1 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
4 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
6 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
8 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
11 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
14 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21 * final.c (last_columnnum, override_columnnum): New variables.
22 (final_start_function): Set last_columnnum, pass it to begin_prologue
23 hook and pass 0 to dwarf2out_begin_prologue.
24 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25 to source_line debug hook.
26 (notice_source_line): Compute last_columnnum and for debug_column_info
27 return true on column changes.
28 * debug.h (struct gcc_debug_hooks): Add column argument to
29 source_line and begin_prologue hooks.
30 (debug_nothing_int_charstar_int_bool): Remove prototype.
31 (debug_nothing_int_int_charstar,
32 debug_nothing_int_int_charstar_int_bool): New prototypes.
33 (dwarf2out_begin_prologue): Add column argument.
34 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36 (debug_nothing_int_charstar_int_bool): Remove.
37 (debug_nothing_int_int_charstar,
38 debug_nothing_int_int_charstar_int_bool): New functions.
39 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
40 through to dwarf2out_source_line.
41 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
42 (dwarf2out_source_line): Add column argument, emit it if requested.
43 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
45 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
46 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
47 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
48 through to dwarf2out_begin_prologue.
49 (vmsdbgout_source_line): Add column argument, pass it through to
50 dwarf2out_source_line.
51 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
52 dbxout_source_line caller.
53 (dbxout_source_line): Add column argument.
55 * common.opt (gno-column-info, gcolumn-info): New options.
56 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
57 (check_die): Also test for multiple DW_AT_decl_column attributes.
58 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
59 DW_AT_decl_column if requested.
60 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
62 (gen_variable_die): Likewise.
63 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
64 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
67 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
68 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
69 (ix86_handle_option): Handle OPT_m3dnowa.
70 * doc/invoke.texi (-m3dnowa): Document.
71 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
72 -m3dnowa instead of -m3dnow -march=athlon.
75 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
76 instead of gcc_assert for K, r and R code checks. Formatting fixes.
78 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
81 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
82 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
83 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
84 generator for vsx_xxpermdi_<mode>_be.
85 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
86 force big-endian semantics.
87 (vsx_xxpermdi_<mode>_be): New define_expand with same
88 implementation as previous version of vsx_xxpermdi_<mode>.
90 2017-02-17 Jakub Jelinek <jakub@redhat.com>
92 PR tree-optimization/79327
93 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
94 variable, its initialization and use.
96 2017-02-17 Julia Koval <julia.koval@intel.com>
98 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
99 (OPTION_MASK_ISA_PKU_UNSET): New.
100 (ix86_handle_option): Handle -mrdpid.
101 * config/i386/cpuid.h (bit_RDPID): New.
102 * config/i386/driver-i386.c (host_detect_local_cpu):
103 Detect RDPID feature.
104 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
105 * config/i386/i386-c.c (ix86_target_macros_internal):
107 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
108 (ix86_valid_target_attribute_inner_p): Add "rdpid".
109 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
110 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
111 * config/i386/i386.md (define_insn "rdpid"): New.
112 * config/i386/i386.opt Add -mrdpid.
113 * config/i386/immintrin.h (_rdpid_u32): New.
115 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
117 PR rtl-optimization/79541
118 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
119 instead of transforming it into USE.
121 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
123 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
124 If HONOR_SNANS (SFmode) force the input to a register.
125 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
126 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
127 an frsp or similar insn.
129 2017-02-17 Martin Liska <mliska@suse.cz>
131 PR rtl-optimization/79577
132 * params.def (selsched-max-sched-times): Increase minimum to 1.
134 2017-02-17 Martin Liska <mliska@suse.cz>
136 PR rtl-optimization/79574
137 * gcse.c (want_to_gcse_p): Prevent integer overflow.
139 2017-02-17 Martin Liska <mliska@suse.cz>
141 PR tree-optimization/79529
142 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
143 ssa_defined_default_def_p to handle cases which are implicitly
145 * tree-ssa.c (ssa_defined_default_def_p): New function.
146 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
147 which are implicitly defined.
148 * tree-ssa.h (ssa_defined_default_def_p): Declare.
150 2017-02-17 Richard Biener <rguenther@suse.de>
153 * params.def (max-ssa-name-query-depth): Limit to 10.
155 2017-02-17 Richard Biener <rguenther@suse.de>
157 PR tree-optimization/79552
158 * tree-ssa-structalias.c (visit_loadstore): Properly verify
161 2017-02-17 Richard Biener <rguenther@suse.de>
164 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
166 2017-02-17 Marek Polacek <polacek@redhat.com>
169 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
170 (fold_negate_expr): New wrapper.
172 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
174 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
175 Correct terminology and de-emphasize pre-standard behavior.
177 2017-02-16 Alan Modra <amodra@gmail.com>
179 PR rtl-optimization/79286
180 * ira.c (def_dominates_uses): New function.
181 (update_equiv_regs): Don't create an equivalence for insns that
182 may trap where the register def does not dominate the use.
184 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
186 PR rtl-optimization/78127
187 * lra.c (lra): Call lra_eliminate before finish the loop after
190 2017-02-16 Richard Biener <rguenther@suse.de>
192 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
194 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
195 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
196 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
197 (isl_val_int_from_wi): New function.
198 (extract_affine_gmp): Rename to ...
199 (extract_affine_wi): ... this, take a widest_int.
200 (extract_affine_int): Just wrap extract_affine_wi.
201 (add_param_constraints): Use isl_val_int_from_wi.
202 (add_loop_constraints): Likewise, and extract_affine_wi.
204 2017-02-15 Jeff Law <law@redhat.com>
207 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
208 ira_init_register_move_cost_if_necessary.
210 2017-02-15 Martin Sebor <msebor@redhat.com>
213 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
214 removed in a prior commit.
216 2017-02-15 Bin Cheng <bin.cheng@arm.com>
218 PR tree-optimization/79347
219 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
220 counters during peeling.
222 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
224 * Makefile.in (site.exp): Remove "set ISLVER".
226 2017-02-15 Jakub Jelinek <jakub@redhat.com>
229 * real.c (real_from_integer): Call real_convert even for decimal.
231 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
234 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
236 2017-02-14 Andrew Pinski <apinski@cavium.com>
238 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
239 cores and change the partno/implementer to be correct.
240 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
241 the 'B" as the implementer.
242 * config/aarch64/aarch64-tune.md: Regenerate.
244 2017-02-14 Carl Love <cel@us.ibm.com>
246 * config/rs6000/rs6000.c: Add case statement entry to make the
247 xvcvuxdsp built-in argument unsigned.
248 * config/rs6000/vsx.md: Fix the source and return operand types so they
249 match the instruction definitions from the ISA document. Fix typo
250 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
253 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
256 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
257 member early_clobber_alts.
258 * lra-lives.c (reg_early_clobber_p): New.
259 (process_bb_lives): Use it.
260 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
261 (debug_operand_data): Initialize early_clobber_alts.
262 (setup_operand_alternative): Set up early_clobber_alts.
263 (collect_non_operand_hard_regs): Ditto. Pass early clobber
264 alternatives to new_insn_reg.
265 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
267 (lra_update_insn_regno_info): Pass the new arg.
269 2017-02-14 Jakub Jelinek <jakub@redhat.com>
272 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
273 (new_oacc_loop_raw): Don't clear already cleared fields.
276 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
277 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
278 _mm512_prefetch_i64gather_ps): New inline functions and macros.
280 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
283 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
285 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
288 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
289 the extra instruction to the right place to store 128-bit constant
292 2017-02-14 Martin Sebor <msebor@redhat.com>
295 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
296 warning for strings of unknown length.
298 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
300 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
302 2017-02-14 Jeff Law <law@redhat.com>
305 * ira-costs.c (scan_one_insn): Initialize register move costs
306 for pseudos seen in USE/CLOBBER insns.
308 PR tree-optimization/79095
309 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
310 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
311 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
312 if the operands are known to be not equal, then the resulting range
314 (intersect_ranges): If the new range is ~[0,0] and the old range is
315 wide, then prefer ~[0,0].
316 * tree-vrp.c (overflow_comparison_p_1): New function.
317 (overflow_comparison_p): New function.
318 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
319 if NAME is used in an overflow test.
320 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
321 overflow check that can be expressed as an equality test, then adjust
322 ops to be that equality test.
324 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
326 * config/s390/s390-builtin-types.def: Remove flags argument.
327 * config/s390/s390.c (s390_init_builtins): Likewise.
329 2017-02-14 Martin Liska <mliska@suse.cz>
331 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
332 vector. Fix trailing white spaces.
334 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
336 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
339 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
341 PR rtl-optimization/68664
342 * config/arm/arm.c (arm_sched_can_speculate_insn):
343 New function. Declare prototype.
344 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
346 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
348 PR rtl-optimization/68664
349 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
351 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
353 2017-02-14 Amit Pawar <amit.pawar@amd.com>
355 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
356 max skip bytes for function, loop and jump.
358 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
360 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
361 ABS_EXPR for gimple dump.
363 2017-02-14 Jakub Jelinek <jakub@redhat.com>
366 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
368 PR tree-optimization/79408
369 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
370 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
371 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
372 also if rhs1 is INTEGER_CST.
374 2017-02-14 Richard Biener <rguenther@suse.de>
377 * tree-into-ssa.c (insert_phi_nodes): When the function can
378 have abnormal edges rewrite SSA names with broken use-def
379 dominance out of SSA and register them for PHI insertion.
381 2017-02-13 Martin Sebor <msebor@redhat.com>
384 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
385 clearing info.nowrite flag when snprintf size argument is a range.
387 2017-02-13 Jakub Jelinek <jakub@redhat.com>
389 * cprop.c (cprop_jump): Add missing space in string literal.
390 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
391 (get_constraint_for_component_ref): Likewise.
392 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
393 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
394 * lra-constraints.c (process_alt_operands): Likewise.
395 * ipa-inline.c (inline_small_functions): Likewise.
396 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
397 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
398 * trans-mem.c (diagnose_tm_1_op): Likewise.
399 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
400 (grid_parallel_clauses_gridifiable): Likewise.
402 * config/nvptx/mkoffload.c (process): Add space in between
405 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
406 "MOD4_SSE_REGS" and "ALL_REGS".
408 * spellcheck.c (test_data): Add , in between "foo" and "food".
410 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
413 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
414 boundary crossing check and subsequent code generation agree.
416 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
418 * config/aarch64/aarch64.c (has_memory_op): Delete.
419 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
422 2017-02-13 Jakub Jelinek <jakub@redhat.com>
424 PR rtl-optimization/79388
425 PR rtl-optimization/79450
426 * combine.c (distribute_notes): When removing TEM_INSN for which
427 corresponding dest has last value recorded, invalidate that last
430 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
432 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
433 of explicit '@'. Add missing assembly comment marker on branch costs
436 2017-02-13 Nathan Sidwell <nathan@acm.org>
438 * gengtype-lex.l (<in_struct>): Add '/'.
440 2017-02-13 Martin Liska <mliska@suse.cz>
443 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
445 2017-02-13 Richard Biener <rguenther@suse.de>
447 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
449 * configure: Re-generate.
450 * config.in: Likewise.
451 * graphite-dependences.c: Simplify as if
452 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
453 * graphite-isl-ast-to-gimple.c: Likewise.
454 * graphite-optimize-isl.c: Likewise.
455 * graphite-poly.c: Likewise.
456 * graphite-sese-to-poly.c: Likewise.
457 * graphite.h: Likewise.
458 * toplev.c: Include isl/version.h and use isl_version () for
459 printing the ISL version.
460 * doc/install.texi: Update ISL requirement.
462 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
464 * doc/standards.texi (Standards): Update reference to
467 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
469 * doc/extend.texi (Named Address Spaces): sourceware.org now
471 * doc/install.texi (Binaries): Ditto.
474 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
476 * doc/cpp.texi: Replace "stringify"/"stringification" with C
477 standard terminology "stringize"/"stringizing" throughout.
478 * doc/cppinternals.texi: Likewise.
480 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
482 * doc/extend.texi: Fix some spelling mistakes and typos.
483 * doc/invoke.texi: Likewise.
485 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
488 * params.def (inline-min-speedup) Change from 10 to 8.
490 2017-02-11 Jakub Jelinek <jakub@redhat.com>
492 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
495 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
498 * ipa-inline-analysis.c (get_minimal_bb): New function.
499 (record_modified): Use it.
500 (remap_edge_change_prob): Handle also ancestor functions.
502 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
504 * doc/contrib.texi (Contributors): Remove broken link into
505 the Mauve CVS repository.
507 2017-02-11 Jakub Jelinek <jakub@redhat.com>
510 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
511 result computation whenever lhs doesn't have vector mode, not
512 just when it has BLKmode.
514 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
516 * doc/makefile.texi (profiledbootstrap): Refer to the
517 installation instructions only in textual form.
519 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
522 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
524 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
526 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
527 (Specific): Update mingw-w64 reference.
529 (Specific): Remove broken link to Renesas RX processor.
531 2017-02-10 Richard Biener <rguenther@suse.de>
533 * toplev.c (process_options): Do not mention obsolete graphite
534 options when printing sorry message about missing graphite support.
535 Mention -floop-nest-optimize.
537 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
539 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
540 (vtst_p16): Likewise.
541 (vtstq_p8): Likewise.
542 (vtstq_p16): Likewise.
544 (vtstq_p64): Likewise.
545 * config/arm/arm_neon.h (vgetq_lane_p64): New.
546 (vset_lane_p64): New.
547 (vsetq_lane_p64): New.
549 2017-02-10 Jakub Jelinek <jakub@redhat.com>
551 PR tree-optimization/79411
552 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
553 stmt operands are SSA_NAMEs used in abnormal phis.
554 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
557 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
560 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
563 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
565 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
567 2017-02-09 Jakub Jelinek <jakub@redhat.com>
569 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
573 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
574 not arbitrary TREE_CONSTANT.
577 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
578 "omp declare target link" attribute unless is_global_var.
579 * omp-offload.c (find_link_var_op): Likewise.
581 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
582 Chung-Lin Tang <cltang@codesourcery.com>
584 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
586 (gimplify_adjust_omp_clauses): Don't delete TILE.
587 (gimplify_omp_for): Deal with TILE.
588 * internal-fn.c (expand_GOACC_TILE): New function.
589 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
591 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
592 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
594 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
595 avoid DIV for outermost collapse var.
596 (expand_oacc_for): Insert tile element loop as needed. Adjust.
597 Remove out of date comments, fix whitespace.
598 * omp-general.c (omp_extract_for_data): Deal with tiling.
599 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
600 adjust OLF_DIM_BASE value.
601 (struct omp_for_data): Add tiling field.
602 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
603 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
604 for auto loops. Remove default auto determining, moved to
605 oacc_loop_fixed_partitions.
606 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
607 stmts, add e_mask field.
608 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
609 (oacc_thread_numbers): Use oacc_dim_call.
610 (oacc_xform_tile): New.
611 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
612 (finish_oacc_loop): Adjust for ifns vector.
613 (oacc_loop_discover_walk): Append loop abstraction sites to list,
614 add case for GOACC_TILE fns.
615 (oacc_loop_xform_loop): Delete.
616 (oacc_loop_process): Iterate over call list directly, and add
617 handling for GOACC_TILE fns.
618 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
620 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
621 vector partitioning to outer loops. Assign 2 partitions to loops
622 when available. Add TILE handling.
623 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
624 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
625 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
626 * tree.c (omp_clause_num_ops): Adjust TILE ops.
627 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
629 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
631 * configure.ac (ACX_BUGURL): Update.
632 * configure: Regenerate.
