1 2017-02-23 Jeff Law <law@redhat.com>
3 PR tree-optimization/79578
4 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
5 in call to operand_equal_p.
7 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
10 * config/i386/cpuid.h: Fix another undefined behavior.
12 2017-02-23 Richard Biener <rguenther@suse.de>
14 PR tree-optimization/79683
15 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
16 vector types for data-refs.
18 2017-02-23 Martin Liska <mliska@suse.cz>
20 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25 * internal-fn.c (get_range_pos_neg): Moved to ...
26 * tree.c (get_range_pos_neg): ... here. No longer static.
27 * tree.h (get_range_pos_neg): New prototype.
28 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29 are known to be in between 0 and signed maximum inclusive, try to
30 expand both unsigned and signed divmod and use the cheaper one from
33 2017-02-22 Jeff Law <law@redhat.com>
35 PR tree-optimization/79578
36 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37 to compare base operands.
39 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
42 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
43 gpc_reg_operand instead of fpr_reg_operand.
45 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
47 * config/mips/mips.c (mips_return_in_memory): Force FP
48 vector types to be returned in memory for o32 ABI.
50 2017-02-22 Jakub Jelinek <jakub@redhat.com>
52 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
53 instead of DW_TAG_member for static data member declarations and don't
54 set no_linkage_name for static inline data members.
55 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
58 2017-02-22 Martin Liska <mliska@suse.cz>
60 * doc/invoke.texi: Replace inequality signs with square brackets
63 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
65 PR tree-optimization/68644
66 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
68 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
71 * lra-constraints.c (simplify_operand_subreg): Handle
72 WORD_REGISTER_OPERATIONS targets.
74 2017-02-22 Jakub Jelinek <jakub@redhat.com>
77 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
78 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
79 elimination by swapping fld*.
81 2017-02-22 Richard Biener <rguenther@suse.de>
83 PR tree-optimization/79673
84 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
85 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
86 irrelevant address-space qualifiers and avoiding a
87 ADDR_SPACE_CONVERT_EXPR from fold_convert.
89 2017-02-22 Richard Biener <rguenther@suse.de>
91 PR tree-optimization/79666
92 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
93 to not symbolically negate if that may introduce undefined
96 2017-02-22 Martin Liska <mliska@suse.cz>
99 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
100 * data-streamer-out.c (streamer_write_gcov_count_stream):
102 * value-prof.c (stream_out_histogram_value): Make assert more
103 precise based on type of counter.
105 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
108 * config/i386/i386.md (standard_x87sse_constant_load splitter):
109 Use nonimmediate_operand instead of memory_operand for operand 1.
110 (float-extend standard_x87sse_constant_load splitter): Ditto.
112 2017-02-21 Jeff Law <law@redhat.com>
114 PR tree-optimization/79621
115 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
116 blocks with edges to themselves.
118 2017-02-21 Jakub Jelinek <jakub@redhat.com>
121 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
122 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
123 Use gimple_call_builtin_p.
126 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
127 on temporarily removed DEBUG_INSNs.
129 PR tree-optimization/79649
130 * tree-loop-distribution.c (classify_partition): Give up on
131 non-generic address space loads/stores.
133 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
135 * doc/loop.texi (Loop manipulation): Remove nonexistent
136 tree_ssa_loop_version from the documentation.
137 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
139 2017-02-21 Jakub Jelinek <jakub@redhat.com>
142 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
143 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
144 * config/rs6000/rs6000.c: Include except.h.
145 (rs6000_expand_split_stack_prologue): Call
146 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
148 2017-02-21 Martin Jambor <mjambor@suse.cz>
151 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
154 2017-02-21 Martin Jambor <mjambor@suse.cz>
156 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
157 for backward compatibility only.
158 * doc/invoke.texi (Option Summary): Remove all references to
161 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
165 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
167 * lra-constraints.c (curr_insn_transform): Handle
168 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
170 2017-02-21 Martin Liska <mliska@suse.cz>
172 * config/i386/i386.opt: Replace -masm-dialect with -masm.
174 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
177 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
179 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
182 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
183 (arm_function_ok_for_sibcall): Return false for an indirect call by
184 descriptor if all the argument registers are used.
185 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
186 alignment of the function.
188 2017-02-21 Jakub Jelinek <jakub@redhat.com>
190 PR tree-optimization/61441
191 * simplify-rtx.c (simplify_const_unary_operation): For
192 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
195 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
197 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
198 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
199 instead of SYSTEM_HEADER_DIR.
201 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
202 Martin Liška <mliska@suse.cz>
204 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
205 Fix typos and grammar, use active voice, and clarify.
207 2017-02-20 Marek Polacek <polacek@redhat.com>
210 * gimplify.c (gimplify_expr): Handle unused *&&L;.
213 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
215 2017-02-20 Jakub Jelinek <jakub@redhat.com>
218 * config/i386/i386.c (ix86_expand_builtin): Handle
219 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
220 ix86_builtins_isa[fcode].isa as a requirement of those
221 flags and any other flag in the bitmask.
222 (ix86_init_mmx_sse_builtins): Use 0 instead of
223 ~OPTION_MASK_ISA_64BIT as mask.
224 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
225 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
226 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
227 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
229 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
232 * lra-constraints.c (split_reg): Check requested split mode
233 is supported by the register.
235 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
237 * lra-constraints.c (simplify_operand_subreg): Remove early
240 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
243 * lra-constraints.c (curr_insn_transform): Tighten condition
244 for converting SUBREG reloads from OP_OUT to OP_INOUT.
246 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
249 * lra-constraints.c (curr_insn_transform): Handle
250 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
252 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
255 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
257 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
259 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
262 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
265 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
267 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
270 2017-02-18 Jakub Jelinek <jakub@redhat.com>
272 * final.c (last_columnnum, override_columnnum): New variables.
273 (final_start_function): Set last_columnnum, pass it to begin_prologue
274 hook and pass 0 to dwarf2out_begin_prologue.
275 (final_scan_insn): Update override_columnnum. Pass last_columnnum
276 to source_line debug hook.
277 (notice_source_line): Compute last_columnnum and for debug_column_info
278 return true on column changes.
279 * debug.h (struct gcc_debug_hooks): Add column argument to
280 source_line and begin_prologue hooks.
281 (debug_nothing_int_charstar_int_bool): Remove prototype.
282 (debug_nothing_int_int_charstar,
283 debug_nothing_int_int_charstar_int_bool): New prototypes.
284 (dwarf2out_begin_prologue): Add column argument.
285 * debug.c (do_nothing_debug_hooks): Adjust source_line and
286 begin_prologue hooks.
287 (debug_nothing_int_charstar_int_bool): Remove.
288 (debug_nothing_int_int_charstar,
289 debug_nothing_int_int_charstar_int_bool): New functions.
290 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
291 through to dwarf2out_source_line.
292 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
293 (dwarf2out_source_line): Add column argument, emit it if requested.
294 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
296 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
297 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
298 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
299 through to dwarf2out_begin_prologue.
300 (vmsdbgout_source_line): Add column argument, pass it through to
301 dwarf2out_source_line.
302 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
303 dbxout_source_line caller.
304 (dbxout_source_line): Add column argument.
306 * common.opt (gno-column-info, gcolumn-info): New options.
307 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
308 (check_die): Also test for multiple DW_AT_decl_column attributes.
309 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
310 DW_AT_decl_column if requested.
311 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
313 (gen_variable_die): Likewise.
314 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
315 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
318 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
319 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
320 (ix86_handle_option): Handle OPT_m3dnowa.
321 * doc/invoke.texi (-m3dnowa): Document.
322 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
323 -m3dnowa instead of -m3dnow -march=athlon.
326 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
327 instead of gcc_assert for K, r and R code checks. Formatting fixes.
329 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
332 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
333 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
334 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
335 generator for vsx_xxpermdi_<mode>_be.
336 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
337 force big-endian semantics.
338 (vsx_xxpermdi_<mode>_be): New define_expand with same
339 implementation as previous version of vsx_xxpermdi_<mode>.
341 2017-02-17 Jakub Jelinek <jakub@redhat.com>
343 PR tree-optimization/79327
344 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
345 variable, its initialization and use.
347 2017-02-17 Julia Koval <julia.koval@intel.com>
349 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
350 (OPTION_MASK_ISA_PKU_UNSET): New.
351 (ix86_handle_option): Handle -mrdpid.
352 * config/i386/cpuid.h (bit_RDPID): New.
353 * config/i386/driver-i386.c (host_detect_local_cpu):
354 Detect RDPID feature.
355 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
356 * config/i386/i386-c.c (ix86_target_macros_internal):
358 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
359 (ix86_valid_target_attribute_inner_p): Add "rdpid".
360 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
361 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
362 * config/i386/i386.md (define_insn "rdpid"): New.
363 * config/i386/i386.opt Add -mrdpid.
364 * config/i386/immintrin.h (_rdpid_u32): New.
366 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
368 PR rtl-optimization/79541
369 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
370 instead of transforming it into USE.
372 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
374 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
375 If HONOR_SNANS (SFmode) force the input to a register.
376 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
377 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
378 an frsp or similar insn.
380 2017-02-17 Martin Liska <mliska@suse.cz>
382 PR rtl-optimization/79577
383 * params.def (selsched-max-sched-times): Increase minimum to 1.
385 2017-02-17 Martin Liska <mliska@suse.cz>
387 PR rtl-optimization/79574
388 * gcse.c (want_to_gcse_p): Prevent integer overflow.
390 2017-02-17 Martin Liska <mliska@suse.cz>
392 PR tree-optimization/79529
393 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
394 ssa_defined_default_def_p to handle cases which are implicitly
396 * tree-ssa.c (ssa_defined_default_def_p): New function.
397 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
398 which are implicitly defined.
399 * tree-ssa.h (ssa_defined_default_def_p): Declare.
401 2017-02-17 Richard Biener <rguenther@suse.de>
404 * params.def (max-ssa-name-query-depth): Limit to 10.
406 2017-02-17 Richard Biener <rguenther@suse.de>
408 PR tree-optimization/79552
409 * tree-ssa-structalias.c (visit_loadstore): Properly verify
412 2017-02-17 Richard Biener <rguenther@suse.de>
415 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
417 2017-02-17 Marek Polacek <polacek@redhat.com>
420 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
421 (fold_negate_expr): New wrapper.
423 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
425 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
426 Correct terminology and de-emphasize pre-standard behavior.
428 2017-02-16 Alan Modra <amodra@gmail.com>
430 PR rtl-optimization/79286
431 * ira.c (def_dominates_uses): New function.
432 (update_equiv_regs): Don't create an equivalence for insns that
433 may trap where the register def does not dominate the use.
435 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
437 PR rtl-optimization/78127
438 * lra.c (lra): Call lra_eliminate before finish the loop after
441 2017-02-16 Richard Biener <rguenther@suse.de>
443 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
445 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
446 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
447 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
448 (isl_val_int_from_wi): New function.
449 (extract_affine_gmp): Rename to ...
450 (extract_affine_wi): ... this, take a widest_int.
451 (extract_affine_int): Just wrap extract_affine_wi.
452 (add_param_constraints): Use isl_val_int_from_wi.
453 (add_loop_constraints): Likewise, and extract_affine_wi.
455 2017-02-15 Jeff Law <law@redhat.com>
458 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
459 ira_init_register_move_cost_if_necessary.
461 2017-02-15 Martin Sebor <msebor@redhat.com>
464 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
465 removed in a prior commit.
467 2017-02-15 Bin Cheng <bin.cheng@arm.com>
469 PR tree-optimization/79347
470 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
471 counters during peeling.
473 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
475 * Makefile.in (site.exp): Remove "set ISLVER".
477 2017-02-15 Jakub Jelinek <jakub@redhat.com>
480 * real.c (real_from_integer): Call real_convert even for decimal.
482 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
485 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
487 2017-02-14 Andrew Pinski <apinski@cavium.com>
489 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
490 cores and change the partno/implementer to be correct.
491 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
492 the 'B" as the implementer.
493 * config/aarch64/aarch64-tune.md: Regenerate.
495 2017-02-14 Carl Love <cel@us.ibm.com>
497 * config/rs6000/rs6000.c: Add case statement entry to make the
498 xvcvuxdsp built-in argument unsigned.
499 * config/rs6000/vsx.md: Fix the source and return operand types so they
500 match the instruction definitions from the ISA document. Fix typo
501 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
504 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
507 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
508 member early_clobber_alts.
509 * lra-lives.c (reg_early_clobber_p): New.
510 (process_bb_lives): Use it.
511 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
512 (debug_operand_data): Initialize early_clobber_alts.
513 (setup_operand_alternative): Set up early_clobber_alts.
514 (collect_non_operand_hard_regs): Ditto. Pass early clobber
515 alternatives to new_insn_reg.
516 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
518 (lra_update_insn_regno_info): Pass the new arg.
520 2017-02-14 Jakub Jelinek <jakub@redhat.com>
523 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
524 (new_oacc_loop_raw): Don't clear already cleared fields.
527 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
528 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
529 _mm512_prefetch_i64gather_ps): New inline functions and macros.
531 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
534 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
536 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
539 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
540 the extra instruction to the right place to store 128-bit constant
543 2017-02-14 Martin Sebor <msebor@redhat.com>
546 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
547 warning for strings of unknown length.
549 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
551 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
553 2017-02-14 Jeff Law <law@redhat.com>
556 * ira-costs.c (scan_one_insn): Initialize register move costs
557 for pseudos seen in USE/CLOBBER insns.
559 PR tree-optimization/79095
560 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
561 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
562 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
563 if the operands are known to be not equal, then the resulting range
565 (intersect_ranges): If the new range is ~[0,0] and the old range is
566 wide, then prefer ~[0,0].
567 * tree-vrp.c (overflow_comparison_p_1): New function.
568 (overflow_comparison_p): New function.
569 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
570 if NAME is used in an overflow test.
571 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
572 overflow check that can be expressed as an equality test, then adjust
573 ops to be that equality test.
575 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
577 * config/s390/s390-builtin-types.def: Remove flags argument.
578 * config/s390/s390.c (s390_init_builtins): Likewise.
580 2017-02-14 Martin Liska <mliska@suse.cz>
582 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
583 vector. Fix trailing white spaces.
585 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
587 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
590 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
592 PR rtl-optimization/68664
593 * config/arm/arm.c (arm_sched_can_speculate_insn):
594 New function. Declare prototype.
595 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
597 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
599 PR rtl-optimization/68664
600 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
602 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
604 2017-02-14 Amit Pawar <amit.pawar@amd.com>
606 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
607 max skip bytes for function, loop and jump.
609 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
611 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
612 ABS_EXPR for gimple dump.
614 2017-02-14 Jakub Jelinek <jakub@redhat.com>
617 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
619 PR tree-optimization/79408
620 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
621 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
622 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
623 also if rhs1 is INTEGER_CST.
625 2017-02-14 Richard Biener <rguenther@suse.de>
628 * tree-into-ssa.c (insert_phi_nodes): When the function can
629 have abnormal edges rewrite SSA names with broken use-def
630 dominance out of SSA and register them for PHI insertion.
632 2017-02-13 Martin Sebor <msebor@redhat.com>
635 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
636 clearing info.nowrite flag when snprintf size argument is a range.
638 2017-02-13 Jakub Jelinek <jakub@redhat.com>
640 * cprop.c (cprop_jump): Add missing space in string literal.
641 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
642 (get_constraint_for_component_ref): Likewise.
643 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
644 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
645 * lra-constraints.c (process_alt_operands): Likewise.
646 * ipa-inline.c (inline_small_functions): Likewise.
647 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
648 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
649 * trans-mem.c (diagnose_tm_1_op): Likewise.
650 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
651 (grid_parallel_clauses_gridifiable): Likewise.
653 * config/nvptx/mkoffload.c (process): Add space in between
656 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
657 "MOD4_SSE_REGS" and "ALL_REGS".
659 * spellcheck.c (test_data): Add , in between "foo" and "food".
661 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
664 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
665 boundary crossing check and subsequent code generation agree.
667 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
669 * config/aarch64/aarch64.c (has_memory_op): Delete.
670 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
673 2017-02-13 Jakub Jelinek <jakub@redhat.com>
675 PR rtl-optimization/79388
676 PR rtl-optimization/79450
677 * combine.c (distribute_notes): When removing TEM_INSN for which
678 corresponding dest has last value recorded, invalidate that last
681 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
683 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
684 of explicit '@'. Add missing assembly comment marker on branch costs
687 2017-02-13 Nathan Sidwell <nathan@acm.org>
689 * gengtype-lex.l (<in_struct>): Add '/'.
