and
3991 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3993 * ipa-cp.c (create_specialized_node): Likewise.
3994 * symtab.c (symtab_node::maybe_create_reference): Handle
3995 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3997 2017-01-20 Martin Liska <mliska@suse.cz>
3999 * read-rtl-function.c (function_reader::create_function): Use
4000 build_decl instread of build_decl_stat.
4002 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
4004 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
4005 * config/i386/avx512dqintrin.h: Ditto.
4006 * config/i386/avx512fintrin.h: Ditto.
4007 * config/i386/i386-builtin-types.def: Add new types.
4008 * gcc/config/i386/i386.c: Handle new types.
4009 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
4010 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
4011 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
4012 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
4013 (__builtin_ia32_kshiftridi): New.
4014 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
4016 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
4020 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
4021 define to rs6000_init_stack_protect_guard.
4022 (rs6000_init_stack_protect_guard): New function.
4024 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4025 Yunqiang Su <yunqiang.su@imgtec.com>
4027 * config.gcc (supported_defaults): Add madd4.
4028 (with_madd4): Add validation.
4029 (all_defaults): Add madd4.
4030 * config/mips/mips.opt (mmadd4): New option.
4031 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4033 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4035 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
4036 (ISA_HAS_FUSED_MADD4): Likewise.
4037 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
4038 * gcc/doc/install.texi (--with-madd4): Document the new option.
4040 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4042 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
4043 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
4044 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
4045 (aarch64_init_pauth_hint_builtins): New.
4046 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
4047 (aarch64_expand_builtin): Expand new builtins.
4049 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4051 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
4052 * combine-stack-adj.c (no_unhandled_cfa): Handle
4053 REG_CFA_TOGGLE_RA_MANGLE.
4054 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
4055 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
4056 info for return address signing.
4057 (aarch64_expand_epilogue): Likewise.
4059 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4061 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
4062 * config/aarch64/aarch64-protos.h
4063 (aarch64_return_address_signing_enabled): New declaration.
4064 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
4066 (aarch64_expand_prologue): Sign return address before it's pushed onto
4068 (aarch64_expand_epilogue): Authenticate return address fetched from
4070 (aarch64_override_options): Sanity check for ILP32 and ISA level.
4071 (aarch64_attributes): New function attributes for "sign-return-address".
4072 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
4073 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
4074 ("*do_return"): Generate combined instructions according to key index.
4075 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
4076 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
4078 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
4079 * config/aarch64/aarch64.opt (msign-return-address=): New.
4080 * doc/extend.texi (AArch64 Function Attributes): Documents
4081 "sign-return-address=".
4082 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
4084 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4086 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
4087 overall option summary.
4089 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4091 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
4092 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
4093 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
4094 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
4096 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
4098 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
4099 -mpower9-minmax by default for -mcpu=power9.
4100 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
4101 128-bit floating point.
4103 2017-01-20 Alan Modra <amodra@gmail.com>
4105 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
4106 optimizing for size.
4108 2017-01-20 Alan Modra <amodra@gmail.com>
4111 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
4112 for strcmp and strncmp from corresponding builtin decl.
4114 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4116 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
4117 instead of i386/rtems-64.h.
4118 * config/i386/rtems-64.h: Remove.
4120 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4124 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
4126 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
4128 2017-01-19 Tamar Christina <tamar.christina@arm.com>
4130 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4131 Change int to HOST_WIDE_INT.
4132 * config/aarch64/aarch64-protos.h
4133 (aarch64_simd_gen_const_vector_dup): Likewise.
4134 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
4136 2017-01-19 David Malcolm <dmalcolm@redhat.com>
4138 * langhooks-def.h (lhd_type_for_size): New decl.
4139 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
4140 * langhooks.c (lhd_type_for_size): New function, taken from
4143 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
4145 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
4146 define_bypass for CR latency.
4147 (power9-cracked-alu): Update bypass latency and remove power9-branch.
4148 (power9-alu2): Add define_bypass for CR latency.
4150 (power9-mul): Update insn latency.
4151 (power9-mul-compare): Update insn latency, bypass latency and remove
4154 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4156 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
4158 * config/aarch64/aarch64.md
4159 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
4160 aarch64_nopcrelative_literal_loads.
4161 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
4163 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
4165 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4167 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4169 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4172 * config.gcc (supported_defaults): Add lxc1-sxc1.
4173 (with_lxc1_sxc1): Add validation.
4174 (all_defaults): Add lxc1-sxc1.
4175 * config/mips/mips.opt (mlxc1-sxc1): New option.
4176 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4178 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4179 __mips_no_lxc1_sxc1.
4180 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4181 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4182 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4184 2017-01-19 Richard Biener <rguenther@suse.de>
4186 PR tree-optimization/72488
4187 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4188 sure to restore SSA info.
4189 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4191 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4193 PR rtl-optimization/79121
4194 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4195 of the inner type when shifting an extended value.
4197 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4200 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4201 interposable aliases.
4203 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4206 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4207 Use the evmergelohi instruction.
4208 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4209 (mov_sitf_e500_subreg8_2_be): Likewise.
4210 (mov_sitf_e500_subreg12_2_le): Likewise.
4211 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4212 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4213 (mov_sitf_e500_subreg8_2_le): Likewise.
4214 (mov_sitf_e500_subreg12_2_be): Likewise.
4216 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4218 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4219 attribute from vecsimple to vecperm.
4220 (altivec_vbpermq2): Likewise.
4222 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4225 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4227 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4228 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4229 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4230 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4231 case where N arg is SIZE_MAX.
4232 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4233 (cmpstrsi): Add pattern.
4235 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4237 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4238 __builtin_vec_revb builtins.
4239 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4240 built-in functions to support generation of the ISA 3.0 XXBR<x>
4241 vector byte reverse instructions.
4242 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4243 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4244 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4245 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4246 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4247 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4248 (P9V_BUILTIN_VEC_REVB): Likewise.
4249 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4250 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4251 (p9_xxbrq_v16qi): Likewise.
4252 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4253 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4254 (p9_xxbrh_v8hi): Likewise.
4255 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4256 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4257 vec_revb built-in functions.
4259 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4261 PR rtl-optimization/78952
4262 * config/i386/i386.md (any_extract): New code iterator.
4263 (*insvqi_2): Use any_extract for source operand.
4264 (*insvqi_3): Use any_shiftrt for source operand.
4266 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4268 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4270 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4272 2017-01-18 Matthias Klose <doko@ubuntu.com>
4274 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4276 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4278 * config/rs6000/altivec.h (vec_bperm): Change #define.
4279 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4280 (altivec_vbpermq2): New define_insn.
4281 (altivec_vbpermd): Likewise.
4282 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4284 (VBPERMD): Likewise.
4285 (VBPERM): New polymorphic function interface.
4286 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4287 Add entries for P9V_BUILTIN_VEC_VBPERM.
4288 * doc/extend.texi: Add interfaces for vec_bperm.
4290 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4292 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4293 first letter of error messages.
4294 (s390_resolve_overloaded_builtin): Likewise.
4295 * config/s390/s390.c (s390_expand_builtin): Likewise.
4296 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4297 (s390_valid_target_attribute_inner_p): Likewise.
4298 * config/s390/s390.md ("tabort"): Likewise.
4300 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4302 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4303 (ISA_AVOID_DIV_HILO): New macro.
4304 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4305 (ISA_HAS_DDIV): Likewise.