634 2017-02-09 Richard Biener <rguenther@suse.de>
636 PR tree-optimization/69823
637 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
638 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
640 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
642 * config/arc/arc-c.def: Add __NPS400__ definition.
643 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
644 (TARGET_NPS400): Define.
646 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
648 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
650 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
652 (arc_cpu_types): Fill the arch_info field with a pointer into the
653 arc_arch_types table.
654 (arc_selected_cpu): Declare.
655 * config/arc/arc.c (arc_selected_cpu): Make global.
656 (arc_selected_arch): Delete.
657 (arc_base_cpu): Delete.
658 (arc_override_options): Remove references to deleted variables,
659 update access to arch information.
660 (ARC_OPT): Update access to arch information.
661 (ARC_OPTX): Likewise.
662 * config/arc/arc.h (arc_base_cpu): Remove declaration.
663 (TARGET_ARC600): Update access to arch information.
664 (TARGET_ARC601): Likewise.
665 (TARGET_ARC700): Likewise.
666 (TARGET_EM): Likewise.
667 (TARGET_HS): Likewise.
668 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
671 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
674 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
675 condition/operands for integer GE/LE/GEU/LEU operations.
677 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
680 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
683 2017-02-08 Martin Jambor <mjambor@suse.cz>
686 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
687 whether allocation happened.
688 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
689 nothing was allocated.
691 2017-02-08 Jakub Jelinek <jakub@redhat.com>
693 PR tree-optimization/79408
694 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
695 constant, but SSA_NAME with a known integer range, use the minimum
696 of that range instead of op1 to determine if modulo can be replaced
697 with its first operand.
699 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
701 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
703 2017-02-08 Richard Biener <rguenther@suse.de>
705 PR tree-optimization/71824
706 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
707 Check all loops contained in the merged region.
709 2017-02-07 Andrew Pinski <apinski@cavium.com>
711 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
713 2017-02-07 Andrew Pinski <apinski@cavium.com>
715 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
716 (thunderxt88): Likewise.
717 (thunderxt81): Disable LSE and change v8.1 to v8.
718 (thunderxt83): Likewise.
720 2017-02-07 Jakub Jelinek <jakub@redhat.com>
721 Richard Biener <rguenther@suse.de>
724 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
725 type from int to size_t.
726 * ira-costs.c (struct_costs_size): Change type from int to size_t.
728 2017-02-07 Jakub Jelinek <jakub@redhat.com>
730 PR rtl-optimization/79386
731 * cprop.c (bypass_conditional_jumps): Initialize
732 bypass_last_basic_block already before splitting bbs after
733 unconditional traps...
734 (bypass_conditional_jumps): ... rather than here.
737 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
738 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
739 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
740 fix -masm=intel patterns.
742 2017-02-07 Richard Biener <rguenther@suse.de>
744 PR tree-optimization/79256
746 * builtins.c (get_object_alignment_2): Use min_align_of_type
747 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
748 and ADJUST_FIELD_ALIGN.
750 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
752 * doc/tm.texi: Regenerate.
753 * stor-layout.c (layout_decl): Adjust.
754 (update_alignment_for_field): Likewise.
755 (place_field): Likewise.
756 (min_align_of_type): Likewise.
757 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
758 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
759 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
760 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
761 * config/frv/frv.c (frv_adjust_field_align): Likewise.
762 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
763 * config/i386/i386.c (x86_field_alignment): Likewise.
764 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
765 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
766 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
767 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
768 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
769 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
773 2017-01-30 Richard Biener <rguenther@suse.de>
775 PR tree-optimization/79256
776 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
777 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
780 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
782 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
783 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
784 builtins to SImode and emit a zero-extend, if necessary.
786 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
788 * docs/invoke.texi (RISC-V Options): Alphabetize.
790 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
792 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
795 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
797 * config/riscv/riscv.c: New file.
798 * gcc/common/config/riscv/riscv-common.c: Likewise.
799 * config.gcc: Likewise.
800 * config/riscv/constraints.md: Likewise.
801 * config/riscv/elf.h: Likewise.
802 * config/riscv/generic.md: Likewise.
803 * config/riscv/linux.h: Likewise.
804 * config/riscv/multilib-generator: Likewise.
805 * config/riscv/peephole.md: Likewise.
806 * config/riscv/pic.md: Likewise.
807 * config/riscv/predicates.md: Likewise.
808 * config/riscv/riscv-builtins.c: Likewise.
809 * config/riscv/riscv-c.c: Likewise.
810 * config/riscv/riscv-ftypes.def: Likewise.
811 * config/riscv/riscv-modes.def: Likewise.
812 * config/riscv/riscv-opts.h: Likewise.
813 * config/riscv/riscv-protos.h: Likewise.
814 * config/riscv/riscv.h: Likewise.
815 * config/riscv/riscv.md: Likewise.
816 * config/riscv/riscv.opt: Likewise.
817 * config/riscv/sync.md: Likewise.
818 * config/riscv/t-elf-multilib: Likewise.
819 * config/riscv/t-linux: Likewise.
820 * config/riscv/t-linux-multilib: Likewise.
821 * config/riscv/t-riscv: Likewise.
822 * configure.ac: Likewise.
823 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
824 Waterman as RISC-V maintainers.
825 * doc/install.texi: Add RISC-V entries.
826 * doc/invoke.texi: Add RISC-V options section.
827 * doc/md.texi: Add RISC-V constraints section.
828 * configure: Regenerated.
830 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
833 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
834 false values to be constant vectors with all 0 or all 1 bits set.
835 (vcondu<mode><mode>): Likewise.
836 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
838 (fpmask_comparison_operator): Update comment.
839 (vecint_comparison_operator): New predicate.
840 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
841 vector conditionals when the true and false values are constant
842 vectors with all 0 bits or all 1 bits set.
844 2017-02-06 Martin Sebor <msebor@redhat.com>
846 PR tree-optimization/79376
847 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
849 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
851 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
852 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
853 to simplify split condition.
855 2017-02-06 Jakub Jelinek <jakub@redhat.com>
857 * omp-expand.c (oxpand_omp_atomic_fetch_op,
858 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
861 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
863 PR rtl-optimization/68664
864 * target.def (can_speculate_insn): New hook.
865 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
866 * doc/tm.texi: Regenerate.
867 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
868 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
869 (rs6000_sched_can_speculate_insn): New function.
871 2017-02-06 Jakub Jelinek <jakub@redhat.com>
873 PR tree-optimization/79284
874 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
875 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
876 vectorizable_mask_load_store, vectorizable_operation,
877 vect_is_simple_cond, get_same_sized_vectype): Use it instead
878 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
879 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
880 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
881 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
882 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
883 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
884 is_gimple_assign (stmt). Replace another such test with
885 is_gimple_assign (stmt).
887 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
890 * config/avr/avr.c (rtl-iter.h): Include it.
891 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
892 (avr_legitimate_combined_insn): ...and implementation.
894 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
896 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
897 * config/s390/s390.c (s390_const_operand_ok)
898 (s390_canonicalize_comparison, s390_extract_part)
899 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
900 (s390_contiguous_bitmask_p, s390_rtx_costs)
901 (legitimize_pic_address): Likewise.
902 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
903 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
904 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
905 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
906 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
908 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
910 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
911 REGNO($0) == REGNO($1).
913 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
915 * config/s390/linux.h(SIZE_TYPE): Add comment.
917 2017-02-06 Julian Brown <julian@codesourcery.com>
918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
919 Virendra Pathak <virendra.pathak@broadcom.com>
921 * config/aarch64/aarch64-cores.def: Change the scheduler
923 * config/aarch64/aarch64.md: Include thunderx2t99.md.
924 * config/aarch64/thunderx2t99.md: New file.
926 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
928 * doc/standards.texi (Go Language): Update link to language
931 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
933 * tree-eh.c (lower_resx): Sanitize profile.
934 (cleanup_empty_eh_move_lp): Likewise.
936 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
939 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
941 * cfgloopmanip.h (loop_version): Update prototype.
942 * modulo-sched.c (sms_schedule): Update call of loop_version.
943 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
944 * tree-parloops.c (gen_parallel_loop): Likewise.
945 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
946 * tree-ssa-loop-split.c (split_loop): Likewise.
947 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
948 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
950 2017-02-05 Martin Liska <mliska@suse.cz>
953 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
955 (print_operand_address): Initialize a struct to zero.
957 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
959 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
960 garbage collector only in textual form.
962 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
964 * doc/extend.texi (x86 specific memory model extensions for
965 transactional memory): Simplify a phrase.
967 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
970 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
971 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
972 (atomic_storedi_1): Likewise.
974 2017-02-04 Jakub Jelinek <jakub@redhat.com>
976 PR tree-optimization/79338
977 * tree-parloops.c (gather_scalar_reductions): Don't call
978 vect_analyze_loop_form for loop->inner before destroying loop's
981 2017-02-03 Martin Sebor <msebor@redhat.com>
983 PR tree-optimization/79327
984 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
985 when precision has resulted in leading zeros.
986 (format_integer): Adjust the likely counter to assume an unknown
987 argument that may be zero is non-zero.
989 2017-02-03 Jason Merrill <jason@redhat.com>
992 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
993 avoid copying non-taken branch.
995 2017-02-03 Jakub Jelinek <jakub@redhat.com>
997 PR tree-optimization/79340
998 * tree-vect-loop.c (vectorizable_reduction): Release
999 vec_defs elements after safe_splicing them into other vectors.
1002 PR tree-optimization/79327
1003 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
1004 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
1006 (format_integer): Use wide_int_to_tree instead of build_int_cst
1007 + to_?hwi. If argmin is NULL, just set argmin and argmax to
1008 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
1009 of shortest and longest sequence.
1011 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
1013 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
1014 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
1016 2017-02-03 Walter Lee <walt@tilera.com>
1019 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
1020 after initial stackframe link reg save.
1021 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
1023 2017-02-03 Jakub Jelinek <jakub@redhat.com>
1026 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
1027 wu for stxssp alternative.
1029 2017-02-03 Martin Sebor <msebor@redhat.com>
1031 PR tree-optimization/79352
1032 * gimple-fold.c (get_range_strlen): Add argument.
1033 (get_range_strlen): Change return type to bool.
1034 (get_maxval_strlen): Pass in a dummy argument.
1035 * gimple-fold.h (get_range_strlen): Change return type to bool.
1036 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
1037 * tree.h (array_at_struct_end_p): Add argument.
1038 * tree.c (array_at_struct_end_p): Handle it.
1040 2017-02-03 Martin Liska <mliska@suse.cz>
1043 * multiple_target.c (create_dispatcher_calls): Redirect edge
1044 from a caller of a dispatcher.
1045 (expand_target_clones): Make the clones local.
1046 (ipa_target_clone): Do both target clones and resolvers.
1047 (ipa_dispatcher_calls): Remove the pass.
1048 (pass_dispatcher_calls::gate): Likewise.
1049 (make_pass_dispatcher_calls): Likewise.
1050 * passes.def (pass_target_clone): Put as very first IPA early
1053 2017-02-03 Martin Liska <mliska@suse.cz>
1055 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
1056 in case of a function with ifunc attribute.
1058 2017-02-03 Martin Liska <mliska@suse.cz>
1060 * cgraph.c (cgraph_node::dump): Dump function version info.
1061 * symtab.c (symtab_node::dump_base): Add missing new line.
1063 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1065 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
1066 (ifcombine_ifandif): Use it.
1068 2017-02-03 Martin Liska <mliska@suse.cz>
1070 * doc/invoke.texi: Document default value for
1071 use-after-scope-direct-emission-threshold.
1073 2017-02-03 Martin Liska <mliska@suse.cz>
1075 PR tree-optimization/79339
1076 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
1077 (format_floating): Likewise.
1079 2017-02-03 Martin Liska <mliska@suse.cz>
1082 * ipa-prop.c (ipa_node_params_t::insert): Remove current
1084 (ipa_node_params_t::remove): Likewise.
1085 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
1086 initialization from removed ipa_node_params_t::insert.
1087 (ipa_node_params::~ipa_node_params): Move from removed
1088 ipa_node_params_t::release.
1089 * symbol-summary.h (symbol_summary::m_released): New member.
1090 Do not release a summary twice. Do not allow to call finalizer
1091 for types of a summary that live in GGC memory.
1093 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1095 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
1098 2017-02-02 Martin Sebor <msebor@redhat.com>
1101 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
1102 (format_string): Tighten up the range of output for non-constant
1103 strings and correct the expected range for wide non-constant strings.
1105 2017-02-02 Martin Sebor <msebor@redhat.com>
1107 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
1110 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
1112 (-fdump-tree-@var): Add to index and document how to come up
1113 with pass-specific option and dump file names.
1114 (-fdump-passes): Clarify where to look for output.
1116 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1119 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
1120 statistics of the analyzed path; allow threading for speed when
1121 any of BBs along the path are optimized for speed.
1123 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
1126 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
1127 settings of the virtual registers.
1130 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
1132 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1133 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
1136 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1138 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
1139 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
1141 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1143 * config/s390/s390.md: Add missing comments with the expanded
1145 * config/s390/vector.md: Likewise.
1146 * config/s390/vx-builtins.md: Likewise.
1148 2017-02-02 Jakub Jelinek <jakub@redhat.com>
1151 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
1152 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
1153 conditions on a single line.
1155 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1157 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1158 __S390_VX__ to __VX__.
1160 2017-02-01 Andrew Pinski <apinski@cavium.com>
1162 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
1163 stmt_info to record_stmt_cost.
1164 (vect_get_known_peeling_cost): Pass stmt_info if known to
1166 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
1167 cpu_vector_cost field into
1168 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
1169 field into vec_int_stmt_cost and vec_fp_stmt_cost.
1170 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
1171 splitting of scalar_stmt_cost and vec_stmt_cost.
1172 (thunderx_vector_cost): Likewise.
1173 (cortexa57_vector_cost): LIkewise.
1174 (exynosm1_vector_cost): Likewise.
1175 (xgene1_vector_cost): Likewise.
1176 (thunderx2t99_vector_cost): Improve after the splitting of the two
1178 (aarch64_builtin_vectorization_cost): Update for the splitting of
1179 scalar_stmt_cost and vec_stmt_cost.
1181 2017-02-01 Torvald Riegel <triegel@redhat.com>
1182 Richard Henderson <rth@redhat.com>
1184 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
1185 conditional on existance of a fast atomic load.
1186 * optabs-query.c (can_atomic_load_p): New function.
1187 * optabs-query.h (can_atomic_load_p): Declare it.
1188 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
1189 no fast atomic load is available for the particular size of access.
1190 (expand_atomic_compare_and_swap): Likewise.
1191 (expand_atomic_load): Likewise.
1192 (expand_atomic_store): Likewise.
1193 (expand_atomic_fetch_op): Likewise.
1194 * testsuite/lib/target-supports.exp
1195 (check_effective_target_sync_int_128): Remove x86 because it provides
1196 no fast atomic load.
1197 (check_effective_target_sync_int_128_runtime): Likewise.
1199 2017-02-01 Richard Biener <rguenther@suse.de>
1201 * graphite.c: Include tree-vectorizer.h for find_loop_location.
1202 (graphite_transform_loops): Provide opt-info for optimized nests.
1203 * tree-parloop.c (parallelize_loops): Provide opt-info for
1206 2017-02-01 Richard Biener <rguenther@suse.de>
1209 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
1212 2017-02-01 Richard Biener <rguenther@suse.de>
1214 PR tree-optimization/71824
1215 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
1216 Verify the loops are valid in the merged SESE region.
1217 (scop_detection::can_represent_loop_1): Check analyzing the
1218 evolution of the number of iterations in the region succeeds.