691 2017-02-13 Martin Liska <mliska@suse.cz>
694 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
696 2017-02-13 Richard Biener <rguenther@suse.de>
698 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
700 * configure: Re-generate.
701 * config.in: Likewise.
702 * graphite-dependences.c: Simplify as if
703 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
704 * graphite-isl-ast-to-gimple.c: Likewise.
705 * graphite-optimize-isl.c: Likewise.
706 * graphite-poly.c: Likewise.
707 * graphite-sese-to-poly.c: Likewise.
708 * graphite.h: Likewise.
709 * toplev.c: Include isl/version.h and use isl_version () for
710 printing the ISL version.
711 * doc/install.texi: Update ISL requirement.
713 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
715 * doc/standards.texi (Standards): Update reference to
718 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
720 * doc/extend.texi (Named Address Spaces): sourceware.org now
722 * doc/install.texi (Binaries): Ditto.
725 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
727 * doc/cpp.texi: Replace "stringify"/"stringification" with C
728 standard terminology "stringize"/"stringizing" throughout.
729 * doc/cppinternals.texi: Likewise.
731 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
733 * doc/extend.texi: Fix some spelling mistakes and typos.
734 * doc/invoke.texi: Likewise.
736 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
739 * params.def (inline-min-speedup) Change from 10 to 8.
741 2017-02-11 Jakub Jelinek <jakub@redhat.com>
743 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
746 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
749 * ipa-inline-analysis.c (get_minimal_bb): New function.
750 (record_modified): Use it.
751 (remap_edge_change_prob): Handle also ancestor functions.
753 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
755 * doc/contrib.texi (Contributors): Remove broken link into
756 the Mauve CVS repository.
758 2017-02-11 Jakub Jelinek <jakub@redhat.com>
761 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
762 result computation whenever lhs doesn't have vector mode, not
763 just when it has BLKmode.
765 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
767 * doc/makefile.texi (profiledbootstrap): Refer to the
768 installation instructions only in textual form.
770 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
773 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
775 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
777 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
778 (Specific): Update mingw-w64 reference.
780 (Specific): Remove broken link to Renesas RX processor.
782 2017-02-10 Richard Biener <rguenther@suse.de>
784 * toplev.c (process_options): Do not mention obsolete graphite
785 options when printing sorry message about missing graphite support.
786 Mention -floop-nest-optimize.
788 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
790 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
791 (vtst_p16): Likewise.
792 (vtstq_p8): Likewise.
793 (vtstq_p16): Likewise.
795 (vtstq_p64): Likewise.
796 * config/arm/arm_neon.h (vgetq_lane_p64): New.
797 (vset_lane_p64): New.
798 (vsetq_lane_p64): New.
800 2017-02-10 Jakub Jelinek <jakub@redhat.com>
802 PR tree-optimization/79411
803 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
804 stmt operands are SSA_NAMEs used in abnormal phis.
805 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
808 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
811 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
814 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
816 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
818 2017-02-09 Jakub Jelinek <jakub@redhat.com>
820 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
824 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
825 not arbitrary TREE_CONSTANT.
828 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
829 "omp declare target link" attribute unless is_global_var.
830 * omp-offload.c (find_link_var_op): Likewise.
832 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
833 Chung-Lin Tang <cltang@codesourcery.com>
835 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
837 (gimplify_adjust_omp_clauses): Don't delete TILE.
838 (gimplify_omp_for): Deal with TILE.
839 * internal-fn.c (expand_GOACC_TILE): New function.
840 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
842 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
843 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
845 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
846 avoid DIV for outermost collapse var.
847 (expand_oacc_for): Insert tile element loop as needed. Adjust.
848 Remove out of date comments, fix whitespace.
849 * omp-general.c (omp_extract_for_data): Deal with tiling.
850 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
851 adjust OLF_DIM_BASE value.
852 (struct omp_for_data): Add tiling field.
853 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
854 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
855 for auto loops. Remove default auto determining, moved to
856 oacc_loop_fixed_partitions.
857 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
858 stmts, add e_mask field.
859 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
860 (oacc_thread_numbers): Use oacc_dim_call.
861 (oacc_xform_tile): New.
862 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
863 (finish_oacc_loop): Adjust for ifns vector.
864 (oacc_loop_discover_walk): Append loop abstraction sites to list,
865 add case for GOACC_TILE fns.
866 (oacc_loop_xform_loop): Delete.
867 (oacc_loop_process): Iterate over call list directly, and add
868 handling for GOACC_TILE fns.
869 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
871 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
872 vector partitioning to outer loops. Assign 2 partitions to loops
873 when available. Add TILE handling.
874 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
875 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
876 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
877 * tree.c (omp_clause_num_ops): Adjust TILE ops.
878 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
880 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
882 * configure.ac (ACX_BUGURL): Update.
883 * configure: Regenerate.
885 2017-02-09 Richard Biener <rguenther@suse.de>
887 PR tree-optimization/69823
888 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
889 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
891 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
893 * config/arc/arc-c.def: Add __NPS400__ definition.
894 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
895 (TARGET_NPS400): Define.
897 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
899 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
901 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
903 (arc_cpu_types): Fill the arch_info field with a pointer into the
904 arc_arch_types table.
905 (arc_selected_cpu): Declare.
906 * config/arc/arc.c (arc_selected_cpu): Make global.
907 (arc_selected_arch): Delete.
908 (arc_base_cpu): Delete.
909 (arc_override_options): Remove references to deleted variables,
910 update access to arch information.
911 (ARC_OPT): Update access to arch information.
912 (ARC_OPTX): Likewise.
913 * config/arc/arc.h (arc_base_cpu): Remove declaration.
914 (TARGET_ARC600): Update access to arch information.
915 (TARGET_ARC601): Likewise.
916 (TARGET_ARC700): Likewise.
917 (TARGET_EM): Likewise.
918 (TARGET_HS): Likewise.
919 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
922 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
925 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
926 condition/operands for integer GE/LE/GEU/LEU operations.
928 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
931 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
934 2017-02-08 Martin Jambor <mjambor@suse.cz>
937 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
938 whether allocation happened.
939 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
940 nothing was allocated.
942 2017-02-08 Jakub Jelinek <jakub@redhat.com>
944 PR tree-optimization/79408
945 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
946 constant, but SSA_NAME with a known integer range, use the minimum
947 of that range instead of op1 to determine if modulo can be replaced
948 with its first operand.
950 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
952 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
954 2017-02-08 Richard Biener <rguenther@suse.de>
956 PR tree-optimization/71824
957 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
958 Check all loops contained in the merged region.
960 2017-02-07 Andrew Pinski <apinski@cavium.com>
962 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
964 2017-02-07 Andrew Pinski <apinski@cavium.com>
966 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
967 (thunderxt88): Likewise.
968 (thunderxt81): Disable LSE and change v8.1 to v8.
969 (thunderxt83): Likewise.
971 2017-02-07 Jakub Jelinek <jakub@redhat.com>
972 Richard Biener <rguenther@suse.de>
975 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
976 type from int to size_t.
977 * ira-costs.c (struct_costs_size): Change type from int to size_t.
979 2017-02-07 Jakub Jelinek <jakub@redhat.com>
981 PR rtl-optimization/79386
982 * cprop.c (bypass_conditional_jumps): Initialize
983 bypass_last_basic_block already before splitting bbs after
984 unconditional traps...
985 (bypass_conditional_jumps): ... rather than here.
988 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
989 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
990 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
991 fix -masm=intel patterns.
993 2017-02-07 Richard Biener <rguenther@suse.de>
995 PR tree-optimization/79256
997 * builtins.c (get_object_alignment_2): Use min_align_of_type
998 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
999 and ADJUST_FIELD_ALIGN.
1001 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
1003 * doc/tm.texi: Regenerate.
1004 * stor-layout.c (layout_decl): Adjust.
1005 (update_alignment_for_field): Likewise.
1006 (place_field): Likewise.
1007 (min_align_of_type): Likewise.
1008 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
1009 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
1010 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
1011 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
1012 * config/frv/frv.c (frv_adjust_field_align): Likewise.
1013 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
1014 * config/i386/i386.c (x86_field_alignment): Likewise.
1015 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
1016 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
1017 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
1018 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
1019 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
1020 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
1024 2017-01-30 Richard Biener <rguenther@suse.de>
1026 PR tree-optimization/79256
1027 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
1028 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
1031 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
1033 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
1034 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
1035 builtins to SImode and emit a zero-extend, if necessary.
1037 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1039 * docs/invoke.texi (RISC-V Options): Alphabetize.
1041 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1043 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
1046 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1048 * config/riscv/riscv.c: New file.
1049 * gcc/common/config/riscv/riscv-common.c: Likewise.
1050 * config.gcc: Likewise.
1051 * config/riscv/constraints.md: Likewise.
1052 * config/riscv/elf.h: Likewise.
1053 * config/riscv/generic.md: Likewise.
1054 * config/riscv/linux.h: Likewise.
1055 * config/riscv/multilib-generator: Likewise.
1056 * config/riscv/peephole.md: Likewise.
1057 * config/riscv/pic.md: Likewise.
1058 * config/riscv/predicates.md: Likewise.
1059 * config/riscv/riscv-builtins.c: Likewise.
1060 * config/riscv/riscv-c.c: Likewise.
1061 * config/riscv/riscv-ftypes.def: Likewise.
1062 * config/riscv/riscv-modes.def: Likewise.
1063 * config/riscv/riscv-opts.h: Likewise.
1064 * config/riscv/riscv-protos.h: Likewise.
1065 * config/riscv/riscv.h: Likewise.
1066 * config/riscv/riscv.md: Likewise.
1067 * config/riscv/riscv.opt: Likewise.
1068 * config/riscv/sync.md: Likewise.
1069 * config/riscv/t-elf-multilib: Likewise.
1070 * config/riscv/t-linux: Likewise.
1071 * config/riscv/t-linux-multilib: Likewise.
1072 * config/riscv/t-riscv: Likewise.
1073 * configure.ac: Likewise.
1074 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
1075 Waterman as RISC-V maintainers.
1076 * doc/install.texi: Add RISC-V entries.
1077 * doc/invoke.texi: Add RISC-V options section.
1078 * doc/md.texi: Add RISC-V constraints section.
1079 * configure: Regenerated.
1081 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1084 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
1085 false values to be constant vectors with all 0 or all 1 bits set.
1086 (vcondu<mode><mode>): Likewise.
1087 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
1089 (fpmask_comparison_operator): Update comment.
1090 (vecint_comparison_operator): New predicate.
1091 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
1092 vector conditionals when the true and false values are constant
1093 vectors with all 0 bits or all 1 bits set.
1095 2017-02-06 Martin Sebor <msebor@redhat.com>
1097 PR tree-optimization/79376
1098 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
1100 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
1102 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
1103 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
1104 to simplify split condition.
1106 2017-02-06 Jakub Jelinek <jakub@redhat.com>
1108 * omp-expand.c (oxpand_omp_atomic_fetch_op,
1109 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
1112 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
1114 PR rtl-optimization/68664
1115 * target.def (can_speculate_insn): New hook.
1116 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
1117 * doc/tm.texi: Regenerate.
1118 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
1119 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
1120 (rs6000_sched_can_speculate_insn): New function.
1122 2017-02-06 Jakub Jelinek <jakub@redhat.com>
1124 PR tree-optimization/79284
1125 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
1126 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
1127 vectorizable_mask_load_store, vectorizable_operation,
1128 vect_is_simple_cond, get_same_sized_vectype): Use it instead
1129 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
1130 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
1131 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
1132 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1133 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
1134 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
1135 is_gimple_assign (stmt). Replace another such test with
1136 is_gimple_assign (stmt).
1138 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
1141 * config/avr/avr.c (rtl-iter.h): Include it.
1142 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
1143 (avr_legitimate_combined_insn): ...and implementation.
1145 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
1147 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
1148 * config/s390/s390.c (s390_const_operand_ok)
1149 (s390_canonicalize_comparison, s390_extract_part)
1150 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
1151 (s390_contiguous_bitmask_p, s390_rtx_costs)
1152 (legitimize_pic_address): Likewise.
1153 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
1154 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
1155 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
1156 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
1157 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
1159 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
1161 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
1162 REGNO($0) == REGNO($1).
1164 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1166 * config/s390/linux.h(SIZE_TYPE): Add comment.
1168 2017-02-06 Julian Brown <julian@codesourcery.com>
1169 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1170 Virendra Pathak <virendra.pathak@broadcom.com>
1172 * config/aarch64/aarch64-cores.def: Change the scheduler
1174 * config/aarch64/aarch64.md: Include thunderx2t99.md.
1175 * config/aarch64/thunderx2t99.md: New file.
1177 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1179 * doc/standards.texi (Go Language): Update link to language
1182 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
1184 * tree-eh.c (lower_resx): Sanitize profile.
1185 (cleanup_empty_eh_move_lp): Likewise.
1187 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
1190 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
1192 * cfgloopmanip.h (loop_version): Update prototype.
1193 * modulo-sched.c (sms_schedule): Update call of loop_version.
1194 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
1195 * tree-parloops.c (gen_parallel_loop): Likewise.
1196 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
1197 * tree-ssa-loop-split.c (split_loop): Likewise.
1198 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
1199 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1201 2017-02-05 Martin Liska <mliska@suse.cz>
1204 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
1206 (print_operand_address): Initialize a struct to zero.
1208 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1210 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
1211 garbage collector only in textual form.
1213 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1215 * doc/extend.texi (x86 specific memory model extensions for
1216 transactional memory): Simplify a phrase.
1218 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
1221 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
1222 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
1223 (atomic_storedi_1): Likewise.
1225 2017-02-04 Jakub Jelinek <jakub@redhat.com>
1227 PR tree-optimization/79338
1228 * tree-parloops.c (gather_scalar_reductions): Don't call
1229 vect_analyze_loop_form for loop->inner before destroying loop's
1232 2017-02-03 Martin Sebor <msebor@redhat.com>
1234 PR tree-optimization/79327
1235 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
1236 when precision has resulted in leading zeros.
1237 (format_integer): Adjust the likely counter to assume an unknown
1238 argument that may be zero is non-zero.
1240 2017-02-03 Jason Merrill <jason@redhat.com>
1243 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
1244 avoid copying non-taken branch.
1246 2017-02-03 Jakub Jelinek <jakub@redhat.com>
1248 PR tree-optimization/79340
1249 * tree-vect-loop.c (vectorizable_reduction): Release
1250 vec_defs elements after safe_splicing them into other vectors.
1253 PR tree-optimization/79327
1254 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
1255 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
1257 (format_integer): Use wide_int_to_tree instead of build_int_cst
1258 + to_?hwi. If argmin is NULL, just set argmin and argmax to
1259 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
1260 of shortest and longest sequence.
1262 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
1264 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
1265 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
1267 2017-02-03 Walter Lee <walt@tilera.com>
1270 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
1271 after initial stackframe link reg save.
1272 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
1274 2017-02-03 Jakub Jelinek <jakub@redhat.com>
1277 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
1278 wu for stxssp alternative.
1280 2017-02-03 Martin Sebor <msebor@redhat.com>
1282 PR tree-optimization/79352
1283 * gimple-fold.c (get_range_strlen): Add argument.
1284 (get_range_strlen): Change return type to bool.
1285 (get_maxval_strlen): Pass in a dummy argument.
1286 * gimple-fold.h (get_range_strlen): Change return type to bool.
1287 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
1288 * tree.h (array_at_struct_end_p): Add argument.
1289 * tree.c (array_at_struct_end_p): Handle it.
1291 2017-02-03 Martin Liska <mliska@suse.cz>
1294 * multiple_target.c (create_dispatcher_calls): Redirect edge
1295 from a caller of a dispatcher.
1296 (expand_target_clones): Make the clones local.
1297 (ipa_target_clone): Do both target clones and resolvers.
1298 (ipa_dispatcher_calls): Remove the pass.
1299 (pass_dispatcher_calls::gate): Likewise.
1300 (make_pass_dispatcher_calls): Likewise.
1301 * passes.def (pass_target_clone): Put as very first IPA early
1304 2017-02-03 Martin Liska <mliska@suse.cz>
1306 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
1307 in case of a function with ifunc attribute.