4307 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4309 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4311 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4313 * doc/invoke.texi (fabi-version): Spelling fix.
4315 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4318 * doc/invoke.texi (fabi-version): Mention mangling fix for
4321 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4324 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4326 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4329 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4332 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4334 (stack_protect_test): Ditto.
4335 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4336 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4338 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4339 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4340 -mstack-protector-guard-offset=.
4341 (RS/6000 and PowerPC Options): Ditto.
4343 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4345 * config/i386/i386.h (MASK_CLASS_P): New define.
4346 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4347 there are no registers from different register sets also when
4348 mask registers are used. Update function comment.
4349 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4350 to (*k/*r) and (*k/*km) alternatives.
4352 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4354 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4355 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4356 (EH_RETURN_HANDLER_RTX): New define.
4357 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4358 Force frame pointer in EH return functions.
4359 (aarch64_expand_epilogue): Add barrier for eh_return.
4360 (aarch64_final_eh_return_addr): Remove.
4361 (aarch64_eh_return_handler_rtx): New function.
4362 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4364 (aarch64_eh_return_handler_rtx): New prototype.
4366 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4368 * config/rs6000/altivec.h (vec_rlmi): New #define.
4369 (vec_vrlnm): Likewise.
4370 (vec_rlnm): Likewise.
4371 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4372 (UNSPEC_VRLNM): Likewise.
4373 (VIlong): New mode iterator.
4374 (altivec_vrl<VI_char>mi): New define_insn.
4375 (altivec_vrl<VI_char>nm): Likewise.
4376 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4379 (RLNM): New polymorphic function entry.
4380 (VRLWMI): New monomorphic function entry.
4382 (RLMI): New polymorphic function entry.
4383 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4384 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4385 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4388 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4391 * dwarf2out.c (field_byte_offset): Restore the
4392 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4393 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4394 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4397 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4400 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4402 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4405 * dwarf2out.c (add_data_member_location_attribute): For constant
4406 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4407 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4408 DW_AT_byte_size attributes.
4410 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4412 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4413 after forcing to constant memory when the code model is medium.
4415 2017-01-17 Julia Koval <julia.koval@intel.com>
4418 * config/i386/avx512fintrin.h
4419 (_mm512_i32gather_ps): Change __addr type to void const*.
4420 (_mm512_mask_i32gather_ps): Ditto.
4421 (_mm512_i32gather_pd): Ditto.
4422 (_mm512_mask_i32gather_pd): Ditto.
4423 (_mm512_i64gather_ps): Ditto.
4424 (_mm512_mask_i64gather_ps): Ditto.
4425 (_mm512_i64gather_pd): Ditto.
4426 (_mm512_mask_i64gather_pd): Ditto.
4427 (_mm512_i32gather_epi32): Ditto.
4428 (_mm512_mask_i32gather_epi32): Ditto.
4429 (_mm512_i32gather_epi64): Ditto.
4430 (_mm512_mask_i32gather_epi64): Ditto.
4431 (_mm512_i64gather_epi32): Ditto.
4432 (_mm512_mask_i64gather_epi32): Ditto.
4433 (_mm512_i64gather_epi64): Ditto.
4434 (_mm512_mask_i64gather_epi64): Ditto.
4435 (_mm512_i32scatter_ps): Change __addr type to void*.
4436 (_mm512_mask_i32scatter_ps): Ditto.
4437 (_mm512_i32scatter_pd): Ditto.
4438 (_mm512_mask_i32scatter_pd): Ditto.
4439 (_mm512_i64scatter_ps): Ditto.
4440 (_mm512_mask_i64scatter_ps): Ditto.
4441 (_mm512_i64scatter_pd): Ditto.
4442 (_mm512_mask_i64scatter_pd): Ditto.
4443 (_mm512_i32scatter_epi32): Ditto.
4444 (_mm512_mask_i32scatter_epi32): Ditto.
4445 (_mm512_i32scatter_epi64): Ditto.
4446 (_mm512_mask_i32scatter_epi64): Ditto.
4447 (_mm512_i64scatter_epi32): Ditto.
4448 (_mm512_mask_i64scatter_epi32): Ditto.
4449 (_mm512_i64scatter_epi64): Ditto.
4450 (_mm512_mask_i64scatter_epi64): Ditto.
4451 * config/i386/avx512pfintrin.h
4452 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4453 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4454 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4455 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4456 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4457 (_mm512_prefetch_i32scatter_ps): Ditto.
4458 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4459 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4460 (_mm512_prefetch_i64scatter_pd): Ditto.
4461 (_mm512_prefetch_i64scatter_ps): Ditto.
4462 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4463 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4464 * config/i386/avx512vlintrin.h
4465 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4466 (_mm_mmask_i32gather_ps): Ditto.
4467 (_mm256_mmask_i32gather_pd): Ditto.
4468 (_mm_mmask_i32gather_pd): Ditto.
4469 (_mm256_mmask_i64gather_ps): Ditto.
4470 (_mm_mmask_i64gather_ps): Ditto.
4471 (_mm256_mmask_i64gather_pd): Ditto.
4472 (_mm_mmask_i64gather_pd): Ditto.
4473 (_mm256_mmask_i32gather_epi32): Ditto.
4474 (_mm_mmask_i32gather_epi32): Ditto.
4475 (_mm256_mmask_i32gather_epi64): Ditto.
4476 (_mm_mmask_i32gather_epi64): Ditto.
4477 (_mm256_mmask_i64gather_epi32): Ditto.
4478 (_mm_mmask_i64gather_epi32): Ditto.
4479 (_mm256_mmask_i64gather_epi64): Ditto.
4480 (_mm_mmask_i64gather_epi64): Ditto.
4481 (_mm256_i32scatter_ps): Change __addr type to void*.
4482 (_mm256_mask_i32scatter_ps): Ditto.
4483 (_mm_i32scatter_ps): Ditto.
4484 (_mm_mask_i32scatter_ps): Ditto.
4485 (_mm256_i32scatter_pd): Ditto.
4486 (_mm256_mask_i32scatter_pd): Ditto.
4487 (_mm_i32scatter_pd): Ditto.
4488 (_mm_mask_i32scatter_pd): Ditto.
4489 (_mm256_i64scatter_ps): Ditto.
4490 (_mm256_mask_i64scatter_ps): Ditto.
4491 (_mm_i64scatter_ps): Ditto.
4492 (_mm_mask_i64scatter_ps): Ditto.
4493 (_mm256_i64scatter_pd): Ditto.
4494 (_mm256_mask_i64scatter_pd): Ditto.
4495 (_mm_i64scatter_pd): Ditto.
4496 (_mm_mask_i64scatter_pd): Ditto.
4497 (_mm256_i32scatter_epi32): Ditto.
4498 (_mm256_mask_i32scatter_epi32): Ditto.
4499 (_mm_i32scatter_epi32): Ditto.
4500 (_mm_mask_i32scatter_epi32): Ditto.
4501 (_mm256_i32scatter_epi64): Ditto.
4502 (_mm256_mask_i32scatter_epi64): Ditto.
4503 (_mm_i32scatter_epi64): Ditto.
4504 (_mm_mask_i32scatter_epi64): Ditto.
4505 (_mm256_i64scatter_epi32): Ditto.
4506 (_mm256_mask_i64scatter_epi32): Ditto.
4507 (_mm_i64scatter_epi32): Ditto.
4508 (_mm_mask_i64scatter_epi32): Ditto.
4509 (_mm256_i64scatter_epi64): Ditto.