1220 2017-01-31 Ian Lance Taylor <iant@golang.org>
1222 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
1223 REG_ARGS_SIZE note to 32-bit push insns and call insn.
1225 2017-01-31 David Malcolm <dmalcolm@redhat.com>
1227 PR preprocessor/79210
1228 * input.c (get_substring_ranges_for_loc): Replace line_width
1229 assertion with error-handling.
1231 2017-01-31 Richard Biener <rguenther@suse.de>
1233 PR tree-optimization/77318
1234 * graphite-sese-to-poly.c (extract_affine): Fix assert.
1235 (create_pw_aff_from_tree): Take loop parameter.
1236 (add_condition_to_pbb): Pass loop of the condition to
1237 create_pw_aff_from_tree.
1239 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1241 * config/s390/s390.c (s390_asan_shadow_offset): New function.
1242 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
1244 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1248 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
1250 (convert_int_to_float128): Likewise.
1251 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
1252 (convert_int_to_float128): Likewise.
1253 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
1254 (UNSPEC_IEEE128_CONVERT): Likewise.
1255 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
1256 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
1257 Use local variables for IBM extended format.
1258 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
1259 (fix_trunc<mode>si2_fprs): Likewise.
1260 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
1261 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
1262 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
1263 to know that we can now have integers of all sizes in vector
1265 (fix<uns>_<mode>di2_hw): Likewise.
1266 (float<uns>_<mode>si2_hw): Likewise.
1267 (fix_<mode>si2_hw): Likewise.
1268 (fixuns_<mode>si2_hw): Likewise.
1269 (float<uns>_<mode>di2_hw): Likewise.
1270 (float_<mode>di2_hw): Likewise.
1271 (float_<mode>si2_hw): Likewise.
1272 (floatuns_<mode>di2_hw): Likewise.
1273 (floatuns_<mode>si2_hw): Likewise.
1274 (xscvqp<su>wz_<mode>): Delete, no longer used.
1275 (xscvqp<su>dz_<mode>): Likewise.
1276 (xscv<su>dqp_<mode>): Likewise.
1277 (ieee128_mfvsrd_64bit): Likewise.
1278 (ieee128_mfvsrd_32bit): Likewise.
1279 (ieee128_mfvsrwz): Likewise.
1280 (ieee128_mtvsrw): Likewise.
1281 (ieee128_mtvsrd_64bit): Likewise.
1282 (ieee128_mtvsrd_32bit): Likewise.
1284 2017-01-31 Martin Liska <mliska@suse.cz>
1287 * ipa-prop.c (ipa_free_all_node_params): Call release method
1288 instead of ~sumbol_summary to not to trigger double times
1291 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
1293 PR tree-optimization/71691
1294 * bitmap.h (class auto_bitmap): New.
1295 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
1296 is_maybe_undefined instead of ssa_undefined_value_p.
1298 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1300 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1301 __S390_ARCH_LEVEL__ to __ARCH__.
1303 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1305 PR tree-optimization/79267
1306 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
1307 if should_remove_lhs_p is true.
1309 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
1312 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
1313 (add_alignment_attribute): New.
1314 (base_type_die): Add alignment attribute.
1315 (subrange_type_die): Likewise.
1316 (modified_type_die): Likewise.
1317 (gen_array_type_die): Likewise.
1318 (gen_descr_array_type_die: Likewise.
1319 (gen_enumeration_type_die): Likewise.
1320 (gen_subprogram_die): Likewise.
1321 (gen_variable_die): Likewise.
1322 (gen_field_die): Likewise.
1323 (gen_ptr_to_mbr_type_die): Likewise.
1324 (gen_struct_or_union_type_die): Likewise.
1325 (gen_subroutine_type_die): Likewise.
1326 (gen_typedef_die): Likewise.
1327 (base_type_cmp): Compare alignment attribute.
1329 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1332 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
1333 (setb_unsigned) New pattern for setb with CCUNS.
1334 * config/rs6000/rs6000.c (expand_block_compare): Use a different
1335 subfc./subfe sequence to avoid overflow problems. Generate a
1336 shorter sequence with cmpld/setb for power9.
1337 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
1338 for generating subfc. instruction.
1339 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
1340 now uses this instruction.
1342 2017-01-30 Ian Lance Taylor <iant@google.com>
1345 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
1346 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
1348 2017-01-30 Martin Sebor <msebor@redhat.com>
1350 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
1351 Move constant to the right of a relational operator.
1352 (get_mpfr_format_length, format_character, format_string): Ditto.
1353 (should_warn_p, maybe_warn): Same.
1355 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
1357 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1360 * asan.c (get_translation_unit_decl): Remove function.
1361 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
1363 2017-01-30 Martin Liska <mliska@suse.cz>
1365 PR gcov-profile/79259
1366 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
1369 2017-01-30 Martin Liska <mliska@suse.cz>
1372 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
1373 Initialize variables with NULL value.
1375 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
1378 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
1380 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
1382 2017-01-30 Richard Biener <rguenther@suse.de>
1384 PR tree-optimization/79276
1385 * tree-vrp.c (process_assert_insertions): Properly adjust common
1386 when removing a duplicate.
1388 * gcc.dg/torture/pr79276.c: New testcase.
1390 2017-01-30 Richard Biener <rguenther@suse.de>
1392 PR tree-optimization/79256
1393 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
1394 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
1396 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
1398 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
1401 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
1402 ("*r<noxa>sbg_<mode>_sll_bitmask")
1403 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
1404 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
1405 Use contiguous_bitmask_nowrap_operand.
1407 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1410 * config/rs6000/altivec.h (vec_xl): Revise #define.
1411 (vec_xst): Likewise.
1413 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
1415 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
1417 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
1419 PR rtl-optimization/79194
1420 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
1421 traps before call to bypass_conditional_jumps.
1423 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1425 PR tree-optimization/71374
1426 * lra-constraints.c (check_conflict_input_operands): New.
1427 (match_reload): Use it.
1429 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1432 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
1433 account to calculate conflict_set.
1435 2017-01-27 Bin Cheng <bin.cheng@arm.com>
1437 PR rtl-optimization/78559
1438 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
1439 other_insn in combine.
1441 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
1443 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
1444 uint16_type_node for BT_UINT16.
1446 2017-01-27 David Malcolm <dmalcolm@redhat.com>
1448 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
1449 "RTL Tests" to menu.
1450 (GIMPLE Tests): New node.
1451 (RTL Tests): New node.
1453 2017-01-27 Richard Biener <rguenther@suse.de>
1455 PR tree-optimization/79245
1456 * tree-loop-distribution.c (distribute_loop): Apply cost
1457 modeling also to detected patterns.
1459 2017-01-27 Richard Biener <rguenther@suse.de>
1461 PR tree-optimization/71433
1462 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
1463 (compare_assert_loc): New function.
1464 (process_assert_insertions): Sort and optimize assert locations
1465 to remove duplicates and push down identical assertions on
1466 edges to their destination block.
1468 2017-01-27 Richard Biener <rguenther@suse.de>
1470 PR tree-optimization/79244
1471 * tree-vrp.c (remove_range_assertions): Forcefully propagate
1472 out SSA names even if abnormal.
1474 2017-01-27 Jakub Jelinek <jakub@redhat.com>
1476 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
1477 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
1478 instead of MPFR_RNDN.
1480 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
1483 * arm.c (arm_option_override): Don't call build_target_option_node
1484 until after doing all option overrides.
1485 (arm_valid_target_attribute_tree): Likewise.
1487 2017-01-27 Martin Liska <mliska@suse.cz>
1489 * doc/invoke.texi (-fprofile-arcs): Document profiling support
1490 for {cd}tors and C++ {cd}tors.
1492 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
1494 * config/s390/s390.md ("*setmem_long_and")
1495 ("*setmem_long_and_31z"): Use zero_extend instead of and.
1497 2017-01-26 Martin Sebor <msebor@redhat.com>
1499 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
1502 2017-01-26 Martin Sebor <msebor@redhat.com>
1504 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
1505 HAVE_DFmode before using XFmode or DFmode.
1506 (parse_directive): Avoid using the z length modifier to avoid
1507 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
1510 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
1511 to accept adjustment as an array.
1512 (get_int_range): New function.
1513 (struct directive): Make width and prec arrays.
1514 (directive::set_width, directive::set_precision): Call get_int_range.
1515 (format_integer, format_floating): Handle width and precision ranges.
1516 (format_string, parse_directive): Same.
1518 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1521 * dwarf2out.c (generate_skeleton_bottom_up): For children with
1522 comdat_type_p set, just clone them, but keep the children in the
1526 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
1527 which have direct callers with -fvar-tracking-assignments enabled
1529 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
1530 inside of type units.
1532 2017-01-26 Martin Sebor <msebor@redhat.com>
1535 * gimple-ssa-sprintf.c (struct result_range): Add likely and
1537 (struct format_result): Replace number_chars, number_chars_min,
1538 and number_chars_max with a single member of struct result_range.
1540 (format_result::operator+=): Adjust.
1541 (struct fmtresult): Remove bounded. Handle likely and unlikely
1543 (fmtresult::adjust_for_width_or_precision): New function.
1544 (fmtresult:type_max_digits): New function.
1545 (bytes_remaining): Handle likely and unlikely counters.
1546 (min_bytes_remaining): Remove.
1547 (format_percent): Simplify.
1548 (format_integer, format_floating): Set likely and unlikely counters.
1549 (get_string_length, format_character, format_string): Same.
1550 (format_plain, should_warn_p): New function.
1551 (maybe_warn): Call should_warn_p. Update diagnostic messages
1552 and handle those for all directives, including plain strings.
1553 (format_directive): Handle likely and unlikely counters.
1554 Remove unnecessary quoting from diagnostics. Add an informational
1556 (add_bytes): Remove.
1557 (pass_sprintf_length::compute_format_length): Simplify.
1558 (try_substitute_return_value): Handle likely and unlikely counters.
1560 2017-01-26 Carl Love <cel@us.ibm.com>
1562 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
1563 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
1565 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
1568 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
1569 endianess for subregs into account.
1570 * lra-constraints.c (lra_constraints): Do risky transformations
1571 always on the first iteration.
1572 * lra-lives.c (check_pseudos_live_through_calls): Add arg
1573 last_call_used_reg_set.
1574 (process_bb_lives): Define and use last_call_used_reg_set.
1575 * lra.c (lra): Always continue after lra_constraints on the first
1578 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
1580 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
1582 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
1584 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1586 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
1587 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
1588 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
1589 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
1590 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
1591 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
1592 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
1593 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
1594 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
1596 2017-01-26 Marek Polacek <polacek@redhat.com>
1599 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
1600 for the third operand.
1602 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1605 * omp-low.c (struct omp_context): Add simt_stmt field.
1606 (scan_omp_for): Return omp_context *.
1607 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
1608 context to the _simt_ SIMD stmt.
1609 (lower_omp_for): For combined SIMD with sibling _simt_
1610 SIMD, make sure to use the same decls in _looptemp_
1611 clauses as in the sibling.
1613 2017-01-26 David Sherwood <david.sherwood@arm.com>
1616 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
1619 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1622 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
1623 emit fld b; fld a; if possible.
1625 * brig-builtins.def: Update copyright years.
1626 * config/arm/arm_acle_builtins.def: Update copyright years.
1628 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1631 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
1632 constraint instead of o for the stxsd instruction.
1634 2017-01-25 Carl Love <cel@us.ibm.com>
1636 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
1637 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
1639 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
1641 * doc/invoke.texi (C++ Dialect Options): Fix typo.
1643 2017-01-25 Richard Biener <rguenther@suse.de>
1645 PR tree-optimization/69264
1646 * target.def (vector_alignment_reachable): Improve documentation.
1647 * doc/tm.texi: Regenerate.
1648 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
1650 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
1651 earlier changes with respect to TYPE_USER_ALIGN.
1652 (vector_alignment_reachable_p): Likewise. Improve dumping.
1654 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1657 * config/arm/arm.md (xordi3): Force constant operand into a register
1660 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1662 * doc/invoke.texi (-fstore-merging): Correct default optimization
1663 levels at which it is enabled.
1664 (-O): Move -fstore-merging from list to...
1667 2017-01-25 Richard Biener <rguenther@suse.de>
1670 * omp-expand.c: Include debug.h.
1671 (expand_omp_taskreg): Make sure to generate early debug before
1672 outlining anything from a function.
1673 (expand_omp_target): Likewise.
1674 (grid_expand_target_grid_body): Likewise.
1676 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
1679 * asan.c (get_translation_unit_decl): New function.
1680 (asan_add_global): Extract modules file name from globals
1681 TRANSLATION_UNIT_DECL name.
1683 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
1686 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
1687 for long calls with APCS frame and VFP.
1689 2017-01-24 David Malcolm <dmalcolm@redhat.com>
1691 * cfg.c (original_copy_tables_initialized_p): New function.
1692 * cfg.h (original_copy_tables_initialized_p): New decl.
1693 * cfgrtl.c (relink_block_chain): Guard the call to
1694 free_original_copy_tables with a call to
1695 original_copy_tables_initialized_p.
1696 * cgraph.h (symtab_node::native_rtl_p): New decl.
1697 * cgraphunit.c (symtab_node::native_rtl_p): New function.
1698 (symtab_node::needed_p): Don't assert for early assembly output
1699 for __RTL functions.
1700 (cgraph_node::finalize_function): Set "force_output" for __RTL
1702 (cgraph_node::analyze): Bail out early for __RTL functions.
1703 (analyze_functions): Update assertion to support __RTL functions.
1704 (cgraph_node::expand): Bail out early for __RTL functions.
1705 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
1707 * function.h (struct function): Update comment for field
1709 * gimple-expr.c: Include "tree-pass.h".
1710 (gimple_has_body_p): Return false for __RTL functions.
1711 * Makefile.in (OBJS): Add run-rtl-passes.o.
1712 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
1714 (gcc::pass_manager::get_clean_slate): New accessor.
1715 * passes.c: Include "insn-addr.h".
1716 (should_skip_pass_p): Add logging. Update logic for running
1717 "expand" to be compatible with both __GIMPLE and __RTL. Guard
1718 property-provider override so it is only done for gimple passes.
1720 (skip_pass): New function.
1721 (execute_one_pass): Call skip_pass when skipping passes.
1722 * read-md.c (md_reader::read_char): Support filtering
1723 the input to a subset of line numbers.
1724 (md_reader::md_reader): Initialize fields
1725 m_first_line and m_last_line.
1726 (md_reader::read_file_fragment): New function.
1727 * read-md.h (md_reader::read_file_fragment): New decl.
1728 (md_reader::m_first_line): New field.
1729 (md_reader::m_last_line): New field.
1730 * read-rtl-function.c (function_reader::create_function): Only
1731 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
1732 curr_properties. Set DECL_INITIAL to a dummy block.
1733 (read_rtl_function_body_from_file_range): New function.
1734 * read-rtl-function.h (read_rtl_function_body_from_file_range):
1736 * run-rtl-passes.c: New file.
1737 * run-rtl-passes.h: New file.
1739 2017-01-24 Jeff Law <law@redhat.com>
1741 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
1744 2017-01-24 Bin Cheng <bin.cheng@arm.com>
1746 PR tree-optimization/79159
1747 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
1748 (record_nonwrapping_iv): Improve boundary using above function if no
1749 value range information.
1751 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
1752 Martin Jambor <mjambor@suse.cz>
1754 * brig-builtins.def: New file.
1755 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
1756 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
1757 (DEF_HSAIL_SAT_BUILTIN): Likewise.
1758 (DEF_HSAIL_INTR_BUILTIN): Likewise.
1759 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
1760 * builtin-types.def (BT_INT8): New.