1309 2017-02-03 Martin Liska <mliska@suse.cz>
1311 * cgraph.c (cgraph_node::dump): Dump function version info.
1312 * symtab.c (symtab_node::dump_base): Add missing new line.
1314 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1316 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
1317 (ifcombine_ifandif): Use it.
1319 2017-02-03 Martin Liska <mliska@suse.cz>
1321 * doc/invoke.texi: Document default value for
1322 use-after-scope-direct-emission-threshold.
1324 2017-02-03 Martin Liska <mliska@suse.cz>
1326 PR tree-optimization/79339
1327 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
1328 (format_floating): Likewise.
1330 2017-02-03 Martin Liska <mliska@suse.cz>
1333 * ipa-prop.c (ipa_node_params_t::insert): Remove current
1335 (ipa_node_params_t::remove): Likewise.
1336 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
1337 initialization from removed ipa_node_params_t::insert.
1338 (ipa_node_params::~ipa_node_params): Move from removed
1339 ipa_node_params_t::release.
1340 * symbol-summary.h (symbol_summary::m_released): New member.
1341 Do not release a summary twice. Do not allow to call finalizer
1342 for types of a summary that live in GGC memory.
1344 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1346 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
1349 2017-02-02 Martin Sebor <msebor@redhat.com>
1352 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
1353 (format_string): Tighten up the range of output for non-constant
1354 strings and correct the expected range for wide non-constant strings.
1356 2017-02-02 Martin Sebor <msebor@redhat.com>
1358 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
1361 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
1363 (-fdump-tree-@var): Add to index and document how to come up
1364 with pass-specific option and dump file names.
1365 (-fdump-passes): Clarify where to look for output.
1367 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1370 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
1371 statistics of the analyzed path; allow threading for speed when
1372 any of BBs along the path are optimized for speed.
1374 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
1377 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
1378 settings of the virtual registers.
1381 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
1383 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1384 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
1387 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1389 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
1390 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
1392 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1394 * config/s390/s390.md: Add missing comments with the expanded
1396 * config/s390/vector.md: Likewise.
1397 * config/s390/vx-builtins.md: Likewise.
1399 2017-02-02 Jakub Jelinek <jakub@redhat.com>
1402 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
1403 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
1404 conditions on a single line.
1406 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1408 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1409 __S390_VX__ to __VX__.
1411 2017-02-01 Andrew Pinski <apinski@cavium.com>
1413 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
1414 stmt_info to record_stmt_cost.
1415 (vect_get_known_peeling_cost): Pass stmt_info if known to
1417 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
1418 cpu_vector_cost field into
1419 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
1420 field into vec_int_stmt_cost and vec_fp_stmt_cost.
1421 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
1422 splitting of scalar_stmt_cost and vec_stmt_cost.
1423 (thunderx_vector_cost): Likewise.
1424 (cortexa57_vector_cost): LIkewise.
1425 (exynosm1_vector_cost): Likewise.
1426 (xgene1_vector_cost): Likewise.
1427 (thunderx2t99_vector_cost): Improve after the splitting of the two
1429 (aarch64_builtin_vectorization_cost): Update for the splitting of
1430 scalar_stmt_cost and vec_stmt_cost.
1432 2017-02-01 Torvald Riegel <triegel@redhat.com>
1433 Richard Henderson <rth@redhat.com>
1435 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
1436 conditional on existance of a fast atomic load.
1437 * optabs-query.c (can_atomic_load_p): New function.
1438 * optabs-query.h (can_atomic_load_p): Declare it.
1439 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
1440 no fast atomic load is available for the particular size of access.
1441 (expand_atomic_compare_and_swap): Likewise.
1442 (expand_atomic_load): Likewise.
1443 (expand_atomic_store): Likewise.
1444 (expand_atomic_fetch_op): Likewise.
1445 * testsuite/lib/target-supports.exp
1446 (check_effective_target_sync_int_128): Remove x86 because it provides
1447 no fast atomic load.
1448 (check_effective_target_sync_int_128_runtime): Likewise.
1450 2017-02-01 Richard Biener <rguenther@suse.de>
1452 * graphite.c: Include tree-vectorizer.h for find_loop_location.
1453 (graphite_transform_loops): Provide opt-info for optimized nests.
1454 * tree-parloop.c (parallelize_loops): Provide opt-info for
1457 2017-02-01 Richard Biener <rguenther@suse.de>
1460 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
1463 2017-02-01 Richard Biener <rguenther@suse.de>
1465 PR tree-optimization/71824
1466 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
1467 Verify the loops are valid in the merged SESE region.
1468 (scop_detection::can_represent_loop_1): Check analyzing the
1469 evolution of the number of iterations in the region succeeds.
1471 2017-01-31 Ian Lance Taylor <iant@golang.org>
1473 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
1474 REG_ARGS_SIZE note to 32-bit push insns and call insn.
1476 2017-01-31 David Malcolm <dmalcolm@redhat.com>
1478 PR preprocessor/79210
1479 * input.c (get_substring_ranges_for_loc): Replace line_width
1480 assertion with error-handling.
1482 2017-01-31 Richard Biener <rguenther@suse.de>
1484 PR tree-optimization/77318
1485 * graphite-sese-to-poly.c (extract_affine): Fix assert.
1486 (create_pw_aff_from_tree): Take loop parameter.
1487 (add_condition_to_pbb): Pass loop of the condition to
1488 create_pw_aff_from_tree.
1490 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1492 * config/s390/s390.c (s390_asan_shadow_offset): New function.
1493 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
1495 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1499 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
1501 (convert_int_to_float128): Likewise.
1502 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
1503 (convert_int_to_float128): Likewise.
1504 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
1505 (UNSPEC_IEEE128_CONVERT): Likewise.
1506 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
1507 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
1508 Use local variables for IBM extended format.
1509 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
1510 (fix_trunc<mode>si2_fprs): Likewise.
1511 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
1512 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
1513 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
1514 to know that we can now have integers of all sizes in vector
1516 (fix<uns>_<mode>di2_hw): Likewise.
1517 (float<uns>_<mode>si2_hw): Likewise.
1518 (fix_<mode>si2_hw): Likewise.
1519 (fixuns_<mode>si2_hw): Likewise.
1520 (float<uns>_<mode>di2_hw): Likewise.
1521 (float_<mode>di2_hw): Likewise.
1522 (float_<mode>si2_hw): Likewise.
1523 (floatuns_<mode>di2_hw): Likewise.
1524 (floatuns_<mode>si2_hw): Likewise.
1525 (xscvqp<su>wz_<mode>): Delete, no longer used.
1526 (xscvqp<su>dz_<mode>): Likewise.
1527 (xscv<su>dqp_<mode>): Likewise.
1528 (ieee128_mfvsrd_64bit): Likewise.
1529 (ieee128_mfvsrd_32bit): Likewise.
1530 (ieee128_mfvsrwz): Likewise.
1531 (ieee128_mtvsrw): Likewise.
1532 (ieee128_mtvsrd_64bit): Likewise.
1533 (ieee128_mtvsrd_32bit): Likewise.
1535 2017-01-31 Martin Liska <mliska@suse.cz>
1538 * ipa-prop.c (ipa_free_all_node_params): Call release method
1539 instead of ~sumbol_summary to not to trigger double times
1542 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
1544 PR tree-optimization/71691
1545 * bitmap.h (class auto_bitmap): New.
1546 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
1547 is_maybe_undefined instead of ssa_undefined_value_p.
1549 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1551 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1552 __S390_ARCH_LEVEL__ to __ARCH__.
1554 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1556 PR tree-optimization/79267
1557 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
1558 if should_remove_lhs_p is true.
1560 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
1563 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
1564 (add_alignment_attribute): New.
1565 (base_type_die): Add alignment attribute.
1566 (subrange_type_die): Likewise.
1567 (modified_type_die): Likewise.
1568 (gen_array_type_die): Likewise.
1569 (gen_descr_array_type_die: Likewise.
1570 (gen_enumeration_type_die): Likewise.
1571 (gen_subprogram_die): Likewise.
1572 (gen_variable_die): Likewise.
1573 (gen_field_die): Likewise.
1574 (gen_ptr_to_mbr_type_die): Likewise.
1575 (gen_struct_or_union_type_die): Likewise.
1576 (gen_subroutine_type_die): Likewise.
1577 (gen_typedef_die): Likewise.
1578 (base_type_cmp): Compare alignment attribute.
1580 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1583 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
1584 (setb_unsigned) New pattern for setb with CCUNS.
1585 * config/rs6000/rs6000.c (expand_block_compare): Use a different
1586 subfc./subfe sequence to avoid overflow problems. Generate a
1587 shorter sequence with cmpld/setb for power9.
1588 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
1589 for generating subfc. instruction.
1590 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
1591 now uses this instruction.
1593 2017-01-30 Ian Lance Taylor <iant@google.com>
1596 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
1597 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
1599 2017-01-30 Martin Sebor <msebor@redhat.com>
1601 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
1602 Move constant to the right of a relational operator.
1603 (get_mpfr_format_length, format_character, format_string): Ditto.
1604 (should_warn_p, maybe_warn): Same.
1606 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
1608 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1611 * asan.c (get_translation_unit_decl): Remove function.
1612 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
1614 2017-01-30 Martin Liska <mliska@suse.cz>
1616 PR gcov-profile/79259
1617 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
1620 2017-01-30 Martin Liska <mliska@suse.cz>
1623 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
1624 Initialize variables with NULL value.
1626 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
1629 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
1631 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
1633 2017-01-30 Richard Biener <rguenther@suse.de>
1635 PR tree-optimization/79276
1636 * tree-vrp.c (process_assert_insertions): Properly adjust common
1637 when removing a duplicate.
1639 * gcc.dg/torture/pr79276.c: New testcase.
1641 2017-01-30 Richard Biener <rguenther@suse.de>
1643 PR tree-optimization/79256
1644 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
1645 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
1647 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
1649 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
1652 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
1653 ("*r<noxa>sbg_<mode>_sll_bitmask")
1654 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
1655 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
1656 Use contiguous_bitmask_nowrap_operand.
1658 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1661 * config/rs6000/altivec.h (vec_xl): Revise #define.
1662 (vec_xst): Likewise.
1664 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
1666 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
1668 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
1670 PR rtl-optimization/79194
1671 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
1672 traps before call to bypass_conditional_jumps.
1674 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1676 PR tree-optimization/71374
1677 * lra-constraints.c (check_conflict_input_operands): New.
1678 (match_reload): Use it.
1680 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1683 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
1684 account to calculate conflict_set.
1686 2017-01-27 Bin Cheng <bin.cheng@arm.com>
1688 PR rtl-optimization/78559
1689 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
1690 other_insn in combine.
1692 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
1694 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
1695 uint16_type_node for BT_UINT16.
1697 2017-01-27 David Malcolm <dmalcolm@redhat.com>
1699 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
1700 "RTL Tests" to menu.
1701 (GIMPLE Tests): New node.
1702 (RTL Tests): New node.
1704 2017-01-27 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/79245
1707 * tree-loop-distribution.c (distribute_loop): Apply cost
1708 modeling also to detected patterns.
1710 2017-01-27 Richard Biener <rguenther@suse.de>
1712 PR tree-optimization/71433
1713 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
1714 (compare_assert_loc): New function.
1715 (process_assert_insertions): Sort and optimize assert locations
1716 to remove duplicates and push down identical assertions on
1717 edges to their destination block.
1719 2017-01-27 Richard Biener <rguenther@suse.de>
1721 PR tree-optimization/79244
1722 * tree-vrp.c (remove_range_assertions): Forcefully propagate
1723 out SSA names even if abnormal.
1725 2017-01-27 Jakub Jelinek <jakub@redhat.com>
1727 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
1728 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
1729 instead of MPFR_RNDN.
1731 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
1734 * arm.c (arm_option_override): Don't call build_target_option_node
1735 until after doing all option overrides.
1736 (arm_valid_target_attribute_tree): Likewise.
1738 2017-01-27 Martin Liska <mliska@suse.cz>
1740 * doc/invoke.texi (-fprofile-arcs): Document profiling support
1741 for {cd}tors and C++ {cd}tors.
1743 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
1745 * config/s390/s390.md ("*setmem_long_and")
1746 ("*setmem_long_and_31z"): Use zero_extend instead of and.
1748 2017-01-26 Martin Sebor <msebor@redhat.com>
1750 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
1753 2017-01-26 Martin Sebor <msebor@redhat.com>
1755 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
1756 HAVE_DFmode before using XFmode or DFmode.
1757 (parse_directive): Avoid using the z length modifier to avoid
1758 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
1761 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
1762 to accept adjustment as an array.
1763 (get_int_range): New function.
1764 (struct directive): Make width and prec arrays.
1765 (directive::set_width, directive::set_precision): Call get_int_range.
1766 (format_integer, format_floating): Handle width and precision ranges.
1767 (format_string, parse_directive): Same.
1769 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1772 * dwarf2out.c (generate_skeleton_bottom_up): For children with
1773 comdat_type_p set, just clone them, but keep the children in the
1777 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
1778 which have direct callers with -fvar-tracking-assignments enabled
1780 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
1781 inside of type units.
1783 2017-01-26 Martin Sebor <msebor@redhat.com>
1786 * gimple-ssa-sprintf.c (struct result_range): Add likely and
1788 (struct format_result): Replace number_chars, number_chars_min,
1789 and number_chars_max with a single member of struct result_range.
1791 (format_result::operator+=): Adjust.
1792 (struct fmtresult): Remove bounded. Handle likely and unlikely
1794 (fmtresult::adjust_for_width_or_precision): New function.
1795 (fmtresult:type_max_digits): New function.
1796 (bytes_remaining): Handle likely and unlikely counters.
1797 (min_bytes_remaining): Remove.
1798 (format_percent): Simplify.
1799 (format_integer, format_floating): Set likely and unlikely counters.
1800 (get_string_length, format_character, format_string): Same.
1801 (format_plain, should_warn_p): New function.
1802 (maybe_warn): Call should_warn_p. Update diagnostic messages
1803 and handle those for all directives, including plain strings.
1804 (format_directive): Handle likely and unlikely counters.
1805 Remove unnecessary quoting from diagnostics. Add an informational
1807 (add_bytes): Remove.
1808 (pass_sprintf_length::compute_format_length): Simplify.
1809 (try_substitute_return_value): Handle likely and unlikely counters.
1811 2017-01-26 Carl Love <cel@us.ibm.com>
1813 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
1814 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
1816 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
1819 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
1820 endianess for subregs into account.
1821 * lra-constraints.c (lra_constraints): Do risky transformations
1822 always on the first iteration.
1823 * lra-lives.c (check_pseudos_live_through_calls): Add arg
1824 last_call_used_reg_set.
1825 (process_bb_lives): Define and use last_call_used_reg_set.
1826 * lra.c (lra): Always continue after lra_constraints on the first
1829 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
1831 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
1833 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
1835 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1837 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
1838 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
1839 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
1840 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
1841 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
1842 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
1843 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
1844 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
1845 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
1847 2017-01-26 Marek Polacek <polacek@redhat.com>
1850 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
1851 for the third operand.
1853 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1856 * omp-low.c (struct omp_context): Add simt_stmt field.
1857 (scan_omp_for): Return omp_context *.
1858 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
1859 context to the _simt_ SIMD stmt.
1860 (lower_omp_for): For combined SIMD with sibling _simt_
1861 SIMD, make sure to use the same decls in _looptemp_
1862 clauses as in the sibling.
1864 2017-01-26 David Sherwood <david.sherwood@arm.com>
1867 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
1870 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1873 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
1874 emit fld b; fld a; if possible.
1876 * brig-builtins.def: Update copyright years.
1877 * config/arm/arm_acle_builtins.def: Update copyright years.
1879 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1882 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
1883 constraint instead of o for the stxsd instruction.
1885 2017-01-25 Carl Love <cel@us.ibm.com>
1887 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
1888 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
1890 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
1892 * doc/invoke.texi (C++ Dialect Options): Fix typo.
1894 2017-01-25 Richard Biener <rguenther@suse.de>
1896 PR tree-optimization/69264
1897 * target.def (vector_alignment_reachable): Improve documentation.
1898 * doc/tm.texi: Regenerate.
1899 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
1901 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
1902 earlier changes with respect to TYPE_USER_ALIGN.
1903 (vector_alignment_reachable_p): Likewise. Improve dumping.