4510 (_mm256_mask_i64scatter_epi64): Ditto.
4511 (_mm_i64scatter_epi64): Ditto.
4512 (_mm_mask_i64scatter_epi64): Ditto.
4513 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4514 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4515 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4516 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4517 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4518 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4519 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4520 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4521 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4522 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4523 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4524 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4525 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4526 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4527 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4528 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4529 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4530 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4531 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4532 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4533 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4534 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4535 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4536 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4537 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4538 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4539 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4540 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4541 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4542 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4543 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4544 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4545 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4546 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4547 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4548 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4549 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4550 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4551 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4552 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4553 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4554 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4555 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4556 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4557 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4558 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4559 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4560 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4561 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4562 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4563 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4564 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4565 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4566 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4567 definitions accordingly.
4569 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4570 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4573 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4576 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4579 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4580 hard regs for inner regno.
4582 2017-01-17 Martin Liska <mliska@suse.cz>
4585 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4586 assumption and add comment.
4588 2017-01-17 Nathan Sidwell <nathan@acm.org>
4590 * ipa-visibility.c (localize_node): New function, broken out of ...
4591 (function_and_variable_visibility): ... here. Call it.
4593 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4596 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4597 correctly set frequency of oudgoing edge.
4598 (duplicate_thread_path): Fix profile updating.
4600 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4603 * configure.ac: Add GCC_BASE_VER.
4604 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4605 version from BASE-VER file.
4606 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4607 (gcc.o): Depend on $(BASEVER).
4608 * common.opt (dumpfullversion): New option.
4609 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4610 * doc/invoke.texi: Document -dumpfullversion.
4611 * doc/install.texi: Document --with-gcc-major-version-only.
4612 * configure: Regenerated.
4614 2017-01-17 Richard Biener <rguenther@suse.de>
4616 PR tree-optimization/71433
4617 * tree-vrp.c (register_new_assert_for): Merge same asserts
4618 on all incoming edges.
4619 (process_assert_insertions_for): Handle insertions at the
4622 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4624 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4625 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4627 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4630 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4633 2017-01-17 Alan Modra <amodra@gmail.com>
4636 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4637 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4638 symbolic stack limit when pic.
4640 2017-01-16 Martin Sebor <msebor@redhat.com>
4642 PR tree-optimization/78608
4643 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4645 2017-01-16 Jeff Law <law@redhat.com>
4648 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4649 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4650 for several include directories that may be relative to sysroot.
4651 * config/i386/x-mingw32 (gplus_includedir): Define.
4652 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4653 (native_system_includedir): Likewise.
4654 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4655 override if TARGET_SYSTEM_ROOT is defined.
4656 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4658 PR tree-optimization/79090
4659 PR tree-optimization/33562
4660 PR tree-optimization/61912
4661 PR tree-optimization/77485
4662 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4663 and computed trims into the dump file.
4665 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4667 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4669 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4672 * gimplify.c (gimplify_init_constructor): If want_value and
4673 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4677 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4678 sequence. Formatting fixes.
4679 (doloop_optimize): Formatting fixes.
4682 * gcc.c (debug_level_greater_than_spec_func): New function.
4683 (static_spec_functions): Add debug-level-gt spec function.
4684 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4686 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4687 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4688 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4689 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4690 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4691 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4693 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4695 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4696 QImode fixups to general and mask registers only.
4698 2017-01-16 Carl Love <cel@us.ibm.com>
4700 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4701 for built-in functions
4702 vector signed char vec_nabs (vector signed char)
4703 vector signed short vec_nabs (vector signed short)
4704 vector signed int vec_nabs (vector signed int)
4705 vector signed long long vec_nabs (vector signed long long)
4706 vector float vec_nabs (vector float)
4707 vector double vec_nabs (vector double)
4708 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4710 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4711 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4712 * doc/extend.texi: Update the documentation file for the new built-in
4715 2017-01-16 Martin Sebor <msebor@redhat.com>
4717 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4720 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4722 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4723 UNSPEC_VSX__XXSPLTD to require special splat handling.
4725 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4728 * system.h: Poison strndup.
4730 2017-01-16 Alan Modra <amodra@gmail.com>
4733 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4736 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4738 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4740 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4742 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4743 call recog here. Assert that INSN_CODE (insn) is non-negative.
4745 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4748 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4750 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4751 in the currently scheduled RTL fragment.
4753 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4755 PR rtl-optimization/78751
4756 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4759 2017-01-14 Jeff Law <law@redhat.com>
4761 PR tree-optimization/79090
4762 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4763 variable length stores.
4764 (compute_trims): Delete dead assignment to *trim_tail.
4765 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4768 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4770 PR rtl-optimization/78626
4771 PR rtl-optimization/78727
4772 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4773 of a block, and split such blocks after everything else is finished.
4775 2017-01-14 Alan Modra <amodra@gmail.com>
4778 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4779 target legitimate_combined_insn.
4780 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4781 (rs6000_legitimate_combined_insn): New function.
4782 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4784 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4785 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4786 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4788 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4790 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4792 2017-01-13 Jeff Law <law@redhat.com>
4794 PR tree-optimization/33562
4795 PR tree-optimization/61912
4796 PR tree-optimization/77485
4797 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4799 (delete_dead_assignment): Likewise.
4800 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4801 statement to delete_dead_call and delete_dead_assignment.
4803 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4806 * substring-locations.c (format_warning_va): Strengthen case 1 so
4807 that both endpoints of the substring must be within the format
4808 range for just the substring to be printed.
4810 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4812 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4813 * config/i386/i386.c (ix86_target_string): Add missing options
4814 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4815 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4816 flags_other and ix86_target_other to flags2_other. Display unknown
4818 (ix86_valid_target_attribute_inner_p): Add missing options and
4819 reorder options by implied ISAs, as in ix86_target_string.
4821 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4823 * hash-table.h (hash_table::too_empty_p): New function.
4824 (hash_table::expand): Use it.
4825 (hash_table::traverse): Likewise.
4826 (hash_table::empty_slot): Use sizeof (value_type) instead of
4827 sizeof (PTR) to convert bytes to elements. Shrink the table
4828 if the current size is excessive for the current number of
4831 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4833 * ira-costs.c (record_reg_classes): Break from the inner loop
4834 early once alt_fail is known to be true. Update outer loop
4835 handling accordingly.
4837 2017-01-13 Jeff Law <law@redhat.com>
4839 * tree-ssa-dse.c (decrement_count): New function.
4840 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4841 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4842 when we know the partially dead statement is a mem* function.
4844 PR tree-optimization/61912
4845 PR tree-optimization/77485
4846 * tree-ssa-dse.c: Include expr.h.
4847 (maybe_trim_constructor_store): New function.
4848 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4850 PR tree-optimization/33562
4851 PR tree-optimization/61912
4852 PR tree-optimization/77485
4853 * doc/invoke.texi: Document new dse-max-object-size param.
4854 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4855 * tree-ssa-dse.c: Include params.h.
4856 (dse_store_status): New enum.
4857 (initialize_ao_ref_for_dse): New, partially extracted from
4859 (valid_ao_ref_for_dse, normalize_ref): New.
4860 (setup_live_bytes_from_ref, compute_trims): Likewise.
4861 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4862 (maybe_trim_partially_dead_store): Likewise.
4863 (maybe_trim_complex_store): Likewise.
4864 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4865 Track what bytes live from the original store. Return tri-state
4866 for dead, partially dead or live.