1761 (BT_INT16): Likewise.
1762 (BT_UINT8): Likewise.
1763 (BT_UINT16): Likewise.
1764 (BT_FN_ULONG): Likewise.
1765 (BT_FN_UINT_INT): Likewise.
1766 (BT_FN_UINT_ULONG): Likewise.
1767 (BT_FN_UINT_LONG): Likewise.
1768 (BT_FN_UINT_PTR): Likewise.
1769 (BT_FN_ULONG_PTR): Likewise.
1770 (BT_FN_INT8_FLOAT): Likewise.
1771 (BT_FN_INT16_FLOAT): Likewise.
1772 (BT_FN_UINT32_FLOAT): Likewise.
1773 (BT_FN_UINT16_FLOAT): Likewise.
1774 (BT_FN_UINT8_FLOAT): Likewise.
1775 (BT_FN_UINT64_FLOAT): Likewise.
1776 (BT_FN_UINT16_UINT32): Likewise.
1777 (BT_FN_UINT32_UINT16): Likewise.
1778 (BT_FN_UINT16_UINT16_UINT16): Likewise.
1779 (BT_FN_INT_PTR_INT): Likewise.
1780 (BT_FN_UINT_PTR_UINT): Likewise.
1781 (BT_FN_LONG_PTR_LONG): Likewise.
1782 (BT_FN_ULONG_PTR_ULONG): Likewise.
1783 (BT_FN_VOID_UINT64_UINT64): Likewise.
1784 (BT_FN_UINT8_UINT8_UINT8): Likewise.
1785 (BT_FN_INT8_INT8_INT8): Likewise.
1786 (BT_FN_INT16_INT16_INT16): Likewise.
1787 (BT_FN_INT_INT_INT): Likewise.
1788 (BT_FN_UINT_FLOAT_UINT): Likewise.
1789 (BT_FN_FLOAT_UINT_UINT): Likewise.
1790 (BT_FN_ULONG_UINT_UINT): Likewise.
1791 (BT_FN_ULONG_UINT_PTR): Likewise.
1792 (BT_FN_ULONG_ULONG_ULONG): Likewise.
1793 (BT_FN_UINT_UINT_UINT): Likewise.
1794 (BT_FN_VOID_UINT_PTR): Likewise.
1795 (BT_FN_UINT_UINT_PTR: Likewise.
1796 (BT_FN_UINT32_UINT64_PTR): Likewise.
1797 (BT_FN_INT_INT_UINT_UINT): Likewise.
1798 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
1799 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
1800 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
1801 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
1802 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
1803 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
1804 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
1805 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
1806 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
1807 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
1808 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
1809 * doc/frontends.texi: List BRIG FE.
1810 * doc/install.texi (Testing): Add BRIG tesring requirements.
1811 * doc/invoke.texi (Overall Options): Mention BRIG.
1812 * doc/standards.texi (Standards): Doucment BRIG HSA version.
1814 2017-01-24 Richard Biener <rguenther@suse.de>
1816 PR translation/79208
1817 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
1819 2017-01-24 Martin Jambor <mjambor@suse.cz>
1822 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
1823 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
1826 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
1829 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
1830 casts from signed to unsigned really don't have a range.
1832 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
1834 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
1835 GMP_RNDx for compatiblity.
1837 2017-01-24 Martin Liska <mliska@suse.cz>
1840 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
1841 that would prevent us to call alloca with -1 as argument.
1843 2017-01-24 Jakub Jelinek <jakub@redhat.com>
1845 * dwarf2out.c (output_compilation_unit_header, output_file_names):
1846 Avoid -Wformat-security warning.
1848 2017-01-23 Andrew Pinski <apinski@cavium.com>
1850 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
1853 2017-01-23 Martin Sebor <msebor@redhat.com>
1856 * gimple-ssa-sprintf.c (warn_level): New global.
1857 (format_integer): Use it here and throughout the rest of the file.
1858 Use the same switch to compute sign as base.
1859 (maybe_warn): New function.
1860 (format_directive): Factor out warnings into maybe_warn.
1861 Add debugging output. Use warn_level.
1862 (add_bytes): Use warn_level.
1863 (pass_sprintf_length::compute_format_length): Add debugging output.
1864 (try_substitute_return_value): Same.
1865 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
1868 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
1869 (struct fmtresult, format_integer, format_floating): Adjust.
1870 (fmtresult::fmtresult): Set max correctly in two argument ctor.
1871 (get_string_length, format_string,format_directive): Same.
1872 (pass_sprintf_length::compute_format_length): Same.
1873 (try_substitute_return_value): Simplify slightly.
1876 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
1877 (fmtresult::operator+=): Outlined.
1878 (struct fmtresult): Add ctors.
1879 (struct conversion_spec): Rename...
1880 (struct directive): ...to this. Add and remove data members.
1881 (directive::set_width, directive::set_precision): New functions.
1882 (format_percent): Use fmtresult ctor.
1883 (get_width_and_precision): Remove.
1884 (format_integer): Make naming changes. Avoid computing width and
1886 (format_floating): Same. Adjust indentation.
1887 (format_character, format_none): New functions.
1888 (format_string): Moved character handling to format_character.
1889 (format_directive): Remove arguments, change return type.
1890 (parse_directive): New function.
1891 (pass_sprintf_length::compute_format_length): Move directive
1892 parsing to parse_directive.
1894 2017-01-23 Jakub Jelinek <jakub@redhat.com>
1896 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
1897 (assign_assembler_name_if_needed): ... this.
1898 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
1899 (assign_assembler_name_if_needed): ... this.
1900 (free_lang_data_in_cgraph): Adjust callers.
1901 * cgraphunit.c (cgraph_node::analyze): Likewise.
1902 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
1905 2017-01-23 Richard Biener <rguenther@suse.de>
1907 PR tree-optimization/79088
1908 PR tree-optimization/79188
1909 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
1910 resetting loop bounds after last path deletion. Reset loop
1911 bounds of the target loop, make code match the comments.
1912 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
1913 Make sure loops need no fixups.
1915 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1917 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
1918 exponent support with double type for first argument.
1919 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
1920 type returned by __builtin_vec_extract_sig,
1921 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
1922 functions from "vector int" to "vector unsigned int" or from
1923 "vector long long int" to "vector unsigned long long int".
1924 Changed type returned by __builtin_vec_extract_exp,
1925 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
1926 functions from "vector int" to "vector unsigned int" or from
1927 "vector long long int" to "vector unsigned long long int".
1928 Changed return type of __builtin_vec_test_data_class,
1929 __builtin_vec_test_data_class_sp, and
1930 __builtin_vec_test_data_class_dp from "vector int" to
1931 "vector bool int" or from "vector long long int" to "vector bool
1932 long long int" and changed second argument type from "unsigned
1933 int" to "int". Added new overloaded function forms "vector float
1934 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
1935 "vector float __builtin_vec_insert_exp_sp (vector float, vector
1936 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
1937 double, vector unsigned long long int)" and "vector double
1938 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
1939 long int)". Changed return type of
1940 __builtin_scalar_test_data_class and
1941 __builtin_scalar_test_data_class_sp and
1942 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
1943 int" and changed second argument from "unsigned int" to "int".
1944 Changed type returned by __builtin_scalar_test_neg,
1945 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
1946 from "int" to "bool int". Added new overloaded function form
1947 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
1948 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
1949 exponent double-precision with floating point first argument.
1950 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
1951 documentation of scalar_test_data_class, scalar_test_neg,
1952 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
1953 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
1954 vec_test_data_class built-in functions to reflect refinements in
1955 their type signatures.
1957 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
1959 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
1961 (aarch64_elf_asm_destructor): Likewise.
1963 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
1965 PR rtl-optimization/78634
1966 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
1967 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
1968 * ifcvt.c (noce_try_cmove): Add missing cost check.
1970 PR rtl-optimization/71724
1971 * combine.c (if_then_else_cond): Look for situations where it is
1972 beneficial to undo the work of one of the recursive calls.
1974 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1976 PR tree-optimization/70754
1977 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1978 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1979 combined stmt before it if not NULL.
1980 (combine_chains): Process refs reversely and compute dominance point
1983 2017-01-23 Martin Liska <mliska@suse.cz>
1985 PR tree-optimization/79196
1986 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
1987 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
1989 (strlen_optimize_stmt): Call the renamed function.
1991 2017-01-23 Michael Matz <matz@suse.de>
1993 PR tree-optimization/78384
1994 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
1996 2017-01-23 Richard Biener <rguenther@suse.de>
1998 PR tree-optimization/79186
1999 * tree-vrp.c (register_new_assert_for): Make sure we've seen
2000 both incoming edges before moving an assert.
2002 2017-01-23 Martin Jambor <mjambor@suse.cz>
2004 * ipa-prop.c (load_from_param_1): Removed.
2005 (load_from_unmodified_param): Bits from load_from_param_1 put back
2007 (load_from_param): Removed.
2008 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
2009 with stmt. Reverted back to use of load_from_unmodified_param.
2011 2017-01-23 Martin Jambor <mjambor@suse.cz>
2014 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
2015 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
2016 field a pointer to garbage collected vector, mark lattices and
2017 ipcp_orig_node with GTY((skip)).
2018 (ipa_get_param_count): Adjust to descriptors being a pointer.
2019 (ipa_get_param): Likewise.
2020 (ipa_get_type): Likewise.
2021 (ipa_get_param_move_cost): Likewise.
2022 (ipa_set_param_used): Likewise.
2023 (ipa_get_controlled_uses): Likewise.
2024 (ipa_set_controlled_uses): Likewise.
2025 (ipa_is_param_used): Likewise.
2026 (ipa_node_params_t): Move into garbage collector. New methods insert
2028 (ipa_node_params_sum): Annotate wth GTY(()).
2029 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
2031 (ipa_load_from_parm_agg): Adjust declaration.
2032 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
2033 * ipa-profile.c (ipa_profile): Likewise.
2034 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
2035 (ipa_populate_param_decls): Make descriptors parameter garbage
2037 (ipa_dump_param): Adjust to descriptors being a pointer.
2038 (ipa_alloc_node_params): Likewise.
2039 (ipa_initialize_node_params): Likewise.
2040 (load_from_param_1): Make descriptors parameter garbage collected.
2041 (load_from_unmodified_param): Likewise.
2042 (load_from_param): Likewise.
2043 (ipa_load_from_parm_agg): Likewise.
2044 (ipa_node_params::~ipa_node_params): Removed.
2045 (ipa_free_all_node_params): Remove call to delete operator.
2046 (ipa_node_params_t::insert): New.
2047 (ipa_node_params_t::remove): Likewise.
2048 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
2049 copy known_csts and known_contexts vectors.
2050 (ipa_read_node_info): Adjust to descriptors being a pointer.
2051 (ipcp_modif_dom_walker): Make m_descriptors field garbage
2053 (ipcp_transform_function): Make descriptors variable garbage
2056 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
2058 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
2059 * config/i386/avx512dqintrin.h: Ditto.
2060 * config/i386/avx512fintrin.h: Ditto.
2061 * gcc/config/i386/i386.c: Handle new builtins.
2062 * config/i386/i386-builtin.def: Add new builtins.
2063 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
2064 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
2066 2017-01-23 Jakub Jelinek <jakub@redhat.com>
2067 Martin Liska <mliska@suse.cz>
2069 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
2070 * asan.c (asan_expand_poison_ifn): Support stores and use
2071 appropriate ASAN report function.
2072 * internal-fn.c (expand_ASAN_POISON_USE): New function.
2073 * internal-fn.def (ASAN_POISON_USE): Declare.
2074 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
2075 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
2076 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
2077 ASAN_POISON calls w/o LHS.
2078 * tree-ssa.c (execute_update_addresses_taken): Create clobber
2079 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
2080 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
2081 * gimplify.c (asan_poison_variables): Add attribute
2082 use_after_scope_memory to variables that really needs to live
2084 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
2085 having the attribute.
2087 2017-01-23 Martin Liska <mliska@suse.cz>
2089 * asan.c (create_asan_shadow_var): New function.
2090 (asan_expand_poison_ifn): Likewise.
2091 * asan.h (asan_expand_poison_ifn): New declaration.
2092 * internal-fn.c (expand_ASAN_POISON): Likewise.
2093 * internal-fn.def (ASAN_POISON): New builtin.
2094 * sanopt.c (pass_sanopt::execute): Expand
2095 asan_expand_poison_ifn.
2096 * tree-inline.c (copy_decl_for_dup_finish): Make function
2098 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
2099 * tree-ssa.c (is_asan_mark_p): New function.
2100 (execute_update_addresses_taken): Rewrite local variables
2101 (identified just by use-after-scope as addressable) into SSA.
2103 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2105 * doc/install.texi (Specific): opensource.apple.com uses https
2106 now. Remove trailing slash.
2108 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2110 * README.Portability: Remove note on an Irix compatibility issue.
2112 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
2114 * gcov.c (INCLUDE_ALGORITHM): Define.
2115 (INCLUDE_VECTOR): Define.
2116 No longer include <vector> and <algorithm> directly.
2118 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
2120 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
2122 * doc/invoke.texi (Code Gen Options): Ditto.
2124 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
2127 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
2129 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
2131 rtl-optimization/79125
2132 * cprop.c (local_cprop_pass): Handle cases where we make an
2135 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
2139 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
2140 read from, for big endian.
2142 2017-01-20 Jiong Wang <jiong.wang@arm.com>
2144 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
2145 register pauth builtins for LP64 only.
2147 2017-01-20 Marek Polacek <polacek@redhat.com>
2150 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
2153 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2155 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
2157 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
2158 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
2159 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
2161 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2164 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
2165 in big-endian mode when they are not a single duplicated value.
2167 2017-01-20 Richard Biener <rguenther@suse.de>
2169 * BASE-VER: Bump to 7.0.1.
2171 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2173 * omp-low.c (omplow_simd_context): New struct. Use it...
2174 (lower_rec_simd_input_clauses): ...here and...
2175 (lower_rec_input_clauses): ...here to hold common data. Adjust all
2176 references to idx, lane, max_vf, is_simt.
2178 2017-01-20 Graham Markall <graham.markall@embecosm.com>
2180 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
2183 2017-01-20 Martin Jambor <mjambor@suse.cz>
2185 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
2186 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
2188 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
2189 (GTFILES): Rename hsa.c to hsa-common.c.
2190 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
2191 * hsa-dump.c: Likewise.
2192 * hsa-gen.c: Likewise.
2193 * hsa-regalloc.c: Likewise.
2194 * ipa-hsa.c: Likewise.
2195 * omp-expand.c: Likewise.
2196 * omp-low.c: Likewise.
2197 * toplev.c: Likewise.
2199 2017-01-20 Marek Polacek <polacek@redhat.com>
2202 * doc/invoke.texi: Document -Wduplicated-branches.
2203 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
2204 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
2205 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
2206 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
2207 return 0 only when not OEP_LEXICOGRAPHIC.
2208 (fold_build_cleanup_point_expr): Use the expression
2209 location when building CLEANUP_POINT_EXPR.
2210 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
2211 * tree.c (add_expr): Handle error_mark_node.
2213 2017-01-20 Martin Liska <mliska@suse.cz>
2216 * tree-profile.c (init_ic_make_global_vars): Do not call
2218 (gimple_init_gcov_profiler): Likewise.
2220 2017-01-20 Martin Liska <mliska@suse.cz>
2223 * cgraph.h (maybe_create_reference): Remove argument and
2225 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
2227 * ipa-cp.c (create_specialized_node): Likewise.