1905 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1908 * config/arm/arm.md (xordi3): Force constant operand into a register
1911 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1913 * doc/invoke.texi (-fstore-merging): Correct default optimization
1914 levels at which it is enabled.
1915 (-O): Move -fstore-merging from list to...
1918 2017-01-25 Richard Biener <rguenther@suse.de>
1921 * omp-expand.c: Include debug.h.
1922 (expand_omp_taskreg): Make sure to generate early debug before
1923 outlining anything from a function.
1924 (expand_omp_target): Likewise.
1925 (grid_expand_target_grid_body): Likewise.
1927 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
1930 * asan.c (get_translation_unit_decl): New function.
1931 (asan_add_global): Extract modules file name from globals
1932 TRANSLATION_UNIT_DECL name.
1934 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
1937 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
1938 for long calls with APCS frame and VFP.
1940 2017-01-24 David Malcolm <dmalcolm@redhat.com>
1942 * cfg.c (original_copy_tables_initialized_p): New function.
1943 * cfg.h (original_copy_tables_initialized_p): New decl.
1944 * cfgrtl.c (relink_block_chain): Guard the call to
1945 free_original_copy_tables with a call to
1946 original_copy_tables_initialized_p.
1947 * cgraph.h (symtab_node::native_rtl_p): New decl.
1948 * cgraphunit.c (symtab_node::native_rtl_p): New function.
1949 (symtab_node::needed_p): Don't assert for early assembly output
1950 for __RTL functions.
1951 (cgraph_node::finalize_function): Set "force_output" for __RTL
1953 (cgraph_node::analyze): Bail out early for __RTL functions.
1954 (analyze_functions): Update assertion to support __RTL functions.
1955 (cgraph_node::expand): Bail out early for __RTL functions.
1956 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
1958 * function.h (struct function): Update comment for field
1960 * gimple-expr.c: Include "tree-pass.h".
1961 (gimple_has_body_p): Return false for __RTL functions.
1962 * Makefile.in (OBJS): Add run-rtl-passes.o.
1963 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
1965 (gcc::pass_manager::get_clean_slate): New accessor.
1966 * passes.c: Include "insn-addr.h".
1967 (should_skip_pass_p): Add logging. Update logic for running
1968 "expand" to be compatible with both __GIMPLE and __RTL. Guard
1969 property-provider override so it is only done for gimple passes.
1971 (skip_pass): New function.
1972 (execute_one_pass): Call skip_pass when skipping passes.
1973 * read-md.c (md_reader::read_char): Support filtering
1974 the input to a subset of line numbers.
1975 (md_reader::md_reader): Initialize fields
1976 m_first_line and m_last_line.
1977 (md_reader::read_file_fragment): New function.
1978 * read-md.h (md_reader::read_file_fragment): New decl.
1979 (md_reader::m_first_line): New field.
1980 (md_reader::m_last_line): New field.
1981 * read-rtl-function.c (function_reader::create_function): Only
1982 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
1983 curr_properties. Set DECL_INITIAL to a dummy block.
1984 (read_rtl_function_body_from_file_range): New function.
1985 * read-rtl-function.h (read_rtl_function_body_from_file_range):
1987 * run-rtl-passes.c: New file.
1988 * run-rtl-passes.h: New file.
1990 2017-01-24 Jeff Law <law@redhat.com>
1992 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
1995 2017-01-24 Bin Cheng <bin.cheng@arm.com>
1997 PR tree-optimization/79159
1998 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
1999 (record_nonwrapping_iv): Improve boundary using above function if no
2000 value range information.
2002 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
2003 Martin Jambor <mjambor@suse.cz>
2005 * brig-builtins.def: New file.
2006 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
2007 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
2008 (DEF_HSAIL_SAT_BUILTIN): Likewise.
2009 (DEF_HSAIL_INTR_BUILTIN): Likewise.
2010 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
2011 * builtin-types.def (BT_INT8): New.
2012 (BT_INT16): Likewise.
2013 (BT_UINT8): Likewise.
2014 (BT_UINT16): Likewise.
2015 (BT_FN_ULONG): Likewise.
2016 (BT_FN_UINT_INT): Likewise.
2017 (BT_FN_UINT_ULONG): Likewise.
2018 (BT_FN_UINT_LONG): Likewise.
2019 (BT_FN_UINT_PTR): Likewise.
2020 (BT_FN_ULONG_PTR): Likewise.
2021 (BT_FN_INT8_FLOAT): Likewise.
2022 (BT_FN_INT16_FLOAT): Likewise.
2023 (BT_FN_UINT32_FLOAT): Likewise.
2024 (BT_FN_UINT16_FLOAT): Likewise.
2025 (BT_FN_UINT8_FLOAT): Likewise.
2026 (BT_FN_UINT64_FLOAT): Likewise.
2027 (BT_FN_UINT16_UINT32): Likewise.
2028 (BT_FN_UINT32_UINT16): Likewise.
2029 (BT_FN_UINT16_UINT16_UINT16): Likewise.
2030 (BT_FN_INT_PTR_INT): Likewise.
2031 (BT_FN_UINT_PTR_UINT): Likewise.
2032 (BT_FN_LONG_PTR_LONG): Likewise.
2033 (BT_FN_ULONG_PTR_ULONG): Likewise.
2034 (BT_FN_VOID_UINT64_UINT64): Likewise.
2035 (BT_FN_UINT8_UINT8_UINT8): Likewise.
2036 (BT_FN_INT8_INT8_INT8): Likewise.
2037 (BT_FN_INT16_INT16_INT16): Likewise.
2038 (BT_FN_INT_INT_INT): Likewise.
2039 (BT_FN_UINT_FLOAT_UINT): Likewise.
2040 (BT_FN_FLOAT_UINT_UINT): Likewise.
2041 (BT_FN_ULONG_UINT_UINT): Likewise.
2042 (BT_FN_ULONG_UINT_PTR): Likewise.
2043 (BT_FN_ULONG_ULONG_ULONG): Likewise.
2044 (BT_FN_UINT_UINT_UINT): Likewise.
2045 (BT_FN_VOID_UINT_PTR): Likewise.
2046 (BT_FN_UINT_UINT_PTR: Likewise.
2047 (BT_FN_UINT32_UINT64_PTR): Likewise.
2048 (BT_FN_INT_INT_UINT_UINT): Likewise.
2049 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
2050 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
2051 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
2052 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
2053 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
2054 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
2055 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
2056 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
2057 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
2058 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
2059 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
2060 * doc/frontends.texi: List BRIG FE.
2061 * doc/install.texi (Testing): Add BRIG tesring requirements.
2062 * doc/invoke.texi (Overall Options): Mention BRIG.
2063 * doc/standards.texi (Standards): Doucment BRIG HSA version.
2065 2017-01-24 Richard Biener <rguenther@suse.de>
2067 PR translation/79208
2068 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
2070 2017-01-24 Martin Jambor <mjambor@suse.cz>
2073 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
2074 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
2077 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
2080 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
2081 casts from signed to unsigned really don't have a range.
2083 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
2085 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
2086 GMP_RNDx for compatiblity.
2088 2017-01-24 Martin Liska <mliska@suse.cz>
2091 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
2092 that would prevent us to call alloca with -1 as argument.
2094 2017-01-24 Jakub Jelinek <jakub@redhat.com>
2096 * dwarf2out.c (output_compilation_unit_header, output_file_names):
2097 Avoid -Wformat-security warning.
2099 2017-01-23 Andrew Pinski <apinski@cavium.com>
2101 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
2104 2017-01-23 Martin Sebor <msebor@redhat.com>
2107 * gimple-ssa-sprintf.c (warn_level): New global.
2108 (format_integer): Use it here and throughout the rest of the file.
2109 Use the same switch to compute sign as base.
2110 (maybe_warn): New function.
2111 (format_directive): Factor out warnings into maybe_warn.
2112 Add debugging output. Use warn_level.
2113 (add_bytes): Use warn_level.
2114 (pass_sprintf_length::compute_format_length): Add debugging output.
2115 (try_substitute_return_value): Same.
2116 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
2119 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
2120 (struct fmtresult, format_integer, format_floating): Adjust.
2121 (fmtresult::fmtresult): Set max correctly in two argument ctor.
2122 (get_string_length, format_string,format_directive): Same.
2123 (pass_sprintf_length::compute_format_length): Same.
2124 (try_substitute_return_value): Simplify slightly.
2127 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
2128 (fmtresult::operator+=): Outlined.
2129 (struct fmtresult): Add ctors.
2130 (struct conversion_spec): Rename...
2131 (struct directive): ...to this. Add and remove data members.
2132 (directive::set_width, directive::set_precision): New functions.
2133 (format_percent): Use fmtresult ctor.
2134 (get_width_and_precision): Remove.
2135 (format_integer): Make naming changes. Avoid computing width and
2137 (format_floating): Same. Adjust indentation.
2138 (format_character, format_none): New functions.
2139 (format_string): Moved character handling to format_character.
2140 (format_directive): Remove arguments, change return type.
2141 (parse_directive): New function.
2142 (pass_sprintf_length::compute_format_length): Move directive
2143 parsing to parse_directive.
2145 2017-01-23 Jakub Jelinek <jakub@redhat.com>
2147 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
2148 (assign_assembler_name_if_needed): ... this.
2149 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
2150 (assign_assembler_name_if_needed): ... this.
2151 (free_lang_data_in_cgraph): Adjust callers.
2152 * cgraphunit.c (cgraph_node::analyze): Likewise.
2153 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
2156 2017-01-23 Richard Biener <rguenther@suse.de>
2158 PR tree-optimization/79088
2159 PR tree-optimization/79188
2160 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
2161 resetting loop bounds after last path deletion. Reset loop
2162 bounds of the target loop, make code match the comments.
2163 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
2164 Make sure loops need no fixups.
2166 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
2168 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
2169 exponent support with double type for first argument.
2170 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
2171 type returned by __builtin_vec_extract_sig,
2172 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
2173 functions from "vector int" to "vector unsigned int" or from
2174 "vector long long int" to "vector unsigned long long int".
2175 Changed type returned by __builtin_vec_extract_exp,
2176 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
2177 functions from "vector int" to "vector unsigned int" or from
2178 "vector long long int" to "vector unsigned long long int".
2179 Changed return type of __builtin_vec_test_data_class,
2180 __builtin_vec_test_data_class_sp, and
2181 __builtin_vec_test_data_class_dp from "vector int" to
2182 "vector bool int" or from "vector long long int" to "vector bool
2183 long long int" and changed second argument type from "unsigned
2184 int" to "int". Added new overloaded function forms "vector float
2185 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
2186 "vector float __builtin_vec_insert_exp_sp (vector float, vector
2187 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
2188 double, vector unsigned long long int)" and "vector double
2189 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
2190 long int)". Changed return type of
2191 __builtin_scalar_test_data_class and
2192 __builtin_scalar_test_data_class_sp and
2193 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
2194 int" and changed second argument from "unsigned int" to "int".
2195 Changed type returned by __builtin_scalar_test_neg,
2196 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
2197 from "int" to "bool int". Added new overloaded function form
2198 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
2199 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
2200 exponent double-precision with floating point first argument.
2201 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
2202 documentation of scalar_test_data_class, scalar_test_neg,
2203 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
2204 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
2205 vec_test_data_class built-in functions to reflect refinements in
2206 their type signatures.
2208 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
2210 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
2212 (aarch64_elf_asm_destructor): Likewise.
2214 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
2216 PR rtl-optimization/78634
2217 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
2218 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
2219 * ifcvt.c (noce_try_cmove): Add missing cost check.
2221 PR rtl-optimization/71724
2222 * combine.c (if_then_else_cond): Look for situations where it is
2223 beneficial to undo the work of one of the recursive calls.
2225 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2227 PR tree-optimization/70754
2228 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2229 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2230 combined stmt before it if not NULL.
2231 (combine_chains): Process refs reversely and compute dominance point
2234 2017-01-23 Martin Liska <mliska@suse.cz>
2236 PR tree-optimization/79196
2237 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
2238 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
2240 (strlen_optimize_stmt): Call the renamed function.
2242 2017-01-23 Michael Matz <matz@suse.de>
2244 PR tree-optimization/78384
2245 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
2247 2017-01-23 Richard Biener <rguenther@suse.de>
2249 PR tree-optimization/79186
2250 * tree-vrp.c (register_new_assert_for): Make sure we've seen
2251 both incoming edges before moving an assert.
2253 2017-01-23 Martin Jambor <mjambor@suse.cz>
2255 * ipa-prop.c (load_from_param_1): Removed.
2256 (load_from_unmodified_param): Bits from load_from_param_1 put back
2258 (load_from_param): Removed.
2259 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
2260 with stmt. Reverted back to use of load_from_unmodified_param.
2262 2017-01-23 Martin Jambor <mjambor@suse.cz>
2265 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
2266 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
2267 field a pointer to garbage collected vector, mark lattices and
2268 ipcp_orig_node with GTY((skip)).
2269 (ipa_get_param_count): Adjust to descriptors being a pointer.
2270 (ipa_get_param): Likewise.
2271 (ipa_get_type): Likewise.
2272 (ipa_get_param_move_cost): Likewise.
2273 (ipa_set_param_used): Likewise.
2274 (ipa_get_controlled_uses): Likewise.
2275 (ipa_set_controlled_uses): Likewise.
2276 (ipa_is_param_used): Likewise.
2277 (ipa_node_params_t): Move into garbage collector. New methods insert
2279 (ipa_node_params_sum): Annotate wth GTY(()).
2280 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
2282 (ipa_load_from_parm_agg): Adjust declaration.
2283 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
2284 * ipa-profile.c (ipa_profile): Likewise.
2285 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
2286 (ipa_populate_param_decls): Make descriptors parameter garbage
2288 (ipa_dump_param): Adjust to descriptors being a pointer.
2289 (ipa_alloc_node_params): Likewise.
2290 (ipa_initialize_node_params): Likewise.
2291 (load_from_param_1): Make descriptors parameter garbage collected.
2292 (load_from_unmodified_param): Likewise.
2293 (load_from_param): Likewise.
2294 (ipa_load_from_parm_agg): Likewise.
2295 (ipa_node_params::~ipa_node_params): Removed.
2296 (ipa_free_all_node_params): Remove call to delete operator.
2297 (ipa_node_params_t::insert): New.
2298 (ipa_node_params_t::remove): Likewise.
2299 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
2300 copy known_csts and known_contexts vectors.
2301 (ipa_read_node_info): Adjust to descriptors being a pointer.
2302 (ipcp_modif_dom_walker): Make m_descriptors field garbage
2304 (ipcp_transform_function): Make descriptors variable garbage
2307 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
2309 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
2310 * config/i386/avx512dqintrin.h: Ditto.
2311 * config/i386/avx512fintrin.h: Ditto.
2312 * gcc/config/i386/i386.c: Handle new builtins.
2313 * config/i386/i386-builtin.def: Add new builtins.
2314 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
2315 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
2317 2017-01-23 Jakub Jelinek <jakub@redhat.com>
2318 Martin Liska <mliska@suse.cz>
2320 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
2321 * asan.c (asan_expand_poison_ifn): Support stores and use
2322 appropriate ASAN report function.
2323 * internal-fn.c (expand_ASAN_POISON_USE): New function.
2324 * internal-fn.def (ASAN_POISON_USE): Declare.
2325 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
2326 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
2327 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
2328 ASAN_POISON calls w/o LHS.
2329 * tree-ssa.c (execute_update_addresses_taken): Create clobber
2330 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
2331 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
2332 * gimplify.c (asan_poison_variables): Add attribute
2333 use_after_scope_memory to variables that really needs to live
2335 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
2336 having the attribute.
2338 2017-01-23 Martin Liska <mliska@suse.cz>
2340 * asan.c (create_asan_shadow_var): New function.
2341 (asan_expand_poison_ifn): Likewise.
2342 * asan.h (asan_expand_poison_ifn): New declaration.
2343 * internal-fn.c (expand_ASAN_POISON): Likewise.
2344 * internal-fn.def (ASAN_POISON): New builtin.
2345 * sanopt.c (pass_sanopt::execute): Expand
2346 asan_expand_poison_ifn.
2347 * tree-inline.c (copy_decl_for_dup_finish): Make function
2349 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
2350 * tree-ssa.c (is_asan_mark_p): New function.
2351 (execute_update_addresses_taken): Rewrite local variables
2352 (identified just by use-after-scope as addressable) into SSA.