4867 (dse_dom_walker): Add constructor, destructor and new private members.
4868 (delete_dead_call, delete_dead_assignment): New extracted from
4870 (dse_optimize_stmt): Make a member of dse_dom_walker.
4871 Use initialize_ao_ref_for_dse.
4873 PR tree-optimization/33562
4874 PR tree-optimization/61912
4875 PR tree-optimization/77485
4876 * sbitmap.h (bitmap_count_bits): Prototype.
4877 (bitmap_clear_range, bitmap_set_range): Likewise.
4878 * sbitmap.c (bitmap_clear_range): New function.
4879 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4881 2017-01-13 Martin Liska <mliska@suse.cz>
4884 * function.c (set_cfun): Add new argument force.
4885 * function.h (set_cfun): Likewise.
4886 * ipa-inline-transform.c (inline_call): Use the function when
4887 strict alising from is dropped for function we inline to.
4889 2017-01-13 Richard Biener <rguenther@suse.de>
4891 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4892 for dumping GIMPLE INTEGER_CSTs.
4894 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4896 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4897 to 201112L since C++17.
4899 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4902 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4903 if -fsanitize=kernel-address is present.
4905 2017-01-13 Richard Biener <rguenther@suse.de>
4907 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4908 as _Literal ( type ) number in case usual suffixes do not
4909 preserve all information.
4911 2017-01-13 Richard Biener <rguenther@suse.de>
4913 PR tree-optimization/77283
4914 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4915 and ssa-iterators.h.
4916 (is_feasible_trace): Implement a cost model based on joiner
4919 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4922 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4923 char or short to __float128/_Float128 directly.
4925 2017-01-12 Martin Sebor <msebor@redhat.com>
4927 to -Wformat-overflow.
4928 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4929 (min_bytes_remaining): Same.
4930 (get_string_length): Same.
4931 (format_string): Same.
4932 (format_directive): Same.
4934 (pass_sprintf_length::handle_gimple_call): Same.
4936 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4938 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4939 info.nowrite calls with no lhs that can't throw. Return bool
4940 whether gsi_remove has been called or not.
4941 (pass_sprintf_length::handle_gimple_call): Return bool whether
4942 try_substitute_return_value called gsi_remove. Formatting fix.
4943 (pass_sprintf_length::execute): Don't use gsi_remove if
4944 handle_gimple_call returned true.
4947 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4948 be removed due to side-effects, don't remove following barrier nor
4949 turn the successor edge into fallthru edge.
4951 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4954 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4955 element-reversing loads and stores as not swappable.
4957 2017-01-12 Nathan Sidwell <nathan@acm.org>
4958 Nicolai Stange <nicstange@gmail.com>
4960 * combine.c (try_combine): Don't ignore result of overlap checking
4961 loop. Combine overlap & asm check into single loop.
4963 2017-01-12 Richard Biener <rguenther@suse.de>
4965 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4966 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4968 2017-01-12 Richard Biener <rguenther@suse.de>
4970 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4971 and TS_TARGET_OPTION directly derive from TS_BASE.
4972 * tree-core.h (tree_optimization_option): Derive from tree_base.
4973 (tree_target_option): Likewise.
4975 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4977 * config/i386/i386.c (memory_address_length): Increase len
4978 only when rip_relative_addr_p returns false.
4980 2017-01-11 Julia Koval <julia.koval@intel.com>
4982 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4983 (OPTION_MASK_ISA_SGX_SET): New.
4984 (ix86_handle_option): Handle OPT_msgx.
4985 * config.gcc: Added sgxintrin.h.
4986 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4987 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4988 * config/i386/i386.c (ix86_target_string): Add -msgx.
4990 (ix86_option_override_internal): Handle new options.
4991 (ix86_valid_target_attribute_inner_p): Add sgx.
4992 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4993 * config/i386/i386.opt: Add msgx.
4994 * config/i386/sgxintrin.h: New file.
4995 * config/i386/x86intrin.h: Add sgxintrin.h.
4997 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5000 * fold-const.c (maybe_nonzero_address): Return 1 for function
5002 (tree_single_nonzero_warnv_p): Don't handle function local objects
5006 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
5009 2017-01-11 David Malcolm <dmalcolm@redhat.com>
5012 * opts.c: Include "spellcheck.h"
5013 (struct string_fragment): New struct.
5014 (struct edit_distance_traits<const string_fragment &>): New
5016 (get_closest_sanitizer_option): New function.
5017 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
5019 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5021 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
5023 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
5024 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
5025 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
5026 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
5027 for initial die_offset if dwarf_split_debug_info.
5028 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
5029 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
5031 (output_skeleton_debug_sections): Formatting fix. Use
5032 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
5033 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
5035 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
5037 * config/arm/cortex-a53.md: Add bypasses for
5039 (cortex_a53_r2f): Only use for transfers.
5040 (cortex_a53_f2r): Likewise.
5041 (cortex_a53_r2f_cvt): Add reservation for conversions.
5042 (cortex_a53_f2r_cvt): Likewise.
5044 2017-01-11 Tamar Christina <tamar.christina@arm.com>
5046 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
5047 to all inlined functions, change static to extern.
5049 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
5052 * config/arm/arm.c (legitimize_pic_address): Handle reference to
5054 (arm_assemble_integer): Likewise.
5056 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
5058 * config.gcc: Use new awk script to check CPU, FPU and architecture
5059 parameters for --with-... options.
5060 * config/arm/parsecpu.awk: New file
5061 * config/arm/arm-cpus.in: New file.
5062 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
5064 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
5066 * config/arm/t-arm: Update dependency rules.
5067 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
5068 of processing .def files.
5069 * config/arm/genopt.sh: Deleted.
5070 * config/arm/gentune.sh: Deleted.
5071 * config/arm/arm-cores.def: Deleted.
5072 * config/arm/arm-arches.def: Deleted.
5073 * config/arm/arm-fpus.def: Deleted.
5074 * config/arm/arm-tune.md: Regenerated.
5075 * config/arm/arm-tables.opt: Regenerated.
5076 * config/arm/arm-cpu.h: New generated file.
5077 * config/arm/arm-cpu-data.h: New generated file.
5078 * config/arm/arm-cpu-cdata.h: New generated file.
5080 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
5083 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
5085 (input_varpool_node): Unpack dynamically_initialized bit.
5087 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
5089 PR rtl-optimization/79032
5090 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
5091 the alignment of the adjusted memory reference against that of MODE,
5092 instead of the alignment of the original memory reference.
5094 2017-01-11 Martin Jambor <mjambor@suse.cz>
5096 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
5098 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
5099 decorated functions.
5101 2017-01-11 Richard Biener <rguenther@suse.de>
5103 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
5104 set range/nonnull info for PHI results. Do not set it on
5105 stmts marked for removal.
5107 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
5109 * expr.c (store_field): In the bitfield case, fetch the return value
5110 from the registers before applying a single big-endian adjustment.
5111 Always do a final load for a BLKmode value not larger than a word.
5113 2017-01-10 David Malcolm <dmalcolm@redhat.com>
5116 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5117 that we correctly handle column numbers greater than
5118 LINE_MAP_MAX_COLUMN_NUMBER.
5120 2017-01-10 Martin Sebor <msebor@redhat.com>
5123 * gimple-ssa-sprintf.c (get_destination_size): Call
5124 {init,fini}object_sizes.
5125 * tree-object-size.c (addr_object_size): Adjust.