2228 * symtab.c (symtab_node::maybe_create_reference): Handle
2229 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
2231 2017-01-20 Martin Liska <mliska@suse.cz>
2233 * read-rtl-function.c (function_reader::create_function): Use
2234 build_decl instread of build_decl_stat.
2236 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
2238 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
2239 * config/i386/avx512dqintrin.h: Ditto.
2240 * config/i386/avx512fintrin.h: Ditto.
2241 * config/i386/i386-builtin-types.def: Add new types.
2242 * gcc/config/i386/i386.c: Handle new types.
2243 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
2244 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
2245 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
2246 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
2247 (__builtin_ia32_kshiftridi): New.
2248 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
2250 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
2254 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
2255 define to rs6000_init_stack_protect_guard.
2256 (rs6000_init_stack_protect_guard): New function.
2258 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2259 Yunqiang Su <yunqiang.su@imgtec.com>
2261 * config.gcc (supported_defaults): Add madd4.
2262 (with_madd4): Add validation.
2263 (all_defaults): Add madd4.
2264 * config/mips/mips.opt (mmadd4): New option.
2265 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2267 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2269 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
2270 (ISA_HAS_FUSED_MADD4): Likewise.
2271 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
2272 * gcc/doc/install.texi (--with-madd4): Document the new option.
2274 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2276 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
2277 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
2278 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
2279 (aarch64_init_pauth_hint_builtins): New.
2280 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
2281 (aarch64_expand_builtin): Expand new builtins.
2283 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2285 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
2286 * combine-stack-adj.c (no_unhandled_cfa): Handle
2287 REG_CFA_TOGGLE_RA_MANGLE.
2288 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
2289 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
2290 info for return address signing.
2291 (aarch64_expand_epilogue): Likewise.
2293 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2295 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
2296 * config/aarch64/aarch64-protos.h
2297 (aarch64_return_address_signing_enabled): New declaration.
2298 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
2300 (aarch64_expand_prologue): Sign return address before it's pushed onto
2302 (aarch64_expand_epilogue): Authenticate return address fetched from
2304 (aarch64_override_options): Sanity check for ILP32 and ISA level.
2305 (aarch64_attributes): New function attributes for "sign-return-address".
2306 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
2307 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
2308 ("*do_return"): Generate combined instructions according to key index.
2309 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
2310 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
2312 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
2313 * config/aarch64/aarch64.opt (msign-return-address=): New.
2314 * doc/extend.texi (AArch64 Function Attributes): Documents
2315 "sign-return-address=".
2316 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
2318 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2320 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
2321 overall option summary.
2323 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2325 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
2326 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
2327 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
2328 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
2330 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
2332 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
2333 -mpower9-minmax by default for -mcpu=power9.
2334 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
2335 128-bit floating point.
2337 2017-01-20 Alan Modra <amodra@gmail.com>
2339 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
2340 optimizing for size.
2342 2017-01-20 Alan Modra <amodra@gmail.com>
2345 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
2346 for strcmp and strncmp from corresponding builtin decl.
2348 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2350 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
2351 instead of i386/rtems-64.h.
2352 * config/i386/rtems-64.h: Remove.
2354 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2358 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
2360 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
2362 2017-01-19 Tamar Christina <tamar.christina@arm.com>
2364 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2365 Change int to HOST_WIDE_INT.
2366 * config/aarch64/aarch64-protos.h
2367 (aarch64_simd_gen_const_vector_dup): Likewise.
2368 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
2370 2017-01-19 David Malcolm <dmalcolm@redhat.com>
2372 * langhooks-def.h (lhd_type_for_size): New decl.
2373 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
2374 * langhooks.c (lhd_type_for_size): New function, taken from
2377 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
2379 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
2380 define_bypass for CR latency.
2381 (power9-cracked-alu): Update bypass latency and remove power9-branch.
2382 (power9-alu2): Add define_bypass for CR latency.
2384 (power9-mul): Update insn latency.
2385 (power9-mul-compare): Update insn latency, bypass latency and remove
2388 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2390 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
2392 * config/aarch64/aarch64.md
2393 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
2394 aarch64_nopcrelative_literal_loads.
2395 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
2397 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
2399 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
2401 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
2403 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
2406 * config.gcc (supported_defaults): Add lxc1-sxc1.
2407 (with_lxc1_sxc1): Add validation.
2408 (all_defaults): Add lxc1-sxc1.
2409 * config/mips/mips.opt (mlxc1-sxc1): New option.
2410 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2412 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2413 __mips_no_lxc1_sxc1.
2414 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
2415 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
2416 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
2418 2017-01-19 Richard Biener <rguenther@suse.de>
2420 PR tree-optimization/72488
2421 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
2422 sure to restore SSA info.
2423 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
2425 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
2427 PR rtl-optimization/79121
2428 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
2429 of the inner type when shifting an extended value.
2431 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
2434 * symtab.c (symtab_node::equal_address_to): Fix comparing of
2435 interposable aliases.
2437 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
2440 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
2441 Use the evmergelohi instruction.
2442 (mov_si<mode>_e500_subreg4_2_le): Likewise.
2443 (mov_sitf_e500_subreg8_2_be): Likewise.
2444 (mov_sitf_e500_subreg12_2_le): Likewise.
2445 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
2446 (mov_si<mode>_e500_subreg4_2_be): Likewise.
2447 (mov_sitf_e500_subreg8_2_le): Likewise.
2448 (mov_sitf_e500_subreg12_2_be): Likewise.
2450 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2452 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
2453 attribute from vecsimple to vecperm.
2454 (altivec_vbpermq2): Likewise.
2456 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2459 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
2461 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2462 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
2463 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
2464 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
2465 case where N arg is SIZE_MAX.
2466 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
2467 (cmpstrsi): Add pattern.
2469 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2472 __builtin_vec_revb builtins.
2473 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
2474 built-in functions to support generation of the ISA 3.0 XXBR<x>
2475 vector byte reverse instructions.
2476 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
2477 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
2478 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
2479 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
2480 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
2481 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
2482 (P9V_BUILTIN_VEC_REVB): Likewise.
2483 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
2484 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
2485 (p9_xxbrq_v16qi): Likewise.
2486 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
2487 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
2488 (p9_xxbrh_v8hi): Likewise.
2489 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
2490 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
2491 vec_revb built-in functions.
2493 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
2495 PR rtl-optimization/78952
2496 * config/i386/i386.md (any_extract): New code iterator.
2497 (*insvqi_2): Use any_extract for source operand.
2498 (*insvqi_3): Use any_shiftrt for source operand.
2500 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
2502 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
2504 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
2506 2017-01-18 Matthias Klose <doko@ubuntu.com>
2508 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
2510 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2512 * config/rs6000/altivec.h (vec_bperm): Change #define.
2513 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
2514 (altivec_vbpermq2): New define_insn.
2515 (altivec_vbpermd): Likewise.
2516 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
2518 (VBPERMD): Likewise.
2519 (VBPERM): New polymorphic function interface.
2520 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
2521 Add entries for P9V_BUILTIN_VEC_VBPERM.
2522 * doc/extend.texi: Add interfaces for vec_bperm.
2524 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2526 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
2527 first letter of error messages.
2528 (s390_resolve_overloaded_builtin): Likewise.
2529 * config/s390/s390.c (s390_expand_builtin): Likewise.
2530 (s390_invalid_arg_for_unprototyped_fn): Likewise.
2531 (s390_valid_target_attribute_inner_p): Likewise.
2532 * config/s390/s390.md ("tabort"): Likewise.
2534 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
2536 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
2537 (ISA_AVOID_DIV_HILO): New macro.
2538 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
2539 (ISA_HAS_DDIV): Likewise.
2541 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2543 * doc/invoke.texi (fabi-version): Correct number of occurrences.
2545 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2547 * doc/invoke.texi (fabi-version): Spelling fix.
2549 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2552 * doc/invoke.texi (fabi-version): Mention mangling fix for
2555 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2558 * doc/invoke.texi (fabi-version): Document discriminator mangling.
2560 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
2563 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
2564 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
2566 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
2568 (stack_protect_test): Ditto.
2569 * config/rs6000/rs6000.opt (mstack-protector-guard=,
2570 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
2572 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
2573 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
2574 -mstack-protector-guard-offset=.
2575 (RS/6000 and PowerPC Options): Ditto.
2577 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2579 * config/i386/i386.h (MASK_CLASS_P): New define.
2580 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
2581 there are no registers from different register sets also when
2582 mask registers are used. Update function comment.
2583 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
2584 to (*k/*r) and (*k/*km) alternatives.
2586 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
2588 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
2589 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
2590 (EH_RETURN_HANDLER_RTX): New define.
2591 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
2592 Force frame pointer in EH return functions.
2593 (aarch64_expand_epilogue): Add barrier for eh_return.
2594 (aarch64_final_eh_return_addr): Remove.
2595 (aarch64_eh_return_handler_rtx): New function.
2596 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
2598 (aarch64_eh_return_handler_rtx): New prototype.
2600 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2602 * config/rs6000/altivec.h (vec_rlmi): New #define.
2603 (vec_vrlnm): Likewise.
2604 (vec_rlnm): Likewise.
2605 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
2606 (UNSPEC_VRLNM): Likewise.
2607 (VIlong): New mode iterator.
2608 (altivec_vrl<VI_char>mi): New define_insn.
2609 (altivec_vrl<VI_char>nm): Likewise.
2610 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
2613 (RLNM): New polymorphic function entry.
2614 (VRLWMI): New monomorphic function entry.
2616 (RLMI): New polymorphic function entry.
2617 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
2618 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
2619 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
2622 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2625 * dwarf2out.c (field_byte_offset): Restore the
2626 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
2627 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
2628 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
2631 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2634 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
2636 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2639 * dwarf2out.c (add_data_member_location_attribute): For constant
2640 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
2641 instead of DW_AT_data_member_location, DW_AT_bit_offset and
2642 DW_AT_byte_size attributes.
2644 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2646 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
2647 after forcing to constant memory when the code model is medium.
2649 2017-01-17 Julia Koval <julia.koval@intel.com>
2652 * config/i386/avx512fintrin.h
2653 (_mm512_i32gather_ps): Change __addr type to void const*.
2654 (_mm512_mask_i32gather_ps): Ditto.
2655 (_mm512_i32gather_pd): Ditto.
2656 (_mm512_mask_i32gather_pd): Ditto.
2657 (_mm512_i64gather_ps): Ditto.
2658 (_mm512_mask_i64gather_ps): Ditto.
2659 (_mm512_i64gather_pd): Ditto.
2660 (_mm512_mask_i64gather_pd): Ditto.
2661 (_mm512_i32gather_epi32): Ditto.
2662 (_mm512_mask_i32gather_epi32): Ditto.
2663 (_mm512_i32gather_epi64): Ditto.
2664 (_mm512_mask_i32gather_epi64): Ditto.
2665 (_mm512_i64gather_epi32): Ditto.
2666 (_mm512_mask_i64gather_epi32): Ditto.
2667 (_mm512_i64gather_epi64): Ditto.
2668 (_mm512_mask_i64gather_epi64): Ditto.
2669 (_mm512_i32scatter_ps): Change __addr type to void*.
2670 (_mm512_mask_i32scatter_ps): Ditto.
2671 (_mm512_i32scatter_pd): Ditto.
2672 (_mm512_mask_i32scatter_pd): Ditto.
2673 (_mm512_i64scatter_ps): Ditto.
2674 (_mm512_mask_i64scatter_ps): Ditto.
2675 (_mm512_i64scatter_pd): Ditto.
2676 (_mm512_mask_i64scatter_pd): Ditto.
2677 (_mm512_i32scatter_epi32): Ditto.
2678 (_mm512_mask_i32scatter_epi32): Ditto.
2679 (_mm512_i32scatter_epi64): Ditto.
2680 (_mm512_mask_i32scatter_epi64): Ditto.
2681 (_mm512_i64scatter_epi32): Ditto.
2682 (_mm512_mask_i64scatter_epi32): Ditto.
2683 (_mm512_i64scatter_epi64): Ditto.
2684 (_mm512_mask_i64scatter_epi64): Ditto.
2685 * config/i386/avx512pfintrin.h
2686 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
2687 (_mm512_mask_prefetch_i32gather_ps): Ditto.
2688 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2689 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2690 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
2691 (_mm512_prefetch_i32scatter_ps): Ditto.
2692 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2693 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2694 (_mm512_prefetch_i64scatter_pd): Ditto.
2695 (_mm512_prefetch_i64scatter_ps): Ditto.
2696 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2697 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2698 * config/i386/avx512vlintrin.h
2699 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
2700 (_mm_mmask_i32gather_ps): Ditto.
2701 (_mm256_mmask_i32gather_pd): Ditto.
2702 (_mm_mmask_i32gather_pd): Ditto.
2703 (_mm256_mmask_i64gather_ps): Ditto.
2704 (_mm_mmask_i64gather_ps): Ditto.
2705 (_mm256_mmask_i64gather_pd): Ditto.
2706 (_mm_mmask_i64gather_pd): Ditto.
2707 (_mm256_mmask_i32gather_epi32): Ditto.
2708 (_mm_mmask_i32gather_epi32): Ditto.
2709 (_mm256_mmask_i32gather_epi64): Ditto.
2710 (_mm_mmask_i32gather_epi64): Ditto.
2711 (_mm256_mmask_i64gather_epi32): Ditto.
2712 (_mm_mmask_i64gather_epi32): Ditto.
2713 (_mm256_mmask_i64gather_epi64): Ditto.
2714 (_mm_mmask_i64gather_epi64): Ditto.
2715 (_mm256_i32scatter_ps): Change __addr type to void*.
2716 (_mm256_mask_i32scatter_ps): Ditto.
2717 (_mm_i32scatter_ps): Ditto.
2718 (_mm_mask_i32scatter_ps): Ditto.
2719 (_mm256_i32scatter_pd): Ditto.
2720 (_mm256_mask_i32scatter_pd): Ditto.
2721 (_mm_i32scatter_pd): Ditto.
2722 (_mm_mask_i32scatter_pd): Ditto.
2723 (_mm256_i64scatter_ps): Ditto.
2724 (_mm256_mask_i64scatter_ps): Ditto.
2725 (_mm_i64scatter_ps): Ditto.
2726 (_mm_mask_i64scatter_ps): Ditto.
2727 (_mm256_i64scatter_pd): Ditto.
2728 (_mm256_mask_i64scatter_pd): Ditto.
2729 (_mm_i64scatter_pd): Ditto.
2730 (_mm_mask_i64scatter_pd): Ditto.
2731 (_mm256_i32scatter_epi32): Ditto.
2732 (_mm256_mask_i32scatter_epi32): Ditto.
2733 (_mm_i32scatter_epi32): Ditto.
2734 (_mm_mask_i32scatter_epi32): Ditto.
2735 (_mm256_i32scatter_epi64): Ditto.
2736 (_mm256_mask_i32scatter_epi64): Ditto.
2737 (_mm_i32scatter_epi64): Ditto.
2738 (_mm_mask_i32scatter_epi64): Ditto.
2739 (_mm256_i64scatter_epi32): Ditto.
2740 (_mm256_mask_i64scatter_epi32): Ditto.
2741 (_mm_i64scatter_epi32): Ditto.
2742 (_mm_mask_i64scatter_epi32): Ditto.
2743 (_mm256_i64scatter_epi64): Ditto.
2744 (_mm256_mask_i64scatter_epi64): Ditto.
2745 (_mm_i64scatter_epi64): Ditto.
2746 (_mm_mask_i64scatter_epi64): Ditto.