2354 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2356 * doc/install.texi (Specific): opensource.apple.com uses https
2357 now. Remove trailing slash.
2359 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2361 * README.Portability: Remove note on an Irix compatibility issue.
2363 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
2365 * gcov.c (INCLUDE_ALGORITHM): Define.
2366 (INCLUDE_VECTOR): Define.
2367 No longer include <vector> and <algorithm> directly.
2369 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
2371 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
2373 * doc/invoke.texi (Code Gen Options): Ditto.
2375 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
2378 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
2380 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
2382 rtl-optimization/79125
2383 * cprop.c (local_cprop_pass): Handle cases where we make an
2386 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
2390 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
2391 read from, for big endian.
2393 2017-01-20 Jiong Wang <jiong.wang@arm.com>
2395 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
2396 register pauth builtins for LP64 only.
2398 2017-01-20 Marek Polacek <polacek@redhat.com>
2401 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
2404 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2406 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
2408 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
2409 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
2410 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
2412 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2415 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
2416 in big-endian mode when they are not a single duplicated value.
2418 2017-01-20 Richard Biener <rguenther@suse.de>
2420 * BASE-VER: Bump to 7.0.1.
2422 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2424 * omp-low.c (omplow_simd_context): New struct. Use it...
2425 (lower_rec_simd_input_clauses): ...here and...
2426 (lower_rec_input_clauses): ...here to hold common data. Adjust all
2427 references to idx, lane, max_vf, is_simt.
2429 2017-01-20 Graham Markall <graham.markall@embecosm.com>
2431 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
2434 2017-01-20 Martin Jambor <mjambor@suse.cz>
2436 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
2437 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
2439 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
2440 (GTFILES): Rename hsa.c to hsa-common.c.
2441 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
2442 * hsa-dump.c: Likewise.
2443 * hsa-gen.c: Likewise.
2444 * hsa-regalloc.c: Likewise.
2445 * ipa-hsa.c: Likewise.
2446 * omp-expand.c: Likewise.
2447 * omp-low.c: Likewise.
2448 * toplev.c: Likewise.
2450 2017-01-20 Marek Polacek <polacek@redhat.com>
2453 * doc/invoke.texi: Document -Wduplicated-branches.
2454 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
2455 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
2456 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
2457 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
2458 return 0 only when not OEP_LEXICOGRAPHIC.
2459 (fold_build_cleanup_point_expr): Use the expression
2460 location when building CLEANUP_POINT_EXPR.
2461 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
2462 * tree.c (add_expr): Handle error_mark_node.
2464 2017-01-20 Martin Liska <mliska@suse.cz>
2467 * tree-profile.c (init_ic_make_global_vars): Do not call
2469 (gimple_init_gcov_profiler): Likewise.
2471 2017-01-20 Martin Liska <mliska@suse.cz>
2474 * cgraph.h (maybe_create_reference): Remove argument and
2476 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
2478 * ipa-cp.c (create_specialized_node): Likewise.
2479 * symtab.c (symtab_node::maybe_create_reference): Handle
2480 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
2482 2017-01-20 Martin Liska <mliska@suse.cz>
2484 * read-rtl-function.c (function_reader::create_function): Use
2485 build_decl instread of build_decl_stat.
2487 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
2489 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
2490 * config/i386/avx512dqintrin.h: Ditto.
2491 * config/i386/avx512fintrin.h: Ditto.
2492 * config/i386/i386-builtin-types.def: Add new types.
2493 * gcc/config/i386/i386.c: Handle new types.
2494 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
2495 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
2496 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
2497 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
2498 (__builtin_ia32_kshiftridi): New.
2499 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
2501 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
2505 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
2506 define to rs6000_init_stack_protect_guard.
2507 (rs6000_init_stack_protect_guard): New function.
2509 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2510 Yunqiang Su <yunqiang.su@imgtec.com>
2512 * config.gcc (supported_defaults): Add madd4.
2513 (with_madd4): Add validation.
2514 (all_defaults): Add madd4.
2515 * config/mips/mips.opt (mmadd4): New option.
2516 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2518 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2520 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
2521 (ISA_HAS_FUSED_MADD4): Likewise.
2522 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
2523 * gcc/doc/install.texi (--with-madd4): Document the new option.
2525 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2527 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
2528 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
2529 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
2530 (aarch64_init_pauth_hint_builtins): New.
2531 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
2532 (aarch64_expand_builtin): Expand new builtins.
2534 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2536 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
2537 * combine-stack-adj.c (no_unhandled_cfa): Handle
2538 REG_CFA_TOGGLE_RA_MANGLE.
2539 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
2540 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
2541 info for return address signing.
2542 (aarch64_expand_epilogue): Likewise.
2544 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2546 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
2547 * config/aarch64/aarch64-protos.h
2548 (aarch64_return_address_signing_enabled): New declaration.
2549 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
2551 (aarch64_expand_prologue): Sign return address before it's pushed onto
2553 (aarch64_expand_epilogue): Authenticate return address fetched from
2555 (aarch64_override_options): Sanity check for ILP32 and ISA level.
2556 (aarch64_attributes): New function attributes for "sign-return-address".
2557 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
2558 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
2559 ("*do_return"): Generate combined instructions according to key index.
2560 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
2561 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
2563 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
2564 * config/aarch64/aarch64.opt (msign-return-address=): New.
2565 * doc/extend.texi (AArch64 Function Attributes): Documents
2566 "sign-return-address=".
2567 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
2569 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2571 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
2572 overall option summary.
2574 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2576 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
2577 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
2578 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
2579 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
2581 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
2583 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
2584 -mpower9-minmax by default for -mcpu=power9.
2585 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
2586 128-bit floating point.
2588 2017-01-20 Alan Modra <amodra@gmail.com>
2590 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
2591 optimizing for size.
2593 2017-01-20 Alan Modra <amodra@gmail.com>
2596 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
2597 for strcmp and strncmp from corresponding builtin decl.
2599 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2601 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
2602 instead of i386/rtems-64.h.
2603 * config/i386/rtems-64.h: Remove.
2605 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2609 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
2611 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
2613 2017-01-19 Tamar Christina <tamar.christina@arm.com>
2615 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2616 Change int to HOST_WIDE_INT.
2617 * config/aarch64/aarch64-protos.h
2618 (aarch64_simd_gen_const_vector_dup): Likewise.
2619 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
2621 2017-01-19 David Malcolm <dmalcolm@redhat.com>
2623 * langhooks-def.h (lhd_type_for_size): New decl.
2624 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
2625 * langhooks.c (lhd_type_for_size): New function, taken from
2628 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
2630 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
2631 define_bypass for CR latency.
2632 (power9-cracked-alu): Update bypass latency and remove power9-branch.
2633 (power9-alu2): Add define_bypass for CR latency.
2635 (power9-mul): Update insn latency.
2636 (power9-mul-compare): Update insn latency, bypass latency and remove
2639 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2641 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
2643 * config/aarch64/aarch64.md
2644 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
2645 aarch64_nopcrelative_literal_loads.
2646 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
2648 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
2650 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
2652 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
2654 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
2657 * config.gcc (supported_defaults): Add lxc1-sxc1.
2658 (with_lxc1_sxc1): Add validation.
2659 (all_defaults): Add lxc1-sxc1.
2660 * config/mips/mips.opt (mlxc1-sxc1): New option.
2661 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2663 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2664 __mips_no_lxc1_sxc1.
2665 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
2666 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
2667 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
2669 2017-01-19 Richard Biener <rguenther@suse.de>
2671 PR tree-optimization/72488
2672 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
2673 sure to restore SSA info.
2674 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
2676 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
2678 PR rtl-optimization/79121
2679 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
2680 of the inner type when shifting an extended value.
2682 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
2685 * symtab.c (symtab_node::equal_address_to): Fix comparing of
2686 interposable aliases.
2688 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
2691 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
2692 Use the evmergelohi instruction.
2693 (mov_si<mode>_e500_subreg4_2_le): Likewise.
2694 (mov_sitf_e500_subreg8_2_be): Likewise.
2695 (mov_sitf_e500_subreg12_2_le): Likewise.
2696 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
2697 (mov_si<mode>_e500_subreg4_2_be): Likewise.
2698 (mov_sitf_e500_subreg8_2_le): Likewise.
2699 (mov_sitf_e500_subreg12_2_be): Likewise.
2701 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2703 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
2704 attribute from vecsimple to vecperm.
2705 (altivec_vbpermq2): Likewise.
2707 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2710 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
2712 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2713 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
2714 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
2715 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
2716 case where N arg is SIZE_MAX.
2717 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
2718 (cmpstrsi): Add pattern.
2720 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2722 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2723 __builtin_vec_revb builtins.
2724 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
2725 built-in functions to support generation of the ISA 3.0 XXBR<x>
2726 vector byte reverse instructions.
2727 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
2728 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
2729 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
2730 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
2731 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
2732 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
2733 (P9V_BUILTIN_VEC_REVB): Likewise.
2734 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
2735 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
2736 (p9_xxbrq_v16qi): Likewise.
2737 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
2738 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
2739 (p9_xxbrh_v8hi): Likewise.
2740 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
2741 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
2742 vec_revb built-in functions.
2744 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
2746 PR rtl-optimization/78952
2747 * config/i386/i386.md (any_extract): New code iterator.
2748 (*insvqi_2): Use any_extract for source operand.
2749 (*insvqi_3): Use any_shiftrt for source operand.
2751 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
2753 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
2755 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
2757 2017-01-18 Matthias Klose <doko@ubuntu.com>
2759 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
2761 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2763 * config/rs6000/altivec.h (vec_bperm): Change #define.
2764 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
2765 (altivec_vbpermq2): New define_insn.
2766 (altivec_vbpermd): Likewise.
2767 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
2769 (VBPERMD): Likewise.
2770 (VBPERM): New polymorphic function interface.
2771 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
2772 Add entries for P9V_BUILTIN_VEC_VBPERM.
2773 * doc/extend.texi: Add interfaces for vec_bperm.
2775 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2777 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
2778 first letter of error messages.
2779 (s390_resolve_overloaded_builtin): Likewise.
2780 * config/s390/s390.c (s390_expand_builtin): Likewise.
2781 (s390_invalid_arg_for_unprototyped_fn): Likewise.
2782 (s390_valid_target_attribute_inner_p): Likewise.
2783 * config/s390/s390.md ("tabort"): Likewise.
2785 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
2787 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
2788 (ISA_AVOID_DIV_HILO): New macro.
2789 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
2790 (ISA_HAS_DDIV): Likewise.
2792 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2794 * doc/invoke.texi (fabi-version): Correct number of occurrences.
2796 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2798 * doc/invoke.texi (fabi-version): Spelling fix.
2800 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2803 * doc/invoke.texi (fabi-version): Mention mangling fix for
2806 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2809 * doc/invoke.texi (fabi-version): Document discriminator mangling.
2811 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
2814 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
2815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
2817 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
2819 (stack_protect_test): Ditto.
2820 * config/rs6000/rs6000.opt (mstack-protector-guard=,
2821 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
2823 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
2824 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
2825 -mstack-protector-guard-offset=.
2826 (RS/6000 and PowerPC Options): Ditto.
2828 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2830 * config/i386/i386.h (MASK_CLASS_P): New define.
2831 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
2832 there are no registers from different register sets also when
2833 mask registers are used. Update function comment.
2834 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
2835 to (*k/*r) and (*k/*km) alternatives.
2837 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
2839 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
2840 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
2841 (EH_RETURN_HANDLER_RTX): New define.
2842 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
2843 Force frame pointer in EH return functions.
2844 (aarch64_expand_epilogue): Add barrier for eh_return.
2845 (aarch64_final_eh_return_addr): Remove.
2846 (aarch64_eh_return_handler_rtx): New function.
2847 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
2849 (aarch64_eh_return_handler_rtx): New prototype.
2851 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2853 * config/rs6000/altivec.h (vec_rlmi): New #define.
2854 (vec_vrlnm): Likewise.
2855 (vec_rlnm): Likewise.
2856 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
2857 (UNSPEC_VRLNM): Likewise.
2858 (VIlong): New mode iterator.
2859 (altivec_vrl<VI_char>mi): New define_insn.
2860 (altivec_vrl<VI_char>nm): Likewise.
2861 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
2864 (RLNM): New polymorphic function entry.
2865 (VRLWMI): New monomorphic function entry.
2867 (RLMI): New polymorphic function entry.
2868 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
2869 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
2870 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
2873 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2876 * dwarf2out.c (field_byte_offset): Restore the
2877 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
2878 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
2879 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
2882 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2885 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
2887 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2890 * dwarf2out.c (add_data_member_location_attribute): For constant
2891 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
2892 instead of DW_AT_data_member_location, DW_AT_bit_offset and
2893 DW_AT_byte_size attributes.
2895 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2897 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
2898 after forcing to constant memory when the code model is medium.
2900 2017-01-17 Julia Koval <julia.koval@intel.com>
2903 * config/i386/avx512fintrin.h
2904 (_mm512_i32gather_ps): Change __addr type to void const*.
2905 (_mm512_mask_i32gather_ps): Ditto.
2906 (_mm512_i32gather_pd): Ditto.
2907 (_mm512_mask_i32gather_pd): Ditto.
2908 (_mm512_i64gather_ps): Ditto.
2909 (_mm512_mask_i64gather_ps): Ditto.
2910 (_mm512_i64gather_pd): Ditto.
2911 (_mm512_mask_i64gather_pd): Ditto.
2912 (_mm512_i32gather_epi32): Ditto.
2913 (_mm512_mask_i32gather_epi32): Ditto.
2914 (_mm512_i32gather_epi64): Ditto.
2915 (_mm512_mask_i32gather_epi64): Ditto.
2916 (_mm512_i64gather_epi32): Ditto.
2917 (_mm512_mask_i64gather_epi32): Ditto.
2918 (_mm512_i64gather_epi64): Ditto.
2919 (_mm512_mask_i64gather_epi64): Ditto.
2920 (_mm512_i32scatter_ps): Change __addr type to void*.
2921 (_mm512_mask_i32scatter_ps): Ditto.
2922 (_mm512_i32scatter_pd): Ditto.
2923 (_mm512_mask_i32scatter_pd): Ditto.
2924 (_mm512_i64scatter_ps): Ditto.
2925 (_mm512_mask_i64scatter_ps): Ditto.
2926 (_mm512_i64scatter_pd): Ditto.
2927 (_mm512_mask_i64scatter_pd): Ditto.
2928 (_mm512_i32scatter_epi32): Ditto.
2929 (_mm512_mask_i32scatter_epi32): Ditto.
2930 (_mm512_i32scatter_epi64): Ditto.
2931 (_mm512_mask_i32scatter_epi64): Ditto.
2932 (_mm512_i64scatter_epi32): Ditto.
2933 (_mm512_mask_i64scatter_epi32): Ditto.
2934 (_mm512_i64scatter_epi64): Ditto.
2935 (_mm512_mask_i64scatter_epi64): Ditto.
2936 * config/i386/avx512pfintrin.h
2937 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
2938 (_mm512_mask_prefetch_i32gather_ps): Ditto.
2939 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2940 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2941 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
2942 (_mm512_prefetch_i32scatter_ps): Ditto.
2943 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2944 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2945 (_mm512_prefetch_i64scatter_pd): Ditto.
2946 (_mm512_prefetch_i64scatter_ps): Ditto.
2947 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2948 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2949 * config/i386/avx512vlintrin.h
2950 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
2951 (_mm_mmask_i32gather_ps): Ditto.
2952 (_mm256_mmask_i32gather_pd): Ditto.
2953 (_mm_mmask_i32gather_pd): Ditto.
2954 (_mm256_mmask_i64gather_ps): Ditto.
2955 (_mm_mmask_i64gather_ps): Ditto.
2956 (_mm256_mmask_i64gather_pd): Ditto.
2957 (_mm_mmask_i64gather_pd): Ditto.
2958 (_mm256_mmask_i32gather_epi32): Ditto.
2959 (_mm_mmask_i32gather_epi32): Ditto.
2960 (_mm256_mmask_i32gather_epi64): Ditto.
2961 (_mm_mmask_i32gather_epi64): Ditto.
2962 (_mm256_mmask_i64gather_epi32): Ditto.
2963 (_mm_mmask_i64gather_epi32): Ditto.
2964 (_mm256_mmask_i64gather_epi64): Ditto.
2965 (_mm_mmask_i64gather_epi64): Ditto.