5126 (pass_through_call): Adjust.
5127 (pass_object_sizes::execute): Adjust.
5128 * tree-object-size.h (fini_object_sizes): Declare.
5130 2017-01-10 Martin Sebor <msebor@redhat.com>
5132 PR tree-optimization/78775
5133 * builtins.c (get_size_range): Move...
5134 * calls.c: ...to here.
5135 (alloc_max_size): Accept zero argument.
5136 (operand_signed_p): Remove.
5137 (maybe_warn_alloc_args_overflow): Call get_size_range.
5138 * calls.h (get_size_range): Declare.
5140 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
5142 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
5143 from TI's devices.csv file as of September 2016.
5144 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
5146 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
5148 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
5149 * doc/invoke.texi: Likewise.
5150 * doc/md.texi: Likewise.
5151 * doc/objc.texi: Likewise.
5153 2017-01-10 Joshua Conner <joshconner@google.com>
5155 * config/arm/fuchsia-elf.h: New file.
5156 * config/fuchsia.h: New file.
5157 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
5158 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
5160 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
5162 2016-01-10 Richard Biener <rguenther@suse.de>
5164 PR tree-optimization/79034
5165 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
5166 Propagate out degenerate PHIs in the joiner.
5168 2017-01-10 Martin Liska <mliska@suse.cz>
5170 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5171 (sort_congruence_classes_by_decl_uid): Likewise.
5172 (sort_congruence_class_groups_by_decl_uid): Likewise.
5173 (sem_item_optimizer::merge_classes): Sort class, groups in these
5174 classes and members in the groups by DECL_UID of declarations.
5175 This would make merge operations stable.
5177 2017-01-10 Martin Liska <mliska@suse.cz>
5179 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5180 usage of m_classes_vec.
5181 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5182 (sem_item_optimizer::get_group_by_hash): Likewise.
5183 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5184 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5185 (sem_item_optimizer::verify_classes): Likewise.
5186 (sem_item_optimizer::process_cong_reduction): Likewise.
5187 (sem_item_optimizer::dump_cong_classes): Likewise.
5188 (sem_item_optimizer::merge_classes): Likewise.
5189 * ipa-icf.h (congruence_class_hash): Rename from
5190 congruence_class_group_hash. Remove declaration of m_classes_vec.
5192 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5194 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5195 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5196 * config.gcc: Add avx512vpopcntdqintrin.h.
5197 * config/i386/avx512vpopcntdqintrin.h: New.
5198 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5199 * config/i386/i386-builtin-types.def: Add new types.
5200 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5201 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5202 __builtin_ia32_vpopcountq_v8di_mask): New.
5203 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5204 __AVX512VPOPCNTDQ__.
5205 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5206 (PTA_AVX512VPOPCNTDQ): Define.
5207 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5208 TARGET_AVX512VPOPCNTDQ_P): Define.
5209 * config/i386/i386.opt: Add mavx512vpopcntdq.
5210 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5211 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5213 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5216 * predict.def (PRED_CALL): Set to 67.
5218 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5220 * expr.c (store_field): In the bitfield case, if the value comes from
5221 a function call and is of an aggregate type returned in registers, do
5222 not modify the field mode; extract the value in all cases if the mode
5223 is BLKmode and the size is not larger than a word.
5225 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5228 * config/i386/cpuid.h: Fix undefined behavior.
5230 2017-01-04 Jeff Law <law@redhat.com>
5232 PR tree-optimization/79007
5233 PR tree-optimization/67955
5234 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5235 conservative for pt.null when flag_non_call_exceptions is on.
5237 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5239 PR translation/79019
5240 PR translation/79020
5241 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5242 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5243 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5245 * config/avr/avr.opt (maccumulate-args): Likewise.
5246 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5247 * common.opt (freport-bug): Likewise.
5248 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5249 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5250 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5251 translatable string.
5252 * config/i386/i386.c (function_value_32): Likewise.
5253 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5254 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5256 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5257 * common/config/msp430/msp430-common.c (msp430_handle_option):
5259 * symtab.c (symtab_node::verify_base): Likewise.
5260 * opts.c (set_debug_level): Likewise.
5261 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5263 missing whitespace to translatable strings.
5264 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5265 * config/sh/superh.h: Likewise.
5266 * config/i386/xopintrin.h: Likewise.
5267 * config/i386/znver1.md: Likewise.
5268 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5269 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5270 * double-int.h (struct double_int): Likewise.
5271 * double-int.c (div_and_round_double): Likewise.
5272 * wide-int.cc: Likewise.
5273 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5274 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5275 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5276 (crossjumps_occurred): ... this.
5277 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5280 PR tree-optimization/78899
5281 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5282 returning bool return struct loop *, NULL for failure and the new
5284 (versionable_outer_loop_p): Don't version outer loop if it has
5285 dont_vectorized bit set.
5286 (tree_if_conversion): When versioning outer loop, ensure
5287 tree_if_conversion is performed also on the inner loop of the
5288 non-vectorizable outer loop copy.
5289 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5290 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5291 prevent vectorization of it.
5292 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5293 the outer loop vectorization of the non-scalar version is attempted
5294 before vectorization of the inner loop in scalar version. If
5295 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5296 vectorization of its inner loop.
5297 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5298 has 2 inner loops, rename also on edges from bb whose single pred
5299 is outer_loop->header. Fix typo in function comment.
5301 2017-01-09 Martin Sebor <msebor@redhat.com>
5304 * asan.c (asan_emit_stack_protection): Increase local buffer size
5305 to avoid snprintf truncation warning.
5307 2017-01-09 Andrew Pinski <apinski@cavium.com>
5309 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5310 to reference thunderx2t99 for the tuning structure
5311 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5313 (thunderx2t99_extra_costs): This.
5314 * config/aarch64/aarch64-tune.md: Regenerate.
5315 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5316 (vulcan_addrcost_table): This.
5317 (vulcan_regmove_cost): Rename to ...
5318 (thunderx2t99_regmove_cost): This.
5319 (vulcan_vector_cost): Rename to ...
5320 (thunderx2t99_vector_cost): this.
5321 (vulcan_branch_cost): Rename to ...
5322 (thunderx2t99_branch_cost): This.
5323 (vulcan_tunings): Rename to ...
5324 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5325 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5327 2017-01-09 Martin Jambor <mjambor@suse.cz>
5331 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5332 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5333 (propagate_vr_accross_jump_function): Use the above function for all
5334 value range computations for pass-through jump functions and type
5335 converasion from explicit value range values.
5336 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5337 parameters from TYPE_ARG_TYPES.
5338 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5339 (ipa_write_node_info): Stream type of the actual argument.
5340 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5342 2017-01-09 Martin Liska <mliska@suse.cz>
5345 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5346 (lookup_compiler): Do not show error message with have_E.
5348 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5350 PR tree-optimization/78938
5351 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5352 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5353 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5354 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5357 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5359 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5362 2017-01-09 Richard Biener <rguenther@suse.de>
5364 PR tree-optimization/78997
5365 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5366 name condition properly.
5368 2017-01-09 Richard Biener <rguenther@suse.de>
5371 * dwarf2out.c (is_cxx): New overload with context.
5372 (is_naming_typedef_decl): Use it.
5374 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5376 * invoke.texi (Option Summary): Correct spacing in option lists
5377 and add line breaks to fix over-long lines.