2747 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
2748 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
2749 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
2750 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
2751 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
2752 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
2753 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
2754 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
2755 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
2756 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
2757 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
2758 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
2759 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
2760 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
2761 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
2762 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
2763 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
2764 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
2765 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
2766 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
2767 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
2768 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
2769 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
2770 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
2771 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
2772 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
2773 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
2774 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
2775 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
2776 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
2777 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
2778 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
2779 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
2780 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
2781 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
2782 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
2783 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
2784 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
2785 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
2786 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
2787 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
2788 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
2789 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
2790 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
2791 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
2792 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
2793 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
2794 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
2795 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
2796 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
2797 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
2798 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
2799 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
2800 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
2801 definitions accordingly.
2803 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
2804 Kuan-Lin Chen <kuanlinchentw@gmail.com>
2807 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
2810 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
2813 * ira-conflicts.c (ira_build_conflicts): Update total conflict
2814 hard regs for inner regno.
2816 2017-01-17 Martin Liska <mliska@suse.cz>
2819 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
2820 assumption and add comment.
2822 2017-01-17 Nathan Sidwell <nathan@acm.org>
2824 * ipa-visibility.c (localize_node): New function, broken out of ...
2825 (function_and_variable_visibility): ... here. Call it.
2827 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
2830 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
2831 correctly set frequency of oudgoing edge.
2832 (duplicate_thread_path): Fix profile updating.
2834 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2837 * configure.ac: Add GCC_BASE_VER.
2838 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
2839 version from BASE-VER file.
2840 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
2841 (gcc.o): Depend on $(BASEVER).
2842 * common.opt (dumpfullversion): New option.
2843 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
2844 * doc/invoke.texi: Document -dumpfullversion.
2845 * doc/install.texi: Document --with-gcc-major-version-only.
2846 * configure: Regenerated.
2848 2017-01-17 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/71433
2851 * tree-vrp.c (register_new_assert_for): Merge same asserts
2852 on all incoming edges.
2853 (process_assert_insertions_for): Handle insertions at the
2856 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
2858 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
2859 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
2861 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
2864 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
2867 2017-01-17 Alan Modra <amodra@gmail.com>
2870 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
2871 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
2872 symbolic stack limit when pic.
2874 2017-01-16 Martin Sebor <msebor@redhat.com>
2876 PR tree-optimization/78608
2877 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
2879 2017-01-16 Jeff Law <law@redhat.com>
2882 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
2883 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
2884 for several include directories that may be relative to sysroot.
2885 * config/i386/x-mingw32 (gplus_includedir): Define.
2886 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
2887 (native_system_includedir): Likewise.
2888 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
2889 override if TARGET_SYSTEM_ROOT is defined.
2890 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
2892 PR tree-optimization/79090
2893 PR tree-optimization/33562
2894 PR tree-optimization/61912
2895 PR tree-optimization/77485
2896 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
2897 and computed trims into the dump file.
2899 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2901 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
2903 2017-01-16 Jakub Jelinek <jakub@redhat.com>
2906 * gimplify.c (gimplify_init_constructor): If want_value and
2907 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
2911 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
2912 sequence. Formatting fixes.
2913 (doloop_optimize): Formatting fixes.
2916 * gcc.c (debug_level_greater_than_spec_func): New function.
2917 (static_spec_functions): Add debug-level-gt spec function.
2918 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
2920 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2921 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2922 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
2923 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
2924 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
2925 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
2927 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
2929 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
2930 QImode fixups to general and mask registers only.
2932 2017-01-16 Carl Love <cel@us.ibm.com>
2934 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
2935 for built-in functions
2936 vector signed char vec_nabs (vector signed char)
2937 vector signed short vec_nabs (vector signed short)
2938 vector signed int vec_nabs (vector signed int)
2939 vector signed long long vec_nabs (vector signed long long)
2940 vector float vec_nabs (vector float)
2941 vector double vec_nabs (vector double)
2942 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
2944 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
2945 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
2946 * doc/extend.texi: Update the documentation file for the new built-in
2949 2017-01-16 Martin Sebor <msebor@redhat.com>
2951 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
2954 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2956 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
2957 UNSPEC_VSX__XXSPLTD to require special splat handling.
2959 2017-01-16 David Malcolm <dmalcolm@redhat.com>
2962 * system.h: Poison strndup.
2964 2017-01-16 Alan Modra <amodra@gmail.com>
2967 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
2970 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
2972 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
2974 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
2976 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
2977 call recog here. Assert that INSN_CODE (insn) is non-negative.
2979 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2982 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
2984 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
2985 in the currently scheduled RTL fragment.
2987 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2989 PR rtl-optimization/78751
2990 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
2993 2017-01-14 Jeff Law <law@redhat.com>
2995 PR tree-optimization/79090
2996 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
2997 variable length stores.
2998 (compute_trims): Delete dead assignment to *trim_tail.
2999 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
3002 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
3004 PR rtl-optimization/78626
3005 PR rtl-optimization/78727
3006 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
3007 of a block, and split such blocks after everything else is finished.
3009 2017-01-14 Alan Modra <amodra@gmail.com>
3012 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
3013 target legitimate_combined_insn.
3014 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
3015 (rs6000_legitimate_combined_insn): New function.
3016 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
3018 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
3019 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
3020 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
3022 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
3024 * doc/frontends.texi (G++ and GCC): Remove references to Java.
3026 2017-01-13 Jeff Law <law@redhat.com>
3028 PR tree-optimization/33562
3029 PR tree-optimization/61912
3030 PR tree-optimization/77485
3031 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
3033 (delete_dead_assignment): Likewise.
3034 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
3035 statement to delete_dead_call and delete_dead_assignment.
3037 2017-01-13 David Malcolm <dmalcolm@redhat.com>
3040 * substring-locations.c (format_warning_va): Strengthen case 1 so
3041 that both endpoints of the substring must be within the format
3042 range for just the substring to be printed.
3044 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
3046 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
3047 * config/i386/i386.c (ix86_target_string): Add missing options
3048 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
3049 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
3050 flags_other and ix86_target_other to flags2_other. Display unknown
3052 (ix86_valid_target_attribute_inner_p): Add missing options and
3053 reorder options by implied ISAs, as in ix86_target_string.
3055 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3057 * hash-table.h (hash_table::too_empty_p): New function.
3058 (hash_table::expand): Use it.
3059 (hash_table::traverse): Likewise.
3060 (hash_table::empty_slot): Use sizeof (value_type) instead of
3061 sizeof (PTR) to convert bytes to elements. Shrink the table
3062 if the current size is excessive for the current number of
3065 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3067 * ira-costs.c (record_reg_classes): Break from the inner loop
3068 early once alt_fail is known to be true. Update outer loop
3069 handling accordingly.
3071 2017-01-13 Jeff Law <law@redhat.com>
3073 * tree-ssa-dse.c (decrement_count): New function.
3074 (increment_start_addr, maybe_trim_memstar_call): Likewise.
3075 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
3076 when we know the partially dead statement is a mem* function.
3078 PR tree-optimization/61912
3079 PR tree-optimization/77485
3080 * tree-ssa-dse.c: Include expr.h.
3081 (maybe_trim_constructor_store): New function.
3082 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
3084 PR tree-optimization/33562
3085 PR tree-optimization/61912
3086 PR tree-optimization/77485
3087 * doc/invoke.texi: Document new dse-max-object-size param.
3088 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
3089 * tree-ssa-dse.c: Include params.h.
3090 (dse_store_status): New enum.
3091 (initialize_ao_ref_for_dse): New, partially extracted from
3093 (valid_ao_ref_for_dse, normalize_ref): New.
3094 (setup_live_bytes_from_ref, compute_trims): Likewise.
3095 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
3096 (maybe_trim_partially_dead_store): Likewise.
3097 (maybe_trim_complex_store): Likewise.
3098 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
3099 Track what bytes live from the original store. Return tri-state
3100 for dead, partially dead or live.
3101 (dse_dom_walker): Add constructor, destructor and new private members.
3102 (delete_dead_call, delete_dead_assignment): New extracted from
3104 (dse_optimize_stmt): Make a member of dse_dom_walker.
3105 Use initialize_ao_ref_for_dse.
3107 PR tree-optimization/33562
3108 PR tree-optimization/61912
3109 PR tree-optimization/77485
3110 * sbitmap.h (bitmap_count_bits): Prototype.
3111 (bitmap_clear_range, bitmap_set_range): Likewise.
3112 * sbitmap.c (bitmap_clear_range): New function.
3113 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
3115 2017-01-13 Martin Liska <mliska@suse.cz>
3118 * function.c (set_cfun): Add new argument force.
3119 * function.h (set_cfun): Likewise.
3120 * ipa-inline-transform.c (inline_call): Use the function when
3121 strict alising from is dropped for function we inline to.
3123 2017-01-13 Richard Biener <rguenther@suse.de>
3125 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
3126 for dumping GIMPLE INTEGER_CSTs.
3128 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3130 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
3131 to 201112L since C++17.
3133 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
3136 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
3137 if -fsanitize=kernel-address is present.
3139 2017-01-13 Richard Biener <rguenther@suse.de>
3141 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
3142 as _Literal ( type ) number in case usual suffixes do not
3143 preserve all information.
3145 2017-01-13 Richard Biener <rguenther@suse.de>
3147 PR tree-optimization/77283
3148 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
3149 and ssa-iterators.h.
3150 (is_feasible_trace): Implement a cost model based on joiner
3153 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3156 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
3157 char or short to __float128/_Float128 directly.
3159 2017-01-12 Martin Sebor <msebor@redhat.com>
3161 to -Wformat-overflow.
3162 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
3163 (min_bytes_remaining): Same.
3164 (get_string_length): Same.
3165 (format_string): Same.
3166 (format_directive): Same.
3168 (pass_sprintf_length::handle_gimple_call): Same.
3170 2017-01-12 Jakub Jelinek <jakub@redhat.com>
3172 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
3173 info.nowrite calls with no lhs that can't throw. Return bool
3174 whether gsi_remove has been called or not.
3175 (pass_sprintf_length::handle_gimple_call): Return bool whether
3176 try_substitute_return_value called gsi_remove. Formatting fix.
3177 (pass_sprintf_length::execute): Don't use gsi_remove if
3178 handle_gimple_call returned true.
3181 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
3182 be removed due to side-effects, don't remove following barrier nor
3183 turn the successor edge into fallthru edge.
3185 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3188 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
3189 element-reversing loads and stores as not swappable.
3191 2017-01-12 Nathan Sidwell <nathan@acm.org>
3192 Nicolai Stange <nicstange@gmail.com>
3194 * combine.c (try_combine): Don't ignore result of overlap checking
3195 loop. Combine overlap & asm check into single loop.
3197 2017-01-12 Richard Biener <rguenther@suse.de>
3199 * tree-pretty-print.c (dump_generic_node): Provide -gimple
3200 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
3202 2017-01-12 Richard Biener <rguenther@suse.de>
3204 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
3205 and TS_TARGET_OPTION directly derive from TS_BASE.
3206 * tree-core.h (tree_optimization_option): Derive from tree_base.
3207 (tree_target_option): Likewise.
3209 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
3211 * config/i386/i386.c (memory_address_length): Increase len
3212 only when rip_relative_addr_p returns false.
3214 2017-01-11 Julia Koval <julia.koval@intel.com>
3216 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
3217 (OPTION_MASK_ISA_SGX_SET): New.
3218 (ix86_handle_option): Handle OPT_msgx.
3219 * config.gcc: Added sgxintrin.h.
3220 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
3221 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
3222 * config/i386/i386.c (ix86_target_string): Add -msgx.
3224 (ix86_option_override_internal): Handle new options.
3225 (ix86_valid_target_attribute_inner_p): Add sgx.
3226 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
3227 * config/i386/i386.opt: Add msgx.
3228 * config/i386/sgxintrin.h: New file.
3229 * config/i386/x86intrin.h: Add sgxintrin.h.
3231 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3234 * fold-const.c (maybe_nonzero_address): Return 1 for function
3236 (tree_single_nonzero_warnv_p): Don't handle function local objects
3240 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
3243 2017-01-11 David Malcolm <dmalcolm@redhat.com>
3246 * opts.c: Include "spellcheck.h"
3247 (struct string_fragment): New struct.
3248 (struct edit_distance_traits<const string_fragment &>): New
3250 (get_closest_sanitizer_option): New function.
3251 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
3253 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3255 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
3257 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
3258 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
3259 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
3260 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
3261 for initial die_offset if dwarf_split_debug_info.
3262 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
3263 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
3265 (output_skeleton_debug_sections): Formatting fix. Use
3266 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
3267 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
3269 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
3271 * config/arm/cortex-a53.md: Add bypasses for
3273 (cortex_a53_r2f): Only use for transfers.
3274 (cortex_a53_f2r): Likewise.
3275 (cortex_a53_r2f_cvt): Add reservation for conversions.
3276 (cortex_a53_f2r_cvt): Likewise.
3278 2017-01-11 Tamar Christina <tamar.christina@arm.com>
3280 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
3281 to all inlined functions, change static to extern.
3283 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
3286 * config/arm/arm.c (legitimize_pic_address): Handle reference to
3288 (arm_assemble_integer): Likewise.
3290 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
3292 * config.gcc: Use new awk script to check CPU, FPU and architecture
3293 parameters for --with-... options.
3294 * config/arm/parsecpu.awk: New file
3295 * config/arm/arm-cpus.in: New file.
3296 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
3298 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
3300 * config/arm/t-arm: Update dependency rules.
3301 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
3302 of processing .def files.
3303 * config/arm/genopt.sh: Deleted.
3304 * config/arm/gentune.sh: Deleted.
3305 * config/arm/arm-cores.def: Deleted.
3306 * config/arm/arm-arches.def: Deleted.
3307 * config/arm/arm-fpus.def: Deleted.
3308 * config/arm/arm-tune.md: Regenerated.
3309 * config/arm/arm-tables.opt: Regenerated.
3310 * config/arm/arm-cpu.h: New generated file.
3311 * config/arm/arm-cpu-data.h: New generated file.
3312 * config/arm/arm-cpu-cdata.h: New generated file.
3314 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
3317 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
3319 (input_varpool_node): Unpack dynamically_initialized bit.
3321 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
3323 PR rtl-optimization/79032
3324 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
3325 the alignment of the adjusted memory reference against that of MODE,
3326 instead of the alignment of the original memory reference.
3328 2017-01-11 Martin Jambor <mjambor@suse.cz>
3330 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
3332 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
3333 decorated functions.
3335 2017-01-11 Richard Biener <rguenther@suse.de>
3337 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
3338 set range/nonnull info for PHI results. Do not set it on
3339 stmts marked for removal.
3341 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
3343 * expr.c (store_field): In the bitfield case, fetch the return value
3344 from the registers before applying a single big-endian adjustment.
3345 Always do a final load for a BLKmode value not larger than a word.
3347 2017-01-10 David Malcolm <dmalcolm@redhat.com>
3350 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3351 that we correctly handle column numbers greater than
3352 LINE_MAP_MAX_COLUMN_NUMBER.
3354 2017-01-10 Martin Sebor <msebor@redhat.com>
3357 * gimple-ssa-sprintf.c (get_destination_size): Call
3358 {init,fini}object_sizes.
3359 * tree-object-size.c (addr_object_size): Adjust.
3360 (pass_through_call): Adjust.
3361 (pass_object_sizes::execute): Adjust.
3362 * tree-object-size.h (fini_object_sizes): Declare.
3364 2017-01-10 Martin Sebor <msebor@redhat.com>
3366 PR tree-optimization/78775
3367 * builtins.c (get_size_range): Move...
3368 * calls.c: ...to here.
3369 (alloc_max_size): Accept zero argument.
3370 (operand_signed_p): Remove.
3371 (maybe_warn_alloc_args_overflow): Call get_size_range.