2966 (_mm256_i32scatter_ps): Change __addr type to void*.
2967 (_mm256_mask_i32scatter_ps): Ditto.
2968 (_mm_i32scatter_ps): Ditto.
2969 (_mm_mask_i32scatter_ps): Ditto.
2970 (_mm256_i32scatter_pd): Ditto.
2971 (_mm256_mask_i32scatter_pd): Ditto.
2972 (_mm_i32scatter_pd): Ditto.
2973 (_mm_mask_i32scatter_pd): Ditto.
2974 (_mm256_i64scatter_ps): Ditto.
2975 (_mm256_mask_i64scatter_ps): Ditto.
2976 (_mm_i64scatter_ps): Ditto.
2977 (_mm_mask_i64scatter_ps): Ditto.
2978 (_mm256_i64scatter_pd): Ditto.
2979 (_mm256_mask_i64scatter_pd): Ditto.
2980 (_mm_i64scatter_pd): Ditto.
2981 (_mm_mask_i64scatter_pd): Ditto.
2982 (_mm256_i32scatter_epi32): Ditto.
2983 (_mm256_mask_i32scatter_epi32): Ditto.
2984 (_mm_i32scatter_epi32): Ditto.
2985 (_mm_mask_i32scatter_epi32): Ditto.
2986 (_mm256_i32scatter_epi64): Ditto.
2987 (_mm256_mask_i32scatter_epi64): Ditto.
2988 (_mm_i32scatter_epi64): Ditto.
2989 (_mm_mask_i32scatter_epi64): Ditto.
2990 (_mm256_i64scatter_epi32): Ditto.
2991 (_mm256_mask_i64scatter_epi32): Ditto.
2992 (_mm_i64scatter_epi32): Ditto.
2993 (_mm_mask_i64scatter_epi32): Ditto.
2994 (_mm256_i64scatter_epi64): Ditto.
2995 (_mm256_mask_i64scatter_epi64): Ditto.
2996 (_mm_i64scatter_epi64): Ditto.
2997 (_mm_mask_i64scatter_epi64): Ditto.
2998 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
2999 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
3000 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
3001 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
3002 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
3003 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
3004 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
3005 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
3006 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
3007 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
3008 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
3009 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
3010 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
3011 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
3012 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
3013 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
3014 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
3015 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
3016 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
3017 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
3018 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
3019 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
3020 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
3021 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
3022 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
3023 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
3024 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
3025 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
3026 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
3027 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
3028 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
3029 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
3030 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
3031 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
3032 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
3033 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
3034 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
3035 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
3036 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
3037 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
3038 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
3039 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
3040 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
3041 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
3042 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
3043 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
3044 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
3045 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
3046 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
3047 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
3048 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
3049 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
3050 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
3051 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
3052 definitions accordingly.
3054 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
3055 Kuan-Lin Chen <kuanlinchentw@gmail.com>
3058 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
3061 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
3064 * ira-conflicts.c (ira_build_conflicts): Update total conflict
3065 hard regs for inner regno.
3067 2017-01-17 Martin Liska <mliska@suse.cz>
3070 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
3071 assumption and add comment.
3073 2017-01-17 Nathan Sidwell <nathan@acm.org>
3075 * ipa-visibility.c (localize_node): New function, broken out of ...
3076 (function_and_variable_visibility): ... here. Call it.
3078 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3081 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
3082 correctly set frequency of oudgoing edge.
3083 (duplicate_thread_path): Fix profile updating.
3085 2017-01-17 Jakub Jelinek <jakub@redhat.com>
3088 * configure.ac: Add GCC_BASE_VER.
3089 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
3090 version from BASE-VER file.
3091 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
3092 (gcc.o): Depend on $(BASEVER).
3093 * common.opt (dumpfullversion): New option.
3094 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
3095 * doc/invoke.texi: Document -dumpfullversion.
3096 * doc/install.texi: Document --with-gcc-major-version-only.
3097 * configure: Regenerated.
3099 2017-01-17 Richard Biener <rguenther@suse.de>
3101 PR tree-optimization/71433
3102 * tree-vrp.c (register_new_assert_for): Merge same asserts
3103 on all incoming edges.
3104 (process_assert_insertions_for): Handle insertions at the
3107 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
3109 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
3110 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
3112 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
3115 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
3118 2017-01-17 Alan Modra <amodra@gmail.com>
3121 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
3122 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
3123 symbolic stack limit when pic.
3125 2017-01-16 Martin Sebor <msebor@redhat.com>
3127 PR tree-optimization/78608
3128 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
3130 2017-01-16 Jeff Law <law@redhat.com>
3133 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
3134 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
3135 for several include directories that may be relative to sysroot.
3136 * config/i386/x-mingw32 (gplus_includedir): Define.
3137 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
3138 (native_system_includedir): Likewise.
3139 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
3140 override if TARGET_SYSTEM_ROOT is defined.
3141 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
3143 PR tree-optimization/79090
3144 PR tree-optimization/33562
3145 PR tree-optimization/61912
3146 PR tree-optimization/77485
3147 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
3148 and computed trims into the dump file.
3150 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
3152 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
3154 2017-01-16 Jakub Jelinek <jakub@redhat.com>
3157 * gimplify.c (gimplify_init_constructor): If want_value and
3158 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
3162 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
3163 sequence. Formatting fixes.
3164 (doloop_optimize): Formatting fixes.
3167 * gcc.c (debug_level_greater_than_spec_func): New function.
3168 (static_spec_functions): Add debug-level-gt spec function.
3169 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
3171 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
3172 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
3173 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
3174 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
3175 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
3176 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
3178 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
3180 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
3181 QImode fixups to general and mask registers only.
3183 2017-01-16 Carl Love <cel@us.ibm.com>
3185 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
3186 for built-in functions
3187 vector signed char vec_nabs (vector signed char)
3188 vector signed short vec_nabs (vector signed short)
3189 vector signed int vec_nabs (vector signed int)
3190 vector signed long long vec_nabs (vector signed long long)
3191 vector float vec_nabs (vector float)
3192 vector double vec_nabs (vector double)
3193 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
3195 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
3196 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
3197 * doc/extend.texi: Update the documentation file for the new built-in
3200 2017-01-16 Martin Sebor <msebor@redhat.com>
3202 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
3205 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3207 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
3208 UNSPEC_VSX__XXSPLTD to require special splat handling.
3210 2017-01-16 David Malcolm <dmalcolm@redhat.com>
3213 * system.h: Poison strndup.
3215 2017-01-16 Alan Modra <amodra@gmail.com>
3218 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
3221 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
3223 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
3225 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
3227 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
3228 call recog here. Assert that INSN_CODE (insn) is non-negative.
3230 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
3233 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
3235 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
3236 in the currently scheduled RTL fragment.
3238 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
3240 PR rtl-optimization/78751
3241 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
3244 2017-01-14 Jeff Law <law@redhat.com>
3246 PR tree-optimization/79090
3247 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
3248 variable length stores.
3249 (compute_trims): Delete dead assignment to *trim_tail.
3250 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
3253 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
3255 PR rtl-optimization/78626
3256 PR rtl-optimization/78727
3257 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
3258 of a block, and split such blocks after everything else is finished.
3260 2017-01-14 Alan Modra <amodra@gmail.com>
3263 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
3264 target legitimate_combined_insn.
3265 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
3266 (rs6000_legitimate_combined_insn): New function.
3267 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
3269 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
3270 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
3271 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
3273 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
3275 * doc/frontends.texi (G++ and GCC): Remove references to Java.
3277 2017-01-13 Jeff Law <law@redhat.com>
3279 PR tree-optimization/33562
3280 PR tree-optimization/61912
3281 PR tree-optimization/77485
3282 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
3284 (delete_dead_assignment): Likewise.
3285 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
3286 statement to delete_dead_call and delete_dead_assignment.
3288 2017-01-13 David Malcolm <dmalcolm@redhat.com>
3291 * substring-locations.c (format_warning_va): Strengthen case 1 so
3292 that both endpoints of the substring must be within the format
3293 range for just the substring to be printed.
3295 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
3297 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
3298 * config/i386/i386.c (ix86_target_string): Add missing options
3299 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
3300 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
3301 flags_other and ix86_target_other to flags2_other. Display unknown
3303 (ix86_valid_target_attribute_inner_p): Add missing options and
3304 reorder options by implied ISAs, as in ix86_target_string.
3306 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3308 * hash-table.h (hash_table::too_empty_p): New function.
3309 (hash_table::expand): Use it.
3310 (hash_table::traverse): Likewise.
3311 (hash_table::empty_slot): Use sizeof (value_type) instead of
3312 sizeof (PTR) to convert bytes to elements. Shrink the table
3313 if the current size is excessive for the current number of
3316 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3318 * ira-costs.c (record_reg_classes): Break from the inner loop
3319 early once alt_fail is known to be true. Update outer loop
3320 handling accordingly.
3322 2017-01-13 Jeff Law <law@redhat.com>
3324 * tree-ssa-dse.c (decrement_count): New function.
3325 (increment_start_addr, maybe_trim_memstar_call): Likewise.
3326 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
3327 when we know the partially dead statement is a mem* function.
3329 PR tree-optimization/61912
3330 PR tree-optimization/77485
3331 * tree-ssa-dse.c: Include expr.h.
3332 (maybe_trim_constructor_store): New function.
3333 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
3335 PR tree-optimization/33562
3336 PR tree-optimization/61912
3337 PR tree-optimization/77485
3338 * doc/invoke.texi: Document new dse-max-object-size param.
3339 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
3340 * tree-ssa-dse.c: Include params.h.
3341 (dse_store_status): New enum.
3342 (initialize_ao_ref_for_dse): New, partially extracted from
3344 (valid_ao_ref_for_dse, normalize_ref): New.
3345 (setup_live_bytes_from_ref, compute_trims): Likewise.
3346 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
3347 (maybe_trim_partially_dead_store): Likewise.
3348 (maybe_trim_complex_store): Likewise.
3349 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
3350 Track what bytes live from the original store. Return tri-state
3351 for dead, partially dead or live.
3352 (dse_dom_walker): Add constructor, destructor and new private members.
3353 (delete_dead_call, delete_dead_assignment): New extracted from
3355 (dse_optimize_stmt): Make a member of dse_dom_walker.
3356 Use initialize_ao_ref_for_dse.
3358 PR tree-optimization/33562
3359 PR tree-optimization/61912
3360 PR tree-optimization/77485
3361 * sbitmap.h (bitmap_count_bits): Prototype.
3362 (bitmap_clear_range, bitmap_set_range): Likewise.
3363 * sbitmap.c (bitmap_clear_range): New function.
3364 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
3366 2017-01-13 Martin Liska <mliska@suse.cz>
3369 * function.c (set_cfun): Add new argument force.
3370 * function.h (set_cfun): Likewise.
3371 * ipa-inline-transform.c (inline_call): Use the function when
3372 strict alising from is dropped for function we inline to.
3374 2017-01-13 Richard Biener <rguenther@suse.de>
3376 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
3377 for dumping GIMPLE INTEGER_CSTs.
3379 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3381 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
3382 to 201112L since C++17.
3384 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
3387 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
3388 if -fsanitize=kernel-address is present.
3390 2017-01-13 Richard Biener <rguenther@suse.de>
3392 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
3393 as _Literal ( type ) number in case usual suffixes do not
3394 preserve all information.
3396 2017-01-13 Richard Biener <rguenther@suse.de>
3398 PR tree-optimization/77283
3399 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
3400 and ssa-iterators.h.
3401 (is_feasible_trace): Implement a cost model based on joiner
3404 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3407 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
3408 char or short to __float128/_Float128 directly.
3410 2017-01-12 Martin Sebor <msebor@redhat.com>
3412 to -Wformat-overflow.
3413 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
3414 (min_bytes_remaining): Same.
3415 (get_string_length): Same.
3416 (format_string): Same.
3417 (format_directive): Same.
3419 (pass_sprintf_length::handle_gimple_call): Same.
3421 2017-01-12 Jakub Jelinek <jakub@redhat.com>
3423 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
3424 info.nowrite calls with no lhs that can't throw. Return bool
3425 whether gsi_remove has been called or not.
3426 (pass_sprintf_length::handle_gimple_call): Return bool whether
3427 try_substitute_return_value called gsi_remove. Formatting fix.
3428 (pass_sprintf_length::execute): Don't use gsi_remove if
3429 handle_gimple_call returned true.
3432 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
3433 be removed due to side-effects, don't remove following barrier nor
3434 turn the successor edge into fallthru edge.
3436 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3439 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
3440 element-reversing loads and stores as not swappable.
3442 2017-01-12 Nathan Sidwell <nathan@acm.org>
3443 Nicolai Stange <nicstange@gmail.com>
3445 * combine.c (try_combine): Don't ignore result of overlap checking
3446 loop. Combine overlap & asm check into single loop.
3448 2017-01-12 Richard Biener <rguenther@suse.de>
3450 * tree-pretty-print.c (dump_generic_node): Provide -gimple
3451 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
3453 2017-01-12 Richard Biener <rguenther@suse.de>
3455 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
3456 and TS_TARGET_OPTION directly derive from TS_BASE.
3457 * tree-core.h (tree_optimization_option): Derive from tree_base.
3458 (tree_target_option): Likewise.
3460 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
3462 * config/i386/i386.c (memory_address_length): Increase len
3463 only when rip_relative_addr_p returns false.
3465 2017-01-11 Julia Koval <julia.koval@intel.com>
3467 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
3468 (OPTION_MASK_ISA_SGX_SET): New.
3469 (ix86_handle_option): Handle OPT_msgx.
3470 * config.gcc: Added sgxintrin.h.
3471 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
3472 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
3473 * config/i386/i386.c (ix86_target_string): Add -msgx.
3475 (ix86_option_override_internal): Handle new options.
3476 (ix86_valid_target_attribute_inner_p): Add sgx.
3477 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
3478 * config/i386/i386.opt: Add msgx.
3479 * config/i386/sgxintrin.h: New file.
3480 * config/i386/x86intrin.h: Add sgxintrin.h.
3482 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3485 * fold-const.c (maybe_nonzero_address): Return 1 for function
3487 (tree_single_nonzero_warnv_p): Don't handle function local objects
3491 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
3494 2017-01-11 David Malcolm <dmalcolm@redhat.com>
3497 * opts.c: Include "spellcheck.h"
3498 (struct string_fragment): New struct.
3499 (struct edit_distance_traits<const string_fragment &>): New
3501 (get_closest_sanitizer_option): New function.
3502 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
3504 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3506 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
3508 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
3509 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
3510 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
3511 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
3512 for initial die_offset if dwarf_split_debug_info.
3513 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
3514 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
3516 (output_skeleton_debug_sections): Formatting fix. Use
3517 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
3518 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
3520 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
3522 * config/arm/cortex-a53.md: Add bypasses for
3524 (cortex_a53_r2f): Only use for transfers.
3525 (cortex_a53_f2r): Likewise.
3526 (cortex_a53_r2f_cvt): Add reservation for conversions.
3527 (cortex_a53_f2r_cvt): Likewise.
3529 2017-01-11 Tamar Christina <tamar.christina@arm.com>
3531 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
3532 to all inlined functions, change static to extern.
3534 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
3537 * config/arm/arm.c (legitimize_pic_address): Handle reference to
3539 (arm_assemble_integer): Likewise.
3541 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
3543 * config.gcc: Use new awk script to check CPU, FPU and architecture
3544 parameters for --with-... options.
3545 * config/arm/parsecpu.awk: New file
3546 * config/arm/arm-cpus.in: New file.
3547 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
3549 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
3551 * config/arm/t-arm: Update dependency rules.
3552 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
3553 of processing .def files.
3554 * config/arm/genopt.sh: Deleted.
3555 * config/arm/gentune.sh: Deleted.
3556 * config/arm/arm-cores.def: Deleted.
3557 * config/arm/arm-arches.def: Deleted.
3558 * config/arm/arm-fpus.def: Deleted.
3559 * config/arm/arm-tune.md: Regenerated.
3560 * config/arm/arm-tables.opt: Regenerated.
3561 * config/arm/arm-cpu.h: New generated file.
3562 * config/arm/arm-cpu-data.h: New generated file.
3563 * config/arm/arm-cpu-cdata.h: New generated file.
3565 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
3568 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
3570 (input_varpool_node): Unpack dynamically_initialized bit.