5379 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5383 * extend.texi (Common Variable Attributes): Add xref to GCC
5384 Internals manual to explain mode attribute keywords.
5386 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5389 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5390 and Preprocessor Options.
5391 (Options for Linking): Document -pthread here....
5392 (RS/6000 and PowerPC Options): ...not here.
5393 (Solaris 2 Options): ...or here.
5394 * doc/cppopts.texi: Document -pthread.
5396 2017-01-08 Martin Sebor <msebor@redhat.com>
5399 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5400 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5401 New member functions.
5402 (format_directive): Used them.
5404 (pass_sprintf_length::handle_gimple_call): Same.
5405 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5406 to avoid truncation for any argument.
5407 (extract_affine_mul): Same.
5408 * tree.c (get_file_function_name): Same.
5410 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5413 * predict.def (PRED_INDIR_CALL): Set to 86.
5415 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5417 PR preprocessor/54124
5418 * doc/cppopts.texi: Reformat -d subtable to list the full name
5419 of the options. Add cross-reference to the docs for the general
5420 compiler -d options.
5421 * doc/invoke.texi (Developer Options): Add cross-reference to the
5422 preprocessor-specific -d option documentation.
5424 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5426 PR preprocessor/13498
5427 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5428 redudant material, and reflect new command-line options.
5429 (System Headers): Likewise.
5431 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5433 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5434 -isystem, and -idirafter. Copy-edit.
5435 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5436 default for -ftrack-macro-expansion. Delete obsolete and
5437 badly-formatted implementation details about -fdebug-cpp output.
5438 * doc/cppwarnopts.texi: Copy-edit.
5440 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5443 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5444 that the transition from a max line width >= 1<<10 to narrower
5445 lines works correctly.
5447 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5449 * doc/options.texi (PerFunction): New.
5450 * opt-functions.awk (switch_flags): Map both Optimization and
5451 PerFunction to CL_OPTIMIZATION.
5452 * opth-gen.awk: Test for PerFunction flag along with
5454 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5455 it only when the latter is present. Skip those that don't in
5456 the hash function generator.
5457 * common.opt (fvar-tracking): Mark as PerFunction instead of
5459 (fvar-tracking-assignments): Likewise.
5460 (fvar-tracking-assignments-toggle): Likewise.
5461 (fvar-tracking-uninit): Likewise.
5463 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5465 PR translation/79018
5466 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5469 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5472 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5473 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5474 TARGET_LONG_JUMP_TABLE_OFFSETS.
5475 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5476 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5477 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5478 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5479 * config/m68k/m68k.md (tablejump expander): Likewise.
5480 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5481 TARGET_LONG_JUMP_TABLE_OFFSETS.
5482 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5483 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5485 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5486 David Holsgrove <david.holsgrove@xilinx.com>
5488 * common/config/microblaze/microblaze-common.c
5489 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5490 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5492 * config/microblaze/microblaze.c (microblaze_must_save_register)
5493 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5495 (microblaze_eh_return): New function.
5496 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5497 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5498 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5499 * config/microblaze/microblaze.md (eh_return): New pattern.
5501 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5503 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5504 GCC_DIAGNOSTIC_STRINGIFY): Define.
5506 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5508 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5510 * config/arm/arm.md (<mcrr>): New.
5512 * config/arm/arm.c (arm_arch5te): New.
5513 (arm_option_override): Set arm_arch5te.
5514 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5516 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5517 (arm_mcrr_qualifiers): ... this. New.
5518 (MRRC_QUALIFIERS): Define to...
5519 (arm_mrrc_qualifiers): ... this. New.
5520 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5522 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5523 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5524 (MRRCI, mrrc, MRRC): New.
5525 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5526 VUNSPEC_MRRC2): New.
5528 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5530 * config/arm/arm.md (<mcr>): New.
5532 * config/arm/arm.c (arm_coproc_builtin_available): Add
5533 support for mcr, mrc, mcr2 and mrc2.
5534 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5535 (arm_mcr_qualifiers): ... this. New.
5536 (MRC_QUALIFIERS): Define to ...
5537 (arm_mrc_qualifiers): ... this. New.
5538 (MCR_QUALIFIERS): Define to ...
5539 (arm_mcr_qualifiers): ... this. New.
5540 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5542 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5543 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5544 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5547 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5549 * config/arm/arm.md (*ldc): New.
5553 * config/arm/arm.c (arm_coproc_builtin_available): Add
5554 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5555 (arm_coproc_ldc_stc_legitimate_address): New.
5556 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5557 'qualifier_const_pointer'.
5558 (LDC_QUALIFIERS): Define to...
5559 (arm_ldc_qualifiers): ... this. New.
5560 (STC_QUALIFIERS): Define to...
5561 (arm_stc_qualifiers): ... this. New.
5562 * config/arm/arm-protos.h
5563 (arm_coproc_ldc_stc_legitimate_address): New.
5564 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5565 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5566 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5567 stc2, stcl, stc2l): New.
5568 * config/arm/constraints.md (Uz): New.
5569 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5570 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5571 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5572 VUNSPEC_STC2L): New.
5574 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5576 * config/arm/arm.md (<cdp>): New.
5577 * config/arm/arm.c (neon_const_bounds): Rename this ...
5578 (arm_const_bounds): ... this.
5579 (arm_coproc_builtin_available): New.
5580 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5581 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5582 (CDP_QUALIFIERS): Define to...
5583 (arm_cdp_qualifiers): ... this. New.
5585 (arm_expand_builtin_args): Add case for 6 arguments.
5586 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5587 (arm_const_bounds): ... this.
5588 (arm_coproc_builtin_available): New.
5589 * config/arm/arm_acle.h (__arm_cdp): New.
5591 * config/arm/arm_acle_builtins.def (cdp): New.
5593 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5594 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5596 * config/arm/types.md (coproc): New.
5597 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5598 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5599 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5600 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5602 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5604 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5605 (UBINOP_QUALIFIERS): New.
5607 (acle_builtin_data): New. Change comment.
5608 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5609 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5610 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5611 arm_acle_builtins.def.
5612 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5613 (arm_init_acle_builtins): New.
5614 (CRC32_BUILTIN): Remove.
5615 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5616 crc32cb, crc32ch and crc32cw.
5617 (arm_init_crc32_builtins): Remove.
5618 (arm_init_builtins): Use arm_init_acle_builtins rather
5619 than arm_init_crc32_builtins.
5620 (arm_expand_acle_builtin): New.
5621 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5622 * config/arm/arm_acle_builtins.def: New.
5624 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5626 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5627 (arm_builtin_datum): ... this.
5628 (arm_init_neon_builtin): Rename to ...
5629 (arm_init_builtin): ... this. Add a new parameters PREFIX
5630 and USE_SIG_IN_NAME.
5631 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5632 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5633 'arm_builtin_datum'.
5634 (arm_init_vfp_builtins): Likewise.
5635 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5636 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5637 (arm_expand_neon_args): Rename to ...
5638 (arm_expand_builtin_args): ... this. Rename builtin_arg
5639 enum values and differentiate between ARG_BUILTIN_MEMORY
5640 and ARG_BUILTIN_NEON_MEMORY.
5641 (arm_expand_neon_builtin_1): Rename to ...
5642 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5643 values, arm_expand_builtin_args and add bool parameter NEON.
5644 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5645 (arm_expand_vfp_builtin): Likewise.
5646 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5648 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5651 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5652 * predict.c (tree_estimate_probability_bb): Reverse direction of
5653 polymorphic call predictor.