3372 * calls.h (get_size_range): Declare.
3374 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
3376 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
3377 from TI's devices.csv file as of September 2016.
3378 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
3380 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
3382 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
3383 * doc/invoke.texi: Likewise.
3384 * doc/md.texi: Likewise.
3385 * doc/objc.texi: Likewise.
3387 2017-01-10 Joshua Conner <joshconner@google.com>
3389 * config/arm/fuchsia-elf.h: New file.
3390 * config/fuchsia.h: New file.
3391 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
3392 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
3394 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
3396 2016-01-10 Richard Biener <rguenther@suse.de>
3398 PR tree-optimization/79034
3399 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
3400 Propagate out degenerate PHIs in the joiner.
3402 2017-01-10 Martin Liska <mliska@suse.cz>
3404 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
3405 (sort_congruence_classes_by_decl_uid): Likewise.
3406 (sort_congruence_class_groups_by_decl_uid): Likewise.
3407 (sem_item_optimizer::merge_classes): Sort class, groups in these
3408 classes and members in the groups by DECL_UID of declarations.
3409 This would make merge operations stable.
3411 2017-01-10 Martin Liska <mliska@suse.cz>
3413 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
3414 usage of m_classes_vec.
3415 (sem_item_optimizer::~sem_item_optimizer): Likewise.
3416 (sem_item_optimizer::get_group_by_hash): Likewise.
3417 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
3418 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
3419 (sem_item_optimizer::verify_classes): Likewise.
3420 (sem_item_optimizer::process_cong_reduction): Likewise.
3421 (sem_item_optimizer::dump_cong_classes): Likewise.
3422 (sem_item_optimizer::merge_classes): Likewise.
3423 * ipa-icf.h (congruence_class_hash): Rename from
3424 congruence_class_group_hash. Remove declaration of m_classes_vec.
3426 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
3428 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
3429 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
3430 * config.gcc: Add avx512vpopcntdqintrin.h.
3431 * config/i386/avx512vpopcntdqintrin.h: New.
3432 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
3433 * config/i386/i386-builtin-types.def: Add new types.
3434 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
3435 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
3436 __builtin_ia32_vpopcountq_v8di_mask): New.
3437 * config/i386/i386-c.c (ix86_target_macros_internal): Define
3438 __AVX512VPOPCNTDQ__.
3439 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
3440 (PTA_AVX512VPOPCNTDQ): Define.
3441 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
3442 TARGET_AVX512VPOPCNTDQ_P): Define.
3443 * config/i386/i386.opt: Add mavx512vpopcntdq.
3444 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
3445 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
3447 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3450 * predict.def (PRED_CALL): Set to 67.
3452 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
3454 * expr.c (store_field): In the bitfield case, if the value comes from
3455 a function call and is of an aggregate type returned in registers, do
3456 not modify the field mode; extract the value in all cases if the mode
3457 is BLKmode and the size is not larger than a word.
3459 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
3462 * config/i386/cpuid.h: Fix undefined behavior.
3464 2017-01-04 Jeff Law <law@redhat.com>
3466 PR tree-optimization/79007
3467 PR tree-optimization/67955
3468 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
3469 conservative for pt.null when flag_non_call_exceptions is on.
3471 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3473 PR translation/79019
3474 PR translation/79020
3475 * params.def (PARAM_INLINE_MIN_SPEEDUP,
3476 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
3477 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
3479 * config/avr/avr.opt (maccumulate-args): Likewise.
3480 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
3481 * common.opt (freport-bug): Likewise.
3482 * cif-code.def (CIF_FINAL_ERROR): Likewise.
3483 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
3484 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
3485 translatable string.
3486 * config/i386/i386.c (function_value_32): Likewise.
3487 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
3488 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
3490 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
3491 * common/config/msp430/msp430-common.c (msp430_handle_option):
3493 * symtab.c (symtab_node::verify_base): Likewise.
3494 * opts.c (set_debug_level): Likewise.
3495 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
3496 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
3497 missing whitespace to translatable strings.
3498 * config/avr/avr.md (bswapsi2): Fix typo in comment.
3499 * config/sh/superh.h: Likewise.
3500 * config/i386/xopintrin.h: Likewise.
3501 * config/i386/znver1.md: Likewise.
3502 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
3503 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
3504 * double-int.h (struct double_int): Likewise.
3505 * double-int.c (div_and_round_double): Likewise.
3506 * wide-int.cc: Likewise.
3507 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
3508 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
3509 * cfgcleanup.c (crossjumps_occured): Renamed to ...
3510 (crossjumps_occurred): ... this.
3511 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
3514 PR tree-optimization/78899
3515 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
3516 returning bool return struct loop *, NULL for failure and the new
3518 (versionable_outer_loop_p): Don't version outer loop if it has
3519 dont_vectorized bit set.
3520 (tree_if_conversion): When versioning outer loop, ensure
3521 tree_if_conversion is performed also on the inner loop of the
3522 non-vectorizable outer loop copy.
3523 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
3524 LOOP_VECTORIZED in inner loop of the scalar outer loop and
3525 prevent vectorization of it.
3526 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
3527 the outer loop vectorization of the non-scalar version is attempted
3528 before vectorization of the inner loop in scalar version. If
3529 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
3530 vectorization of its inner loop.
3531 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
3532 has 2 inner loops, rename also on edges from bb whose single pred
3533 is outer_loop->header. Fix typo in function comment.
3535 2017-01-09 Martin Sebor <msebor@redhat.com>
3538 * asan.c (asan_emit_stack_protection): Increase local buffer size
3539 to avoid snprintf truncation warning.
3541 2017-01-09 Andrew Pinski <apinski@cavium.com>
3543 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
3544 to reference thunderx2t99 for the tuning structure
3545 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
3547 (thunderx2t99_extra_costs): This.
3548 * config/aarch64/aarch64-tune.md: Regenerate.
3549 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
3550 (vulcan_addrcost_table): This.
3551 (vulcan_regmove_cost): Rename to ...
3552 (thunderx2t99_regmove_cost): This.
3553 (vulcan_vector_cost): Rename to ...
3554 (thunderx2t99_vector_cost): this.
3555 (vulcan_branch_cost): Rename to ...
3556 (thunderx2t99_branch_cost): This.
3557 (vulcan_tunings): Rename to ...
3558 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
3559 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
3561 2017-01-09 Martin Jambor <mjambor@suse.cz>
3565 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
3566 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
3567 (propagate_vr_accross_jump_function): Use the above function for all
3568 value range computations for pass-through jump functions and type
3569 converasion from explicit value range values.
3570 (ipcp_propagate_stage): Do not attempt to deduce types of formal
3571 parameters from TYPE_ARG_TYPES.
3572 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
3573 (ipa_write_node_info): Stream type of the actual argument.
3574 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
3576 2017-01-09 Martin Liska <mliska@suse.cz>
3579 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
3580 (lookup_compiler): Do not show error message with have_E.
3582 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3584 PR tree-optimization/78938
3585 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
3586 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
3587 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
3588 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
3591 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3593 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
3596 2017-01-09 Richard Biener <rguenther@suse.de>
3598 PR tree-optimization/78997
3599 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
3600 name condition properly.
3602 2017-01-09 Richard Biener <rguenther@suse.de>
3605 * dwarf2out.c (is_cxx): New overload with context.
3606 (is_naming_typedef_decl): Use it.
3608 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3610 * invoke.texi (Option Summary): Correct spacing in option lists
3611 and add line breaks to fix over-long lines.
3613 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3617 * extend.texi (Common Variable Attributes): Add xref to GCC
3618 Internals manual to explain mode attribute keywords.
3620 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3623 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
3624 and Preprocessor Options.
3625 (Options for Linking): Document -pthread here....
3626 (RS/6000 and PowerPC Options): ...not here.
3627 (Solaris 2 Options): ...or here.
3628 * doc/cppopts.texi: Document -pthread.
3630 2017-01-08 Martin Sebor <msebor@redhat.com>
3633 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
3634 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
3635 New member functions.
3636 (format_directive): Used them.
3638 (pass_sprintf_length::handle_gimple_call): Same.
3639 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
3640 to avoid truncation for any argument.
3641 (extract_affine_mul): Same.
3642 * tree.c (get_file_function_name): Same.
3644 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3647 * predict.def (PRED_INDIR_CALL): Set to 86.
3649 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3651 PR preprocessor/54124
3652 * doc/cppopts.texi: Reformat -d subtable to list the full name
3653 of the options. Add cross-reference to the docs for the general
3654 compiler -d options.
3655 * doc/invoke.texi (Developer Options): Add cross-reference to the
3656 preprocessor-specific -d option documentation.
3658 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3660 PR preprocessor/13498
3661 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
3662 redudant material, and reflect new command-line options.
3663 (System Headers): Likewise.
3665 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3667 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
3668 -isystem, and -idirafter. Copy-edit.
3669 * doc/cppopts.texi: Copy-edit. Remove contradiction about
3670 default for -ftrack-macro-expansion. Delete obsolete and
3671 badly-formatted implementation details about -fdebug-cpp output.
3672 * doc/cppwarnopts.texi: Copy-edit.
3674 2017-01-07 David Malcolm <dmalcolm@redhat.com>
3677 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3678 that the transition from a max line width >= 1<<10 to narrower
3679 lines works correctly.
3681 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
3683 * doc/options.texi (PerFunction): New.
3684 * opt-functions.awk (switch_flags): Map both Optimization and
3685 PerFunction to CL_OPTIMIZATION.
3686 * opth-gen.awk: Test for PerFunction flag along with
3688 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
3689 it only when the latter is present. Skip those that don't in
3690 the hash function generator.
3691 * common.opt (fvar-tracking): Mark as PerFunction instead of
3693 (fvar-tracking-assignments): Likewise.
3694 (fvar-tracking-assignments-toggle): Likewise.
3695 (fvar-tracking-uninit): Likewise.
3697 2017-01-07 Jakub Jelinek <jakub@redhat.com>
3699 PR translation/79018
3700 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
3703 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
3706 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
3707 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
3708 TARGET_LONG_JUMP_TABLE_OFFSETS.
3709 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
3710 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
3711 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
3712 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
3713 * config/m68k/m68k.md (tablejump expander): Likewise.
3714 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
3715 TARGET_LONG_JUMP_TABLE_OFFSETS.
3716 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
3717 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
3719 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3720 David Holsgrove <david.holsgrove@xilinx.com>
3722 * common/config/microblaze/microblaze-common.c
3723 (TARGET_EXCEPT_UNWIND_INFO): Remove.
3724 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
3726 * config/microblaze/microblaze.c (microblaze_must_save_register)
3727 (microblaze_expand_epilogue, microblaze_return_addr): Handle
3729 (microblaze_eh_return): New function.
3730 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
3731 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
3732 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
3733 * config/microblaze/microblaze.md (eh_return): New pattern.
3735 2017-01-06 Jakub Jelinek <jakub@redhat.com>
3737 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
3738 GCC_DIAGNOSTIC_STRINGIFY): Define.
3740 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
3742 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3744 * config/arm/arm.md (<mcrr>): New.
3746 * config/arm/arm.c (arm_arch5te): New.
3747 (arm_option_override): Set arm_arch5te.
3748 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
3750 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
3751 (arm_mcrr_qualifiers): ... this. New.
3752 (MRRC_QUALIFIERS): Define to...
3753 (arm_mrrc_qualifiers): ... this. New.
3754 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
3756 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
3757 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
3758 (MRRCI, mrrc, MRRC): New.
3759 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
3760 VUNSPEC_MRRC2): New.
3762 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3764 * config/arm/arm.md (<mcr>): New.
3766 * config/arm/arm.c (arm_coproc_builtin_available): Add
3767 support for mcr, mrc, mcr2 and mrc2.
3768 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
3769 (arm_mcr_qualifiers): ... this. New.
3770 (MRC_QUALIFIERS): Define to ...
3771 (arm_mrc_qualifiers): ... this. New.
3772 (MCR_QUALIFIERS): Define to ...
3773 (arm_mcr_qualifiers): ... this. New.
3774 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
3776 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
3777 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
3778 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
3781 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3783 * config/arm/arm.md (*ldc): New.
3787 * config/arm/arm.c (arm_coproc_builtin_available): Add
3788 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
3789 (arm_coproc_ldc_stc_legitimate_address): New.
3790 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
3791 'qualifier_const_pointer'.
3792 (LDC_QUALIFIERS): Define to...
3793 (arm_ldc_qualifiers): ... this. New.
3794 (STC_QUALIFIERS): Define to...
3795 (arm_stc_qualifiers): ... this. New.
3796 * config/arm/arm-protos.h
3797 (arm_coproc_ldc_stc_legitimate_address): New.
3798 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
3799 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
3800 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
3801 stc2, stcl, stc2l): New.
3802 * config/arm/constraints.md (Uz): New.
3803 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
3804 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
3805 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
3806 VUNSPEC_STC2L): New.
3808 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3810 * config/arm/arm.md (<cdp>): New.
3811 * config/arm/arm.c (neon_const_bounds): Rename this ...
3812 (arm_const_bounds): ... this.
3813 (arm_coproc_builtin_available): New.
3814 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
3815 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
3816 (CDP_QUALIFIERS): Define to...
3817 (arm_cdp_qualifiers): ... this. New.
3819 (arm_expand_builtin_args): Add case for 6 arguments.
3820 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
3821 (arm_const_bounds): ... this.
3822 (arm_coproc_builtin_available): New.
3823 * config/arm/arm_acle.h (__arm_cdp): New.
3825 * config/arm/arm_acle_builtins.def (cdp): New.
3827 * config/arm/iterators.md (CDPI,CDP,cdp): New.
3828 * config/arm/neon.md: Rename all 'neon_const_bounds' to
3830 * config/arm/types.md (coproc): New.
3831 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
3832 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
3833 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
3834 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
3836 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3838 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
3839 (UBINOP_QUALIFIERS): New.
3841 (acle_builtin_data): New. Change comment.
3842 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
3843 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
3844 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
3845 arm_acle_builtins.def.
3846 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
3847 (arm_init_acle_builtins): New.
3848 (CRC32_BUILTIN): Remove.
3849 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
3850 crc32cb, crc32ch and crc32cw.
3851 (arm_init_crc32_builtins): Remove.
3852 (arm_init_builtins): Use arm_init_acle_builtins rather
3853 than arm_init_crc32_builtins.
3854 (arm_expand_acle_builtin): New.
3855 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
3856 * config/arm/arm_acle_builtins.def: New.
3858 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3860 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
3861 (arm_builtin_datum): ... this.
3862 (arm_init_neon_builtin): Rename to ...
3863 (arm_init_builtin): ... this. Add a new parameters PREFIX
3864 and USE_SIG_IN_NAME.
3865 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
3866 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
3867 'arm_builtin_datum'.
3868 (arm_init_vfp_builtins): Likewise.
3869 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
3870 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
3871 (arm_expand_neon_args): Rename to ...
3872 (arm_expand_builtin_args): ... this. Rename builtin_arg
3873 enum values and differentiate between ARG_BUILTIN_MEMORY
3874 and ARG_BUILTIN_NEON_MEMORY.
3875 (arm_expand_neon_builtin_1): Rename to ...
3876 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
3877 values, arm_expand_builtin_args and add bool parameter NEON.
3878 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
3879 (arm_expand_vfp_builtin): Likewise.
3880 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
3882 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3885 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
3886 * predict.c (tree_estimate_probability_bb): Reverse direction of
3887 polymorphic call predictor.
3889 2017-01-06 David Malcolm <dmalcolm@redhat.com>
3891 * passes.c (execute_one_pass): Split out pass-skipping logic into...