3572 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
3574 PR rtl-optimization/79032
3575 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
3576 the alignment of the adjusted memory reference against that of MODE,
3577 instead of the alignment of the original memory reference.
3579 2017-01-11 Martin Jambor <mjambor@suse.cz>
3581 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
3583 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
3584 decorated functions.
3586 2017-01-11 Richard Biener <rguenther@suse.de>
3588 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
3589 set range/nonnull info for PHI results. Do not set it on
3590 stmts marked for removal.
3592 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
3594 * expr.c (store_field): In the bitfield case, fetch the return value
3595 from the registers before applying a single big-endian adjustment.
3596 Always do a final load for a BLKmode value not larger than a word.
3598 2017-01-10 David Malcolm <dmalcolm@redhat.com>
3601 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3602 that we correctly handle column numbers greater than
3603 LINE_MAP_MAX_COLUMN_NUMBER.
3605 2017-01-10 Martin Sebor <msebor@redhat.com>
3608 * gimple-ssa-sprintf.c (get_destination_size): Call
3609 {init,fini}object_sizes.
3610 * tree-object-size.c (addr_object_size): Adjust.
3611 (pass_through_call): Adjust.
3612 (pass_object_sizes::execute): Adjust.
3613 * tree-object-size.h (fini_object_sizes): Declare.
3615 2017-01-10 Martin Sebor <msebor@redhat.com>
3617 PR tree-optimization/78775
3618 * builtins.c (get_size_range): Move...
3619 * calls.c: ...to here.
3620 (alloc_max_size): Accept zero argument.
3621 (operand_signed_p): Remove.
3622 (maybe_warn_alloc_args_overflow): Call get_size_range.
3623 * calls.h (get_size_range): Declare.
3625 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
3627 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
3628 from TI's devices.csv file as of September 2016.
3629 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
3631 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
3633 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
3634 * doc/invoke.texi: Likewise.
3635 * doc/md.texi: Likewise.
3636 * doc/objc.texi: Likewise.
3638 2017-01-10 Joshua Conner <joshconner@google.com>
3640 * config/arm/fuchsia-elf.h: New file.
3641 * config/fuchsia.h: New file.
3642 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
3643 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
3645 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
3647 2016-01-10 Richard Biener <rguenther@suse.de>
3649 PR tree-optimization/79034
3650 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
3651 Propagate out degenerate PHIs in the joiner.
3653 2017-01-10 Martin Liska <mliska@suse.cz>
3655 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
3656 (sort_congruence_classes_by_decl_uid): Likewise.
3657 (sort_congruence_class_groups_by_decl_uid): Likewise.
3658 (sem_item_optimizer::merge_classes): Sort class, groups in these
3659 classes and members in the groups by DECL_UID of declarations.
3660 This would make merge operations stable.
3662 2017-01-10 Martin Liska <mliska@suse.cz>
3664 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
3665 usage of m_classes_vec.
3666 (sem_item_optimizer::~sem_item_optimizer): Likewise.
3667 (sem_item_optimizer::get_group_by_hash): Likewise.
3668 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
3669 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
3670 (sem_item_optimizer::verify_classes): Likewise.
3671 (sem_item_optimizer::process_cong_reduction): Likewise.
3672 (sem_item_optimizer::dump_cong_classes): Likewise.
3673 (sem_item_optimizer::merge_classes): Likewise.
3674 * ipa-icf.h (congruence_class_hash): Rename from
3675 congruence_class_group_hash. Remove declaration of m_classes_vec.
3677 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
3679 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
3680 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
3681 * config.gcc: Add avx512vpopcntdqintrin.h.
3682 * config/i386/avx512vpopcntdqintrin.h: New.
3683 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
3684 * config/i386/i386-builtin-types.def: Add new types.
3685 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
3686 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
3687 __builtin_ia32_vpopcountq_v8di_mask): New.
3688 * config/i386/i386-c.c (ix86_target_macros_internal): Define
3689 __AVX512VPOPCNTDQ__.
3690 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
3691 (PTA_AVX512VPOPCNTDQ): Define.
3692 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
3693 TARGET_AVX512VPOPCNTDQ_P): Define.
3694 * config/i386/i386.opt: Add mavx512vpopcntdq.
3695 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
3696 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
3698 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3701 * predict.def (PRED_CALL): Set to 67.
3703 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
3705 * expr.c (store_field): In the bitfield case, if the value comes from
3706 a function call and is of an aggregate type returned in registers, do
3707 not modify the field mode; extract the value in all cases if the mode
3708 is BLKmode and the size is not larger than a word.
3710 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
3713 * config/i386/cpuid.h: Fix undefined behavior.
3715 2017-01-04 Jeff Law <law@redhat.com>
3717 PR tree-optimization/79007
3718 PR tree-optimization/67955
3719 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
3720 conservative for pt.null when flag_non_call_exceptions is on.
3722 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3724 PR translation/79019
3725 PR translation/79020
3726 * params.def (PARAM_INLINE_MIN_SPEEDUP,
3727 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
3728 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
3730 * config/avr/avr.opt (maccumulate-args): Likewise.
3731 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
3732 * common.opt (freport-bug): Likewise.
3733 * cif-code.def (CIF_FINAL_ERROR): Likewise.
3734 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
3735 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
3736 translatable string.
3737 * config/i386/i386.c (function_value_32): Likewise.
3738 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
3739 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
3741 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
3742 * common/config/msp430/msp430-common.c (msp430_handle_option):
3744 * symtab.c (symtab_node::verify_base): Likewise.
3745 * opts.c (set_debug_level): Likewise.
3746 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
3747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
3748 missing whitespace to translatable strings.
3749 * config/avr/avr.md (bswapsi2): Fix typo in comment.
3750 * config/sh/superh.h: Likewise.
3751 * config/i386/xopintrin.h: Likewise.
3752 * config/i386/znver1.md: Likewise.
3753 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
3754 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
3755 * double-int.h (struct double_int): Likewise.
3756 * double-int.c (div_and_round_double): Likewise.
3757 * wide-int.cc: Likewise.
3758 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
3759 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
3760 * cfgcleanup.c (crossjumps_occured): Renamed to ...
3761 (crossjumps_occurred): ... this.
3762 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
3765 PR tree-optimization/78899
3766 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
3767 returning bool return struct loop *, NULL for failure and the new
3769 (versionable_outer_loop_p): Don't version outer loop if it has
3770 dont_vectorized bit set.
3771 (tree_if_conversion): When versioning outer loop, ensure
3772 tree_if_conversion is performed also on the inner loop of the
3773 non-vectorizable outer loop copy.
3774 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
3775 LOOP_VECTORIZED in inner loop of the scalar outer loop and
3776 prevent vectorization of it.
3777 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
3778 the outer loop vectorization of the non-scalar version is attempted
3779 before vectorization of the inner loop in scalar version. If
3780 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
3781 vectorization of its inner loop.
3782 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
3783 has 2 inner loops, rename also on edges from bb whose single pred
3784 is outer_loop->header. Fix typo in function comment.
3786 2017-01-09 Martin Sebor <msebor@redhat.com>
3789 * asan.c (asan_emit_stack_protection): Increase local buffer size
3790 to avoid snprintf truncation warning.
3792 2017-01-09 Andrew Pinski <apinski@cavium.com>
3794 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
3795 to reference thunderx2t99 for the tuning structure
3796 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
3798 (thunderx2t99_extra_costs): This.
3799 * config/aarch64/aarch64-tune.md: Regenerate.
3800 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
3801 (vulcan_addrcost_table): This.
3802 (vulcan_regmove_cost): Rename to ...
3803 (thunderx2t99_regmove_cost): This.
3804 (vulcan_vector_cost): Rename to ...
3805 (thunderx2t99_vector_cost): this.
3806 (vulcan_branch_cost): Rename to ...
3807 (thunderx2t99_branch_cost): This.
3808 (vulcan_tunings): Rename to ...
3809 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
3810 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
3812 2017-01-09 Martin Jambor <mjambor@suse.cz>
3816 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
3817 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
3818 (propagate_vr_accross_jump_function): Use the above function for all
3819 value range computations for pass-through jump functions and type
3820 converasion from explicit value range values.
3821 (ipcp_propagate_stage): Do not attempt to deduce types of formal
3822 parameters from TYPE_ARG_TYPES.
3823 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
3824 (ipa_write_node_info): Stream type of the actual argument.
3825 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
3827 2017-01-09 Martin Liska <mliska@suse.cz>
3830 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
3831 (lookup_compiler): Do not show error message with have_E.
3833 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3835 PR tree-optimization/78938
3836 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
3837 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
3838 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
3839 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
3842 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3844 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
3847 2017-01-09 Richard Biener <rguenther@suse.de>
3849 PR tree-optimization/78997
3850 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
3851 name condition properly.
3853 2017-01-09 Richard Biener <rguenther@suse.de>
3856 * dwarf2out.c (is_cxx): New overload with context.
3857 (is_naming_typedef_decl): Use it.
3859 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3861 * invoke.texi (Option Summary): Correct spacing in option lists
3862 and add line breaks to fix over-long lines.
3864 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3868 * extend.texi (Common Variable Attributes): Add xref to GCC
3869 Internals manual to explain mode attribute keywords.
3871 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3874 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
3875 and Preprocessor Options.
3876 (Options for Linking): Document -pthread here....
3877 (RS/6000 and PowerPC Options): ...not here.
3878 (Solaris 2 Options): ...or here.
3879 * doc/cppopts.texi: Document -pthread.
3881 2017-01-08 Martin Sebor <msebor@redhat.com>
3884 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
3885 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
3886 New member functions.
3887 (format_directive): Used them.
3889 (pass_sprintf_length::handle_gimple_call): Same.
3890 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
3891 to avoid truncation for any argument.
3892 (extract_affine_mul): Same.
3893 * tree.c (get_file_function_name): Same.
3895 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3898 * predict.def (PRED_INDIR_CALL): Set to 86.
3900 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3902 PR preprocessor/54124
3903 * doc/cppopts.texi: Reformat -d subtable to list the full name
3904 of the options. Add cross-reference to the docs for the general
3905 compiler -d options.
3906 * doc/invoke.texi (Developer Options): Add cross-reference to the
3907 preprocessor-specific -d option documentation.
3909 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3911 PR preprocessor/13498
3912 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
3913 redudant material, and reflect new command-line options.
3914 (System Headers): Likewise.
3916 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3918 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
3919 -isystem, and -idirafter. Copy-edit.
3920 * doc/cppopts.texi: Copy-edit. Remove contradiction about
3921 default for -ftrack-macro-expansion. Delete obsolete and
3922 badly-formatted implementation details about -fdebug-cpp output.
3923 * doc/cppwarnopts.texi: Copy-edit.
3925 2017-01-07 David Malcolm <dmalcolm@redhat.com>
3928 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3929 that the transition from a max line width >= 1<<10 to narrower
3930 lines works correctly.
3932 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
3934 * doc/options.texi (PerFunction): New.
3935 * opt-functions.awk (switch_flags): Map both Optimization and
3936 PerFunction to CL_OPTIMIZATION.
3937 * opth-gen.awk: Test for PerFunction flag along with
3939 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
3940 it only when the latter is present. Skip those that don't in
3941 the hash function generator.
3942 * common.opt (fvar-tracking): Mark as PerFunction instead of
3944 (fvar-tracking-assignments): Likewise.
3945 (fvar-tracking-assignments-toggle): Likewise.
3946 (fvar-tracking-uninit): Likewise.
3948 2017-01-07 Jakub Jelinek <jakub@redhat.com>
3950 PR translation/79018
3951 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
3954 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
3957 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
3958 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
3959 TARGET_LONG_JUMP_TABLE_OFFSETS.
3960 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
3961 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
3962 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
3963 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
3964 * config/m68k/m68k.md (tablejump expander): Likewise.
3965 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
3966 TARGET_LONG_JUMP_TABLE_OFFSETS.
3967 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
3968 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
3970 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3971 David Holsgrove <david.holsgrove@xilinx.com>
3973 * common/config/microblaze/microblaze-common.c
3974 (TARGET_EXCEPT_UNWIND_INFO): Remove.
3975 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
3977 * config/microblaze/microblaze.c (microblaze_must_save_register)
3978 (microblaze_expand_epilogue, microblaze_return_addr): Handle
3980 (microblaze_eh_return): New function.
3981 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
3982 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
3983 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
3984 * config/microblaze/microblaze.md (eh_return): New pattern.
3986 2017-01-06 Jakub Jelinek <jakub@redhat.com>
3988 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
3989 GCC_DIAGNOSTIC_STRINGIFY): Define.
3991 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
3993 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3995 * config/arm/arm.md (<mcrr>): New.
3997 * config/arm/arm.c (arm_arch5te): New.
3998 (arm_option_override): Set arm_arch5te.
3999 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
4001 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
4002 (arm_mcrr_qualifiers): ... this. New.
4003 (MRRC_QUALIFIERS): Define to...
4004 (arm_mrrc_qualifiers): ... this. New.
4005 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
4007 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
4008 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
4009 (MRRCI, mrrc, MRRC): New.
4010 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
4011 VUNSPEC_MRRC2): New.
4013 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4015 * config/arm/arm.md (<mcr>): New.
4017 * config/arm/arm.c (arm_coproc_builtin_available): Add
4018 support for mcr, mrc, mcr2 and mrc2.
4019 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
4020 (arm_mcr_qualifiers): ... this. New.
4021 (MRC_QUALIFIERS): Define to ...
4022 (arm_mrc_qualifiers): ... this. New.
4023 (MCR_QUALIFIERS): Define to ...
4024 (arm_mcr_qualifiers): ... this. New.
4025 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
4027 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
4028 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
4029 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
4032 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4034 * config/arm/arm.md (*ldc): New.
4038 * config/arm/arm.c (arm_coproc_builtin_available): Add
4039 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
4040 (arm_coproc_ldc_stc_legitimate_address): New.
4041 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
4042 'qualifier_const_pointer'.
4043 (LDC_QUALIFIERS): Define to...
4044 (arm_ldc_qualifiers): ... this. New.
4045 (STC_QUALIFIERS): Define to...
4046 (arm_stc_qualifiers): ... this. New.
4047 * config/arm/arm-protos.h
4048 (arm_coproc_ldc_stc_legitimate_address): New.
4049 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
4050 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
4051 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
4052 stc2, stcl, stc2l): New.
4053 * config/arm/constraints.md (Uz): New.
4054 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
4055 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
4056 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
4057 VUNSPEC_STC2L): New.
4059 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4061 * config/arm/arm.md (<cdp>): New.
4062 * config/arm/arm.c (neon_const_bounds): Rename this ...
4063 (arm_const_bounds): ... this.
4064 (arm_coproc_builtin_available): New.
4065 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
4066 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
4067 (CDP_QUALIFIERS): Define to...
4068 (arm_cdp_qualifiers): ... this. New.
4070 (arm_expand_builtin_args): Add case for 6 arguments.
4071 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
4072 (arm_const_bounds): ... this.
4073 (arm_coproc_builtin_available): New.
4074 * config/arm/arm_acle.h (__arm_cdp): New.
4076 * config/arm/arm_acle_builtins.def (cdp): New.
4078 * config/arm/iterators.md (CDPI,CDP,cdp): New.
4079 * config/arm/neon.md: Rename all 'neon_const_bounds' to
4081 * config/arm/types.md (coproc): New.
4082 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
4083 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
4084 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
4085 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
4087 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4089 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
4090 (UBINOP_QUALIFIERS): New.
4092 (acle_builtin_data): New. Change comment.
4093 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
4094 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
4095 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
4096 arm_acle_builtins.def.
4097 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
4098 (arm_init_acle_builtins): New.
4099 (CRC32_BUILTIN): Remove.
4100 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
4101 crc32cb, crc32ch and crc32cw.
4102 (arm_init_crc32_builtins): Remove.
4103 (arm_init_builtins): Use arm_init_acle_builtins rather
4104 than arm_init_crc32_builtins.
4105 (arm_expand_acle_builtin): New.
4106 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
4107 * config/arm/arm_acle_builtins.def: New.
4109 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4111 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
4112 (arm_builtin_datum): ... this.
4113 (arm_init_neon_builtin): Rename to ...
4114 (arm_init_builtin): ... this. Add a new parameters PREFIX
4115 and USE_SIG_IN_NAME.
4116 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
4117 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
4118 'arm_builtin_datum'.
4119 (arm_init_vfp_builtins): Likewise.