5655 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5657 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5658 (determine_pass_name_match): ...this new function and...
5659 (should_skip_pass_p): ...this new function.
5661 2017-01-06 Nathan Sidwell <nathan@acm.org>
5663 * ipa-visibility.c (function_and_variable_visibility): Reformat
5664 comments and long lines. Remove extrneous if.
5665 * symtab.c (symtab_node::make_decl_local): Fix code format.
5666 (symtab_node::set_section_for_node): Fix comment typo.
5668 2017-01-06 Martin Liska <mliska@suse.cz>
5671 * lra-constraints.c: Rename invariant to lra_invariant.
5672 * predict.c (set_even_probabilities): Initialize e to NULL.
5674 2017-01-05 Martin Sebor <msebor@redhat.com>
5676 PR tree-optimization/78910
5677 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5678 (format_integer): Correct off-by-one error in the handling
5679 of precision with negative numbers in signed conversions..
5681 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5683 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5685 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5687 PR tree-optimization/71016
5688 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5689 factor_out_conditional_conversion. Formatting fix.
5690 (factor_out_conditional_conversion): Add cond_stmt argument.
5691 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5692 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5695 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5697 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5698 read-rtl-function.o, and selftest-rtl.o.
5699 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5700 (selftest::aarch64_test_loading_full_dump): New function.
5701 (selftest::aarch64_run_selftests): New function.
5702 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5703 selftest::aarch64_run_selftests.
5704 * config/i386/i386.c
5705 (selftest::ix86_test_loading_dump_fragment_1): New function.
5706 (selftest::ix86_test_loading_call_insn): New function.
5707 (selftest::ix86_test_loading_full_dump): New function.
5708 (selftest::ix86_test_loading_unspec): New function.
5709 (selftest::ix86_run_selftests): Call the new functions.
5710 * emit-rtl.c (maybe_set_max_label_num): New function.
5711 * emit-rtl.h (maybe_set_max_label_num): New decl.
5712 * function.c (instantiate_decls): Guard call to
5713 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5714 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5716 * gensupport.c (gen_reader::gen_reader): Pass "false"
5717 for new "compact" param of rtx_reader.
5718 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5719 rather than an empty string for NULL strings.
5720 * read-md.c: Potentially include config.h rather than bconfig.h.
5721 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5722 (have_error): New global, copied from errors.c.
5723 (md_reader::read_name): Rename to...
5724 (md_reader::read_name_1): ...this, adding "out_loc" param,
5725 and converting "missing name or number" to returning false, rather
5727 (md_reader::read_name): Reimplement in terms of read_name_1.
5728 (md_reader::read_name_or_nil): New function.
5729 (md_reader::read_string): Handle "(nil)" by returning NULL.
5730 (md_reader::md_reader): Add new param "compact".
5731 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5732 (md_reader::read_file): New method.
5733 * read-md.h (md_reader::md_reader): Add new param "compact".
5734 (md_reader::read_file): New method.
5735 (md_reader::is_compact): New accessor.
5736 (md_reader::read_name): Convert return type from void to file_location.
5737 (md_reader::read_name_or_nil): New decl.
5738 (md_reader::read_name_1): New decl.
5739 (md_reader::m_compact): New field.
5740 (noop_reader::noop_reader): Pass "false" for new "compact" param
5742 (rtx_reader::rtx_reader): Add new "compact" param.
5743 (rtx_reader::read_rtx_operand): Make virtual and convert return
5744 type from void to rtx.
5745 (rtx_reader::read_until): New decl.
5746 (rtx_reader::handle_any_trailing_information): New virtual function.
5747 (rtx_reader::postprocess): New virtual function.
5748 (rtx_reader::finalize_string): New virtual function.
5749 (rtx_reader::m_in_call_function_usage): New field.
5750 (rtx_reader::m_reuse_rtx_by_id): New field.
5751 * read-rtl-function.c: New file.
5752 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5753 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5754 (selftest::verify_three_block_rtl_cfg): New decl.
5755 * read-rtl-function.h: New file.
5756 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5757 For host, include function.h, memmodel.h, and emit-rtl.h.
5758 (one_time_initialization): New function.
5759 (struct compact_insn_name): New struct.
5760 (compact_insn_names): New array.
5761 (find_code): Handle insn codes in compact dumps.
5762 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5763 (bind_subst_iter_and_attr): Likewise.
5764 (add_condition_to_string): Likewise.
5765 (add_condition_to_rtx): Likewise.
5766 (apply_attribute_uses): Likewise.
5767 (add_current_iterators): Likewise.
5768 (apply_iterators): Likewise.
5769 (initialize_iterators): Guard usage of apply_subst_iterator with
5770 #ifdef GENERATOR_FILE.
5771 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5772 (md_reader::read_mapping): Likewise.
5773 (add_define_attr_for_define_subst): Likewise.
5774 (add_define_subst_attr): Likewise.
5775 (read_subst_mapping): Likewise.
5776 (check_code_iterator): Likewise.
5777 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5779 (one_time_initialization): New function.
5780 (rtx_reader::read_until): New method.
5781 (read_flags): New function.
5782 (parse_reg_note_name): New function.
5783 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5784 Handle reuse_rtx ids.
5785 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5786 Add parsing support for RTL dumps, mirroring the special-cases in
5787 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5788 values, and calling handle_any_trailing_information.
5789 (rtx_reader::read_rtx_operand): Convert return type from void
5790 to rtx, returning return_rtx. Handle case 'e'. Call
5791 finalize_string on XSTR and XTMPL fields.
5792 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5793 "(nil)" values were omitted. Call the postprocess vfunc on the
5795 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5796 class ctor. Initialize m_in_call_function_usage. Call
5797 one_time_initialization.
5798 * rtl-tests.c (selftest::test_uncond_jump): Call
5799 set_new_first_and_last_insn.
5800 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5801 * selftest-rtl.c: New file.
5802 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5803 (selftest::get_insn_by_uid): New decl.
5804 * selftest-run-tests.c (selftest::run_tests): Call
5805 read_rtl_function_c_tests.
5806 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5807 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5810 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5812 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5813 operands in a special way. Assert that pos+len <= mode precision.
5815 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5817 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5818 3 argument Alias with unlimited for the negative form.
5819 (fno-vect-cost-model): Removed.
5821 2017-01-05 Martin Liska <mliska@suse.cz>
5823 * hsa-gen.c (gen_hsa_divmod): New function.
5824 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5826 2017-01-05 Martin Liska <mliska@suse.cz>
5829 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5832 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5834 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5835 small constant length operands.
5837 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5839 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5840 between loop iterations.
5842 2017-01-05 Martin Liska <mliska@suse.cz>
5845 * gimplify.c (gimplify_decl_expr): Compare to
5846 asan_poisoned_variables instread of checking flags.
5847 (gimplify_target_expr): Likewise.
5848 (gimplify_expr): Likewise.
5849 (gimplify_function_tree): Conditionally initialize
5850 asan_poisoned_variables.
5852 2017-01-04 Jeff Law <law@redhat.com>
5854 PR tree-optimizatin/78812
5855 * rtl.h (contains_mem_rtx_p): Prototype.
5856 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5857 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5858 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5859 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5860 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5862 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5864 * input.c (assert_char_at_range): Default-initialize actual_range.
5866 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5868 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5869 to match the caller.
5871 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5873 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5874 insns after final jump in test to emit dummy move.
5876 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5878 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5879 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5881 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5883 * multiple_target.c (create_dispatcher_calls): Init e_next.