3892 (determine_pass_name_match): ...this new function and...
3893 (should_skip_pass_p): ...this new function.
3895 2017-01-06 Nathan Sidwell <nathan@acm.org>
3897 * ipa-visibility.c (function_and_variable_visibility): Reformat
3898 comments and long lines. Remove extrneous if.
3899 * symtab.c (symtab_node::make_decl_local): Fix code format.
3900 (symtab_node::set_section_for_node): Fix comment typo.
3902 2017-01-06 Martin Liska <mliska@suse.cz>
3905 * lra-constraints.c: Rename invariant to lra_invariant.
3906 * predict.c (set_even_probabilities): Initialize e to NULL.
3908 2017-01-05 Martin Sebor <msebor@redhat.com>
3910 PR tree-optimization/78910
3911 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
3912 (format_integer): Correct off-by-one error in the handling
3913 of precision with negative numbers in signed conversions..
3915 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
3917 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
3919 2017-01-05 Jakub Jelinek <jakub@redhat.com>
3921 PR tree-optimization/71016
3922 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
3923 factor_out_conditional_conversion. Formatting fix.
3924 (factor_out_conditional_conversion): Add cond_stmt argument.
3925 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
3926 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
3929 2017-01-05 David Malcolm <dmalcolm@redhat.com>
3931 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
3932 read-rtl-function.o, and selftest-rtl.o.
3933 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
3934 (selftest::aarch64_test_loading_full_dump): New function.
3935 (selftest::aarch64_run_selftests): New function.
3936 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
3937 selftest::aarch64_run_selftests.
3938 * config/i386/i386.c
3939 (selftest::ix86_test_loading_dump_fragment_1): New function.
3940 (selftest::ix86_test_loading_call_insn): New function.
3941 (selftest::ix86_test_loading_full_dump): New function.
3942 (selftest::ix86_test_loading_unspec): New function.
3943 (selftest::ix86_run_selftests): Call the new functions.
3944 * emit-rtl.c (maybe_set_max_label_num): New function.
3945 * emit-rtl.h (maybe_set_max_label_num): New decl.
3946 * function.c (instantiate_decls): Guard call to
3947 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
3948 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
3950 * gensupport.c (gen_reader::gen_reader): Pass "false"
3951 for new "compact" param of rtx_reader.
3952 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
3953 rather than an empty string for NULL strings.
3954 * read-md.c: Potentially include config.h rather than bconfig.h.
3955 Wrap include of errors.h with #ifdef GENERATOR_FILE.
3956 (have_error): New global, copied from errors.c.
3957 (md_reader::read_name): Rename to...
3958 (md_reader::read_name_1): ...this, adding "out_loc" param,
3959 and converting "missing name or number" to returning false, rather
3961 (md_reader::read_name): Reimplement in terms of read_name_1.
3962 (md_reader::read_name_or_nil): New function.
3963 (md_reader::read_string): Handle "(nil)" by returning NULL.
3964 (md_reader::md_reader): Add new param "compact".
3965 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
3966 (md_reader::read_file): New method.
3967 * read-md.h (md_reader::md_reader): Add new param "compact".
3968 (md_reader::read_file): New method.
3969 (md_reader::is_compact): New accessor.
3970 (md_reader::read_name): Convert return type from void to file_location.
3971 (md_reader::read_name_or_nil): New decl.
3972 (md_reader::read_name_1): New decl.
3973 (md_reader::m_compact): New field.
3974 (noop_reader::noop_reader): Pass "false" for new "compact" param
3976 (rtx_reader::rtx_reader): Add new "compact" param.
3977 (rtx_reader::read_rtx_operand): Make virtual and convert return
3978 type from void to rtx.
3979 (rtx_reader::read_until): New decl.
3980 (rtx_reader::handle_any_trailing_information): New virtual function.
3981 (rtx_reader::postprocess): New virtual function.
3982 (rtx_reader::finalize_string): New virtual function.
3983 (rtx_reader::m_in_call_function_usage): New field.
3984 (rtx_reader::m_reuse_rtx_by_id): New field.
3985 * read-rtl-function.c: New file.
3986 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
3987 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
3988 (selftest::verify_three_block_rtl_cfg): New decl.
3989 * read-rtl-function.h: New file.
3990 * read-rtl.c: Potentially include config.h rather than bconfig.h.
3991 For host, include function.h, memmodel.h, and emit-rtl.h.
3992 (one_time_initialization): New function.
3993 (struct compact_insn_name): New struct.
3994 (compact_insn_names): New array.
3995 (find_code): Handle insn codes in compact dumps.
3996 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
3997 (bind_subst_iter_and_attr): Likewise.
3998 (add_condition_to_string): Likewise.
3999 (add_condition_to_rtx): Likewise.
4000 (apply_attribute_uses): Likewise.
4001 (add_current_iterators): Likewise.
4002 (apply_iterators): Likewise.
4003 (initialize_iterators): Guard usage of apply_subst_iterator with
4004 #ifdef GENERATOR_FILE.
4005 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
4006 (md_reader::read_mapping): Likewise.
4007 (add_define_attr_for_define_subst): Likewise.
4008 (add_define_subst_attr): Likewise.
4009 (read_subst_mapping): Likewise.
4010 (check_code_iterator): Likewise.
4011 (rtx_reader::read_rtx): Likewise. Move one-time initialization
4013 (one_time_initialization): New function.
4014 (rtx_reader::read_until): New method.
4015 (read_flags): New function.
4016 (parse_reg_note_name): New function.
4017 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
4018 Handle reuse_rtx ids.
4019 Wrap iterator lookup within #ifdef GENERATOR_FILE.
4020 Add parsing support for RTL dumps, mirroring the special-cases in
4021 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
4022 values, and calling handle_any_trailing_information.
4023 (rtx_reader::read_rtx_operand): Convert return type from void
4024 to rtx, returning return_rtx. Handle case 'e'. Call
4025 finalize_string on XSTR and XTMPL fields.
4026 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
4027 "(nil)" values were omitted. Call the postprocess vfunc on the
4029 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
4030 class ctor. Initialize m_in_call_function_usage. Call
4031 one_time_initialization.
4032 * rtl-tests.c (selftest::test_uncond_jump): Call
4033 set_new_first_and_last_insn.
4034 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
4035 * selftest-rtl.c: New file.
4036 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
4037 (selftest::get_insn_by_uid): New decl.
4038 * selftest-run-tests.c (selftest::run_tests): Call
4039 read_rtl_function_c_tests.
4040 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
4041 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
4044 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
4046 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
4047 operands in a special way. Assert that pos+len <= mode precision.
4049 2017-01-05 Jakub Jelinek <jakub@redhat.com>
4051 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
4052 3 argument Alias with unlimited for the negative form.
4053 (fno-vect-cost-model): Removed.
4055 2017-01-05 Martin Liska <mliska@suse.cz>
4057 * hsa-gen.c (gen_hsa_divmod): New function.
4058 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
4060 2017-01-05 Martin Liska <mliska@suse.cz>
4063 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
4066 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4068 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
4069 small constant length operands.
4071 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4073 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
4074 between loop iterations.
4076 2017-01-05 Martin Liska <mliska@suse.cz>
4079 * gimplify.c (gimplify_decl_expr): Compare to
4080 asan_poisoned_variables instread of checking flags.
4081 (gimplify_target_expr): Likewise.
4082 (gimplify_expr): Likewise.
4083 (gimplify_function_tree): Conditionally initialize
4084 asan_poisoned_variables.
4086 2017-01-04 Jeff Law <law@redhat.com>
4088 PR tree-optimizatin/78812
4089 * rtl.h (contains_mem_rtx_p): Prototype.
4090 * ifcvt.c (containts_mem_rtx_p): Move from here to...
4091 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
4092 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
4093 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
4094 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
4096 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4098 * input.c (assert_char_at_range): Default-initialize actual_range.
4100 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4102 * df-scan.c (df_ref_create_structure): Make regno unsigned,
4103 to match the caller.
4105 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4107 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
4108 insns after final jump in test to emit dummy move.
4110 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4112 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
4113 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
4115 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4117 * multiple_target.c (create_dispatcher_calls): Init e_next.
4118 * tree-ssa-loop-split.c (split_loop): Init border.
4119 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
4122 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
4127 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
4128 (altivec_register_operand): Do not return true if the operand
4129 contains a SUBREG mixing SImode and SFmode.
4130 (vsx_register_operand): Likewise.
4131 (vsx_reg_sfsubreg_ok): New predicate.
4132 (vfloat_operand): Do not return true if the operand contains a
4133 SUBREG mixing SImode and SFmode.
4134 (vint_operand): Likewise.
4135 (vlogical_operand): Likewise.
4136 (gpc_reg_operand): Likewise.
4137 (int_reg_operand): Likewise.
4138 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
4139 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
4140 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
4142 (rs6000_emit_move_si_sf_subreg): New helper function.
4143 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
4144 fixup SUBREGs involving SImode and SFmode.
4145 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
4146 numbers for the new peephole2 optimization.
4147 (peephole2 for SFmode unions): New peephole2 to optimize cases in
4148 the GLIBC math library that do AND/IOR/XOR operations on single
4149 precision floating point.
4150 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
4151 target macros to say whether we need to avoid SUBREGs mixing
4153 (TARGET_ALLOW_SF_SUBREG): Likewise.
4154 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
4155 (UNSPEC_SI_FROM_SF): Likewise.
4156 (iorxor): Change spacing.
4157 (and_ior_xor): New iterator for AND, IOR, and XOR.
4158 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
4159 (movdi_from_sf_zero_ext): Likewise.
4160 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
4161 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
4162 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
4163 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
4164 (fms<mode>4): Likewise.
4165 (fnma<mode>4): Likewise.
4166 (fnms<mode>4): Likewise.
4167 (nfma<mode>4): Likewise.
4168 (nfms<mode>4): Likewise.
4170 2017-01-04 Marek Polacek <polacek@redhat.com>
4173 * doc/invoke.texi: Document -Wpointer-compare.
4175 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4177 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
4180 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
4181 descriptions for -gdwarf-5 and emit them as uleb128 instead of
4184 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
4187 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
4188 documentation of the powerpc_popcntb_ok attribute.
4189 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4190 code to issue warning messages if a requested CPU configuration is
4191 not supported by the binary (assembler and loader) toolchain.
4192 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
4193 made to define a built-in function that has been disabled.
4194 (paired_init_builtins): Add assertion to prevent ICE if attempt is
4195 made to define a built-in function that has been disabled.
4196 (altivec_init_builtins): Add comment explaining why definition
4197 of the DST built-in functions is not preceded by an assertion
4198 check. Add assertions to prevent ICE if attempts are made to
4199 define an altivec predicate or an abs* built-in function that has
4201 (htm_init_builtins): Add comment explaining why definition of the
4202 htm built-in functions is not preceded by an assertion check.
4204 2017-01-04 Jeff Law <law@redhat.com>
4206 PR tree-optimizatin/67955
4207 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
4208 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
4209 the points-to solution does not include pt_null. Use DECL_PT_UID
4212 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
4214 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
4215 Use gen_int_mode instead of gen_lopwart for const_int operands.
4217 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4219 PR tree-optimization/71563
4220 * match.pd: Simplify X << Y into X if Y is known to be 0 or
4221 out of range value - has low bits known to be zero.
4223 2017-01-04 Alan Modra <amodra@gmail.com>
4225 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
4226 * configure: Regenerate.
4227 * config.in: Regenerate.
4229 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4232 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
4233 a substring of the message, but strcmp with the whole message. Ifdef
4234 ENABLE_NLS, translate the message first using dgettext.
4236 2017-01-03 Jeff Law <law@redhat.com>
4238 PR tree-optimizatin/78856
4239 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
4240 (mark_threaded_blocks): Remove code to truncate thread paths that
4241 cross multiple loop headers. Instead invalidate the cached loop
4242 iteration information and handle case of a thread path walking
4243 into an irreducible region.
4245 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4248 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
4249 assertions. Add support for doing the signbit if the IEEE 128-bit
4250 floating point value is in a GPR.
4251 * config/rs6000/rs6000.md (Fsignbit): Delete.
4252 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
4253 Update the length attribute if the value is in a GPR.
4254 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
4255 the sign or zero extension instruction, since the value is always 0/1.
4256 (signbit<mode>2_dm2): Delete using <Fsignbit>.
4259 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
4260 extracting SImode to a GPR register so that we can generate a
4261 store, limit the vector to be in a traditional Altivec register
4262 for the vextuwrx instruction.
4264 2017-01-03 Ian Lance Taylor <iant@google.com>
4266 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
4268 2017-01-03 Martin Sebor <msebor@redhat.com>
4270 PR tree-optimization/78696
4271 * gimple-ssa-sprintf.c (format_floating): Correct handling of
4272 precision. Use MPFR for %f for greater fidelity. Correct handling
4274 (pass_sprintf_length::compute_format_length): Set width and precision
4275 specified by asrerisk to void_node for vararg functions.
4276 (try_substitute_return_value): Adjust dump output.
4278 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
4280 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
4282 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
4284 * doc/invoke.texi (SPARC options): Document -mlra as the default.
4285 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
4286 -mlra/-mno-lra was passed to the compiler.
4288 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
4290 PR rtl-optimization/65618
4291 * emit-rtl.c (try_split): Move initialization of "before" and
4292 "after" to just before the call to emit_insn_after_setloc.
4294 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
4296 * doc/md.texi (Standard Names): Remove reference to Java frontend.
4298 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
4300 * dwarf2out.c (gen_enumeration_type_die): When
4301 -gno-strict-dwarf, add a DW_AT_encoding attribute.
4303 2017-01-03 Jakub Jelinek <jakub@redhat.com>
4305 PR tree-optimization/78965
4306 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
4307 Change first argument from const call_info & to call_info &. For %n
4308 set info.nowrite to false.
4311 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
4312 possibly throwing calls.
4314 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
4315 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
4316 and fns handling, rather than in a separate case SSA_NAME.
4318 2017-01-02 Jeff Law <law@redhat.com>
4320 * config/darwin-driver.c (darwin_driver_init): Const-correctness
4321 fixes for first_period and second_period variables.
4323 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
4326 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
4327 (*insvqi_1): New insn pattern.
4328 (*insvqi_1_mem_rex64): Ditto.
4330 (*insvqi_3): Rename from *insvqi.
4332 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
4334 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
4336 * doc/cfg.texi (Edges): Remove reference to Java.
4337 (Maintaining the CFG): Ditto.
4339 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4342 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
4343 transparent aliases.
4345 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4348 * predict.def (PRED_CALL): Update hitrate.
4349 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
4350 * predict.c (tree_estimate_probability_bb): Split CALL predictor
4351 into direct/indirect/polymorphic variants.
4353 2017-01-01 Jakub Jelinek <jakub@redhat.com>
4355 Update copyright years.
4357 * gcc.c (process_command): Update copyright notice dates.
4358 * gcov-dump.c (print_version): Ditto.
4359 * gcov.c (print_version): Ditto.
4360 * gcov-tool.c (print_version): Ditto.
4361 * gengtype.c (create_file): Ditto.
4362 * doc/cpp.texi: Bump @copying's copyright year.
4363 * doc/cppinternals.texi: Ditto.
4364 * doc/gcc.texi: Ditto.
4365 * doc/gccint.texi: Ditto.
4366 * doc/gcov.texi: Ditto.
4367 * doc/install.texi: Ditto.
4368 * doc/invoke.texi: Ditto.
4370 Copyright (C) 2017 Free Software Foundation, Inc.
4372 Copying and distribution of this file, with or without modification,
4373 are permitted in any medium without royalty provided the copyright
4374 notice and this notice are preserved.