4120 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
4121 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
4122 (arm_expand_neon_args): Rename to ...
4123 (arm_expand_builtin_args): ... this. Rename builtin_arg
4124 enum values and differentiate between ARG_BUILTIN_MEMORY
4125 and ARG_BUILTIN_NEON_MEMORY.
4126 (arm_expand_neon_builtin_1): Rename to ...
4127 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
4128 values, arm_expand_builtin_args and add bool parameter NEON.
4129 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
4130 (arm_expand_vfp_builtin): Likewise.
4131 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
4133 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4136 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
4137 * predict.c (tree_estimate_probability_bb): Reverse direction of
4138 polymorphic call predictor.
4140 2017-01-06 David Malcolm <dmalcolm@redhat.com>
4142 * passes.c (execute_one_pass): Split out pass-skipping logic into...
4143 (determine_pass_name_match): ...this new function and...
4144 (should_skip_pass_p): ...this new function.
4146 2017-01-06 Nathan Sidwell <nathan@acm.org>
4148 * ipa-visibility.c (function_and_variable_visibility): Reformat
4149 comments and long lines. Remove extrneous if.
4150 * symtab.c (symtab_node::make_decl_local): Fix code format.
4151 (symtab_node::set_section_for_node): Fix comment typo.
4153 2017-01-06 Martin Liska <mliska@suse.cz>
4156 * lra-constraints.c: Rename invariant to lra_invariant.
4157 * predict.c (set_even_probabilities): Initialize e to NULL.
4159 2017-01-05 Martin Sebor <msebor@redhat.com>
4161 PR tree-optimization/78910
4162 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
4163 (format_integer): Correct off-by-one error in the handling
4164 of precision with negative numbers in signed conversions..
4166 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
4168 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
4170 2017-01-05 Jakub Jelinek <jakub@redhat.com>
4172 PR tree-optimization/71016
4173 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
4174 factor_out_conditional_conversion. Formatting fix.
4175 (factor_out_conditional_conversion): Add cond_stmt argument.
4176 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
4177 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
4180 2017-01-05 David Malcolm <dmalcolm@redhat.com>
4182 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
4183 read-rtl-function.o, and selftest-rtl.o.
4184 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
4185 (selftest::aarch64_test_loading_full_dump): New function.
4186 (selftest::aarch64_run_selftests): New function.
4187 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
4188 selftest::aarch64_run_selftests.
4189 * config/i386/i386.c
4190 (selftest::ix86_test_loading_dump_fragment_1): New function.
4191 (selftest::ix86_test_loading_call_insn): New function.
4192 (selftest::ix86_test_loading_full_dump): New function.
4193 (selftest::ix86_test_loading_unspec): New function.
4194 (selftest::ix86_run_selftests): Call the new functions.
4195 * emit-rtl.c (maybe_set_max_label_num): New function.
4196 * emit-rtl.h (maybe_set_max_label_num): New decl.
4197 * function.c (instantiate_decls): Guard call to
4198 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
4199 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
4201 * gensupport.c (gen_reader::gen_reader): Pass "false"
4202 for new "compact" param of rtx_reader.
4203 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
4204 rather than an empty string for NULL strings.
4205 * read-md.c: Potentially include config.h rather than bconfig.h.
4206 Wrap include of errors.h with #ifdef GENERATOR_FILE.
4207 (have_error): New global, copied from errors.c.
4208 (md_reader::read_name): Rename to...
4209 (md_reader::read_name_1): ...this, adding "out_loc" param,
4210 and converting "missing name or number" to returning false, rather
4212 (md_reader::read_name): Reimplement in terms of read_name_1.
4213 (md_reader::read_name_or_nil): New function.
4214 (md_reader::read_string): Handle "(nil)" by returning NULL.
4215 (md_reader::md_reader): Add new param "compact".
4216 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
4217 (md_reader::read_file): New method.
4218 * read-md.h (md_reader::md_reader): Add new param "compact".
4219 (md_reader::read_file): New method.
4220 (md_reader::is_compact): New accessor.
4221 (md_reader::read_name): Convert return type from void to file_location.
4222 (md_reader::read_name_or_nil): New decl.
4223 (md_reader::read_name_1): New decl.
4224 (md_reader::m_compact): New field.
4225 (noop_reader::noop_reader): Pass "false" for new "compact" param
4227 (rtx_reader::rtx_reader): Add new "compact" param.
4228 (rtx_reader::read_rtx_operand): Make virtual and convert return
4229 type from void to rtx.
4230 (rtx_reader::read_until): New decl.
4231 (rtx_reader::handle_any_trailing_information): New virtual function.
4232 (rtx_reader::postprocess): New virtual function.
4233 (rtx_reader::finalize_string): New virtual function.
4234 (rtx_reader::m_in_call_function_usage): New field.
4235 (rtx_reader::m_reuse_rtx_by_id): New field.
4236 * read-rtl-function.c: New file.
4237 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
4238 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
4239 (selftest::verify_three_block_rtl_cfg): New decl.
4240 * read-rtl-function.h: New file.
4241 * read-rtl.c: Potentially include config.h rather than bconfig.h.
4242 For host, include function.h, memmodel.h, and emit-rtl.h.
4243 (one_time_initialization): New function.
4244 (struct compact_insn_name): New struct.
4245 (compact_insn_names): New array.
4246 (find_code): Handle insn codes in compact dumps.
4247 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
4248 (bind_subst_iter_and_attr): Likewise.
4249 (add_condition_to_string): Likewise.
4250 (add_condition_to_rtx): Likewise.
4251 (apply_attribute_uses): Likewise.
4252 (add_current_iterators): Likewise.
4253 (apply_iterators): Likewise.
4254 (initialize_iterators): Guard usage of apply_subst_iterator with
4255 #ifdef GENERATOR_FILE.
4256 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
4257 (md_reader::read_mapping): Likewise.
4258 (add_define_attr_for_define_subst): Likewise.
4259 (add_define_subst_attr): Likewise.
4260 (read_subst_mapping): Likewise.
4261 (check_code_iterator): Likewise.
4262 (rtx_reader::read_rtx): Likewise. Move one-time initialization
4264 (one_time_initialization): New function.
4265 (rtx_reader::read_until): New method.
4266 (read_flags): New function.
4267 (parse_reg_note_name): New function.
4268 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
4269 Handle reuse_rtx ids.
4270 Wrap iterator lookup within #ifdef GENERATOR_FILE.
4271 Add parsing support for RTL dumps, mirroring the special-cases in
4272 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
4273 values, and calling handle_any_trailing_information.
4274 (rtx_reader::read_rtx_operand): Convert return type from void
4275 to rtx, returning return_rtx. Handle case 'e'. Call
4276 finalize_string on XSTR and XTMPL fields.
4277 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
4278 "(nil)" values were omitted. Call the postprocess vfunc on the
4280 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
4281 class ctor. Initialize m_in_call_function_usage. Call
4282 one_time_initialization.
4283 * rtl-tests.c (selftest::test_uncond_jump): Call
4284 set_new_first_and_last_insn.
4285 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
4286 * selftest-rtl.c: New file.
4287 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
4288 (selftest::get_insn_by_uid): New decl.
4289 * selftest-run-tests.c (selftest::run_tests): Call
4290 read_rtl_function_c_tests.
4291 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
4292 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
4295 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
4297 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
4298 operands in a special way. Assert that pos+len <= mode precision.
4300 2017-01-05 Jakub Jelinek <jakub@redhat.com>
4302 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
4303 3 argument Alias with unlimited for the negative form.
4304 (fno-vect-cost-model): Removed.
4306 2017-01-05 Martin Liska <mliska@suse.cz>
4308 * hsa-gen.c (gen_hsa_divmod): New function.
4309 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
4311 2017-01-05 Martin Liska <mliska@suse.cz>
4314 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
4317 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4319 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
4320 small constant length operands.
4322 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4324 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
4325 between loop iterations.
4327 2017-01-05 Martin Liska <mliska@suse.cz>
4330 * gimplify.c (gimplify_decl_expr): Compare to
4331 asan_poisoned_variables instread of checking flags.
4332 (gimplify_target_expr): Likewise.
4333 (gimplify_expr): Likewise.
4334 (gimplify_function_tree): Conditionally initialize
4335 asan_poisoned_variables.
4337 2017-01-04 Jeff Law <law@redhat.com>
4339 PR tree-optimizatin/78812
4340 * rtl.h (contains_mem_rtx_p): Prototype.
4341 * ifcvt.c (containts_mem_rtx_p): Move from here to...
4342 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
4343 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
4344 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
4345 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
4347 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4349 * input.c (assert_char_at_range): Default-initialize actual_range.
4351 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4353 * df-scan.c (df_ref_create_structure): Make regno unsigned,
4354 to match the caller.
4356 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4358 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
4359 insns after final jump in test to emit dummy move.
4361 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4363 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
4364 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
4366 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4368 * multiple_target.c (create_dispatcher_calls): Init e_next.
4369 * tree-ssa-loop-split.c (split_loop): Init border.
4370 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
4373 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
4378 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
4379 (altivec_register_operand): Do not return true if the operand
4380 contains a SUBREG mixing SImode and SFmode.
4381 (vsx_register_operand): Likewise.
4382 (vsx_reg_sfsubreg_ok): New predicate.
4383 (vfloat_operand): Do not return true if the operand contains a
4384 SUBREG mixing SImode and SFmode.
4385 (vint_operand): Likewise.
4386 (vlogical_operand): Likewise.
4387 (gpc_reg_operand): Likewise.
4388 (int_reg_operand): Likewise.
4389 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
4390 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
4391 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
4393 (rs6000_emit_move_si_sf_subreg): New helper function.
4394 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
4395 fixup SUBREGs involving SImode and SFmode.
4396 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
4397 numbers for the new peephole2 optimization.
4398 (peephole2 for SFmode unions): New peephole2 to optimize cases in
4399 the GLIBC math library that do AND/IOR/XOR operations on single
4400 precision floating point.
4401 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
4402 target macros to say whether we need to avoid SUBREGs mixing
4404 (TARGET_ALLOW_SF_SUBREG): Likewise.
4405 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
4406 (UNSPEC_SI_FROM_SF): Likewise.
4407 (iorxor): Change spacing.
4408 (and_ior_xor): New iterator for AND, IOR, and XOR.
4409 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
4410 (movdi_from_sf_zero_ext): Likewise.
4411 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
4412 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
4413 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
4414 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
4415 (fms<mode>4): Likewise.
4416 (fnma<mode>4): Likewise.
4417 (fnms<mode>4): Likewise.
4418 (nfma<mode>4): Likewise.
4419 (nfms<mode>4): Likewise.
4421 2017-01-04 Marek Polacek <polacek@redhat.com>
4424 * doc/invoke.texi: Document -Wpointer-compare.
4426 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4428 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
4431 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
4432 descriptions for -gdwarf-5 and emit them as uleb128 instead of
4435 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
4438 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
4439 documentation of the powerpc_popcntb_ok attribute.
4440 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4441 code to issue warning messages if a requested CPU configuration is
4442 not supported by the binary (assembler and loader) toolchain.
4443 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
4444 made to define a built-in function that has been disabled.
4445 (paired_init_builtins): Add assertion to prevent ICE if attempt is
4446 made to define a built-in function that has been disabled.
4447 (altivec_init_builtins): Add comment explaining why definition
4448 of the DST built-in functions is not preceded by an assertion
4449 check. Add assertions to prevent ICE if attempts are made to
4450 define an altivec predicate or an abs* built-in function that has
4452 (htm_init_builtins): Add comment explaining why definition of the
4453 htm built-in functions is not preceded by an assertion check.
4455 2017-01-04 Jeff Law <law@redhat.com>
4457 PR tree-optimizatin/67955
4458 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
4459 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
4460 the points-to solution does not include pt_null. Use DECL_PT_UID
4463 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
4465 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
4466 Use gen_int_mode instead of gen_lopwart for const_int operands.
4468 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4470 PR tree-optimization/71563
4471 * match.pd: Simplify X << Y into X if Y is known to be 0 or
4472 out of range value - has low bits known to be zero.
4474 2017-01-04 Alan Modra <amodra@gmail.com>
4476 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
4477 * configure: Regenerate.
4478 * config.in: Regenerate.
4480 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4483 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
4484 a substring of the message, but strcmp with the whole message. Ifdef
4485 ENABLE_NLS, translate the message first using dgettext.
4487 2017-01-03 Jeff Law <law@redhat.com>
4489 PR tree-optimizatin/78856
4490 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
4491 (mark_threaded_blocks): Remove code to truncate thread paths that
4492 cross multiple loop headers. Instead invalidate the cached loop
4493 iteration information and handle case of a thread path walking
4494 into an irreducible region.
4496 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4499 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
4500 assertions. Add support for doing the signbit if the IEEE 128-bit
4501 floating point value is in a GPR.
4502 * config/rs6000/rs6000.md (Fsignbit): Delete.
4503 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
4504 Update the length attribute if the value is in a GPR.
4505 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
4506 the sign or zero extension instruction, since the value is always 0/1.
4507 (signbit<mode>2_dm2): Delete using <Fsignbit>.
4510 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
4511 extracting SImode to a GPR register so that we can generate a
4512 store, limit the vector to be in a traditional Altivec register
4513 for the vextuwrx instruction.
4515 2017-01-03 Ian Lance Taylor <iant@google.com>
4517 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
4519 2017-01-03 Martin Sebor <msebor@redhat.com>
4521 PR tree-optimization/78696
4522 * gimple-ssa-sprintf.c (format_floating): Correct handling of
4523 precision. Use MPFR for %f for greater fidelity. Correct handling
4525 (pass_sprintf_length::compute_format_length): Set width and precision
4526 specified by asrerisk to void_node for vararg functions.
4527 (try_substitute_return_value): Adjust dump output.
4529 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
4531 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
4533 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
4535 * doc/invoke.texi (SPARC options): Document -mlra as the default.
4536 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
4537 -mlra/-mno-lra was passed to the compiler.
4539 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
4541 PR rtl-optimization/65618
4542 * emit-rtl.c (try_split): Move initialization of "before" and
4543 "after" to just before the call to emit_insn_after_setloc.
4545 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
4547 * doc/md.texi (Standard Names): Remove reference to Java frontend.
4549 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
4551 * dwarf2out.c (gen_enumeration_type_die): When
4552 -gno-strict-dwarf, add a DW_AT_encoding attribute.
4554 2017-01-03 Jakub Jelinek <jakub@redhat.com>
4556 PR tree-optimization/78965
4557 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
4558 Change first argument from const call_info & to call_info &. For %n
4559 set info.nowrite to false.
4562 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
4563 possibly throwing calls.
4565 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
4566 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
4567 and fns handling, rather than in a separate case SSA_NAME.
4569 2017-01-02 Jeff Law <law@redhat.com>
4571 * config/darwin-driver.c (darwin_driver_init): Const-correctness
4572 fixes for first_period and second_period variables.
4574 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
4577 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
4578 (*insvqi_1): New insn pattern.
4579 (*insvqi_1_mem_rex64): Ditto.
4581 (*insvqi_3): Rename from *insvqi.
4583 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
4585 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
4587 * doc/cfg.texi (Edges): Remove reference to Java.
4588 (Maintaining the CFG): Ditto.
4590 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4593 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
4594 transparent aliases.
4596 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4599 * predict.def (PRED_CALL): Update hitrate.
4600 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
4601 * predict.c (tree_estimate_probability_bb): Split CALL predictor
4602 into direct/indirect/polymorphic variants.
4604 2017-01-01 Jakub Jelinek <jakub@redhat.com>
4606 Update copyright years.
4608 * gcc.c (process_command): Update copyright notice dates.
4609 * gcov-dump.c (print_version): Ditto.
4610 * gcov.c (print_version): Ditto.
4611 * gcov-tool.c (print_version): Ditto.
4612 * gengtype.c (create_file): Ditto.
4613 * doc/cpp.texi: Bump @copying's copyright year.
4614 * doc/cppinternals.texi: Ditto.
4615 * doc/gcc.texi: Ditto.
4616 * doc/gccint.texi: Ditto.
4617 * doc/gcov.texi: Ditto.
4618 * doc/install.texi: Ditto.
4619 * doc/invoke.texi: Ditto.
4621 Copyright (C) 2017 Free Software Foundation, Inc.
4623 Copying and distribution of this file, with or without modification,
4624 are permitted in any medium without royalty provided the copyright
4625 notice and this notice are preserved.