5884 * tree-ssa-loop-split.c (split_loop): Init border.
5885 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5888 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5893 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5894 (altivec_register_operand): Do not return true if the operand
5895 contains a SUBREG mixing SImode and SFmode.
5896 (vsx_register_operand): Likewise.
5897 (vsx_reg_sfsubreg_ok): New predicate.
5898 (vfloat_operand): Do not return true if the operand contains a
5899 SUBREG mixing SImode and SFmode.
5900 (vint_operand): Likewise.
5901 (vlogical_operand): Likewise.
5902 (gpc_reg_operand): Likewise.
5903 (int_reg_operand): Likewise.
5904 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5905 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5906 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5908 (rs6000_emit_move_si_sf_subreg): New helper function.
5909 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5910 fixup SUBREGs involving SImode and SFmode.
5911 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5912 numbers for the new peephole2 optimization.
5913 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5914 the GLIBC math library that do AND/IOR/XOR operations on single
5915 precision floating point.
5916 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5917 target macros to say whether we need to avoid SUBREGs mixing
5919 (TARGET_ALLOW_SF_SUBREG): Likewise.
5920 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5921 (UNSPEC_SI_FROM_SF): Likewise.
5922 (iorxor): Change spacing.
5923 (and_ior_xor): New iterator for AND, IOR, and XOR.
5924 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5925 (movdi_from_sf_zero_ext): Likewise.
5926 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5927 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5928 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5929 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5930 (fms<mode>4): Likewise.
5931 (fnma<mode>4): Likewise.
5932 (fnms<mode>4): Likewise.
5933 (nfma<mode>4): Likewise.
5934 (nfms<mode>4): Likewise.
5936 2017-01-04 Marek Polacek <polacek@redhat.com>
5939 * doc/invoke.texi: Document -Wpointer-compare.
5941 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5943 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5946 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5947 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5950 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5953 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5954 documentation of the powerpc_popcntb_ok attribute.
5955 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5956 code to issue warning messages if a requested CPU configuration is
5957 not supported by the binary (assembler and loader) toolchain.
5958 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5959 made to define a built-in function that has been disabled.
5960 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5961 made to define a built-in function that has been disabled.
5962 (altivec_init_builtins): Add comment explaining why definition
5963 of the DST built-in functions is not preceded by an assertion
5964 check. Add assertions to prevent ICE if attempts are made to
5965 define an altivec predicate or an abs* built-in function that has
5967 (htm_init_builtins): Add comment explaining why definition of the
5968 htm built-in functions is not preceded by an assertion check.
5970 2017-01-04 Jeff Law <law@redhat.com>
5972 PR tree-optimizatin/67955
5973 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5974 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5975 the points-to solution does not include pt_null. Use DECL_PT_UID
5978 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5980 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5981 Use gen_int_mode instead of gen_lopwart for const_int operands.
5983 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5985 PR tree-optimization/71563
5986 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5987 out of range value - has low bits known to be zero.
5989 2017-01-04 Alan Modra <amodra@gmail.com>
5991 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5992 * configure: Regenerate.
5993 * config.in: Regenerate.
5995 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5998 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5999 a substring of the message, but strcmp with the whole message. Ifdef
6000 ENABLE_NLS, translate the message first using dgettext.
6002 2017-01-03 Jeff Law <law@redhat.com>
6004 PR tree-optimizatin/78856
6005 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
6006 (mark_threaded_blocks): Remove code to truncate thread paths that
6007 cross multiple loop headers. Instead invalidate the cached loop
6008 iteration information and handle case of a thread path walking
6009 into an irreducible region.
6011 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6014 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
6015 assertions. Add support for doing the signbit if the IEEE 128-bit
6016 floating point value is in a GPR.
6017 * config/rs6000/rs6000.md (Fsignbit): Delete.
6018 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
6019 Update the length attribute if the value is in a GPR.
6020 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
6021 the sign or zero extension instruction, since the value is always 0/1.
6022 (signbit<mode>2_dm2): Delete using <Fsignbit>.
6025 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
6026 extracting SImode to a GPR register so that we can generate a
6027 store, limit the vector to be in a traditional Altivec register
6028 for the vextuwrx instruction.
6030 2017-01-03 Ian Lance Taylor <iant@google.com>
6032 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
6034 2017-01-03 Martin Sebor <msebor@redhat.com>
6036 PR tree-optimization/78696
6037 * gimple-ssa-sprintf.c (format_floating): Correct handling of
6038 precision. Use MPFR for %f for greater fidelity. Correct handling
6040 (pass_sprintf_length::compute_format_length): Set width and precision
6041 specified by asrerisk to void_node for vararg functions.
6042 (try_substitute_return_value): Adjust dump output.
6044 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
6046 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
6048 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
6050 * doc/invoke.texi (SPARC options): Document -mlra as the default.
6051 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
6052 -mlra/-mno-lra was passed to the compiler.
6054 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
6056 PR rtl-optimization/65618
6057 * emit-rtl.c (try_split): Move initialization of "before" and
6058 "after" to just before the call to emit_insn_after_setloc.
6060 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
6062 * doc/md.texi (Standard Names): Remove reference to Java frontend.
6064 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
6066 * dwarf2out.c (gen_enumeration_type_die): When
6067 -gno-strict-dwarf, add a DW_AT_encoding attribute.
6069 2017-01-03 Jakub Jelinek <jakub@redhat.com>
6071 PR tree-optimization/78965
6072 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
6073 Change first argument from const call_info & to call_info &. For %n
6074 set info.nowrite to false.
6077 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
6078 possibly throwing calls.
6080 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
6081 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
6082 and fns handling, rather than in a separate case SSA_NAME.
6084 2017-01-02 Jeff Law <law@redhat.com>
6086 * config/darwin-driver.c (darwin_driver_init): Const-correctness
6087 fixes for first_period and second_period variables.
6089 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
6092 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
6093 (*insvqi_1): New insn pattern.
6094 (*insvqi_1_mem_rex64): Ditto.
6096 (*insvqi_3): Rename from *insvqi.
6098 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
6100 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
6102 * doc/cfg.texi (Edges): Remove reference to Java.
6103 (Maintaining the CFG): Ditto.
6105 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6108 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
6109 transparent aliases.
6111 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6114 * predict.def (PRED_CALL): Update hitrate.
6115 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
6116 * predict.c (tree_estimate_probability_bb): Split CALL predictor
6117 into direct/indirect/polymorphic variants.
6119 2017-01-01 Jakub Jelinek <jakub@redhat.com>
6121 Update copyright years.
6123 * gcc.c (process_command): Update copyright notice dates.
6124 * gcov-dump.c (print_version): Ditto.
6125 * gcov.c (print_version): Ditto.
6126 * gcov-tool.c (print_version): Ditto.
6127 * gengtype.c (create_file): Ditto.
6128 * doc/cpp.texi: Bump @copying's copyright year.
6129 * doc/cppinternals.texi: Ditto.
6130 * doc/gcc.texi: Ditto.
6131 * doc/gccint.texi: Ditto.
6132 * doc/gcov.texi: Ditto.
6133 * doc/install.texi: Ditto.
6134 * doc/invoke.texi: Ditto.
6136 Copyright (C) 2017 Free Software Foundation, Inc.
6138 Copying and distribution of this file, with or without modification,
6139 are permitted in any medium without royalty provided the copyright
6140 notice and this notice are preserved.