1 2017-02-24 Jeff Law <law@redhat.com>
3 PR rtl-optimizatoin/79286
4 * ira.c (update_equiv_regs): Drop may_trap_p exception to
7 2017-02-24 Richard Biener <rguenther@suse.de>
9 PR tree-optimization/79389
10 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
13 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
15 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
16 function comment to reflect reality.
17 (loop_exits_before_overflow): Fix typo in function description.
19 2017-02-24 Richard Biener <rguenther@suse.de>
21 PR tree-optimization/79389
22 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23 properly that a threading opportunity exists. Detect conditional
24 copy/constant propagation opportunities.
26 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28 * config/visium/visium.md (type): Add trap.
29 (b): New mode attribute.
30 (*btst): Rename into...
31 (*btst<mode>): ...this and adjust.
32 (*cbranchsi4_btst_insn): Rename into...
33 (*cbranch<mode>4_btst_insn): ...this and adjust.
34 (trap): New define_insn.
36 2017-02-23 Jakub Jelinek <jakub@redhat.com>
38 PR tree-optimization/79389
39 * ifcvt.c (struct noce_if_info): Add rev_cond field.
40 (noce_reversed_cond_code): New function.
41 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
42 reversed_comparison_code. Formatting fix.
43 (noce_try_store_flag): Test rev_cond != NULL in addition to
44 reversed_comparison_code.
45 (noce_try_store_flag_constants): Likewise.
46 (noce_try_store_flag_mask): Likewise.
47 (noce_try_addcc): Use rev_cond if non-NULL instead of
48 reversed_comparison_code.
49 (noce_try_cmove_arith): Likewise. Formatting fixes.
50 (noce_try_minmax, noce_try_abs): Clear rev_cond.
51 (noce_find_if_block): Initialize rev_cond.
52 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
53 instead of false as last argument never attempt to reverse it
56 2017-02-23 Bin Cheng <bin.cheng@arm.com>
58 PR tree-optimization/79663
59 * tree-predcom.c (combine_chains): Process refs in reverse order
60 only for ZERO length chains, and add explaining comment.
62 2017-02-23 Jeff Law <law@redhat.com>
64 PR tree-optimization/79578
65 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
66 in call to operand_equal_p.
68 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
71 * config/i386/cpuid.h: Fix another undefined behavior.
73 2017-02-23 Richard Biener <rguenther@suse.de>
75 PR tree-optimization/79683
76 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
77 vector types for data-refs.
79 2017-02-23 Martin Liska <mliska@suse.cz>
81 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
83 2017-02-23 Jakub Jelinek <jakub@redhat.com>
86 * internal-fn.c (get_range_pos_neg): Moved to ...
87 * tree.c (get_range_pos_neg): ... here. No longer static.
88 * tree.h (get_range_pos_neg): New prototype.
89 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
90 are known to be in between 0 and signed maximum inclusive, try to
91 expand both unsigned and signed divmod and use the cheaper one from
94 2017-02-22 Jeff Law <law@redhat.com>
96 PR tree-optimization/79578
97 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
98 to compare base operands.
100 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
103 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
104 gpc_reg_operand instead of fpr_reg_operand.
106 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
108 * config/mips/mips.c (mips_return_in_memory): Force FP
109 vector types to be returned in memory for o32 ABI.
111 2017-02-22 Jakub Jelinek <jakub@redhat.com>
113 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
114 instead of DW_TAG_member for static data member declarations and don't
115 set no_linkage_name for static inline data members.
116 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
119 2017-02-22 Martin Liska <mliska@suse.cz>
121 * doc/invoke.texi: Replace inequality signs with square brackets
124 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
126 PR tree-optimization/68644
127 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
129 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
132 * lra-constraints.c (simplify_operand_subreg): Handle
133 WORD_REGISTER_OPERATIONS targets.
135 2017-02-22 Jakub Jelinek <jakub@redhat.com>
138 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
139 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
140 elimination by swapping fld*.
142 2017-02-22 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/79673
145 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
146 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
147 irrelevant address-space qualifiers and avoiding a
148 ADDR_SPACE_CONVERT_EXPR from fold_convert.
150 2017-02-22 Richard Biener <rguenther@suse.de>
152 PR tree-optimization/79666
153 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
154 to not symbolically negate if that may introduce undefined
157 2017-02-22 Martin Liska <mliska@suse.cz>
160 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
161 * data-streamer-out.c (streamer_write_gcov_count_stream):
163 * value-prof.c (stream_out_histogram_value): Make assert more
164 precise based on type of counter.
166 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
169 * config/i386/i386.md (standard_x87sse_constant_load splitter):
170 Use nonimmediate_operand instead of memory_operand for operand 1.
171 (float-extend standard_x87sse_constant_load splitter): Ditto.
173 2017-02-21 Jeff Law <law@redhat.com>
175 PR tree-optimization/79621
176 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
177 blocks with edges to themselves.
179 2017-02-21 Jakub Jelinek <jakub@redhat.com>
182 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
183 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
184 Use gimple_call_builtin_p.
187 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
188 on temporarily removed DEBUG_INSNs.
190 PR tree-optimization/79649
191 * tree-loop-distribution.c (classify_partition): Give up on
192 non-generic address space loads/stores.
194 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
196 * doc/loop.texi (Loop manipulation): Remove nonexistent
197 tree_ssa_loop_version from the documentation.
198 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
200 2017-02-21 Jakub Jelinek <jakub@redhat.com>
203 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
204 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
205 * config/rs6000/rs6000.c: Include except.h.
206 (rs6000_expand_split_stack_prologue): Call
207 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
209 2017-02-21 Martin Jambor <mjambor@suse.cz>
212 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
215 2017-02-21 Martin Jambor <mjambor@suse.cz>
217 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
218 for backward compatibility only.
219 * doc/invoke.texi (Option Summary): Remove all references to
222 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
226 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
228 * lra-constraints.c (curr_insn_transform): Handle
229 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
231 2017-02-21 Martin Liska <mliska@suse.cz>
233 * config/i386/i386.opt: Replace -masm-dialect with -masm.
235 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
238 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
240 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
243 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
244 (arm_function_ok_for_sibcall): Return false for an indirect call by
245 descriptor if all the argument registers are used.
246 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
247 alignment of the function.
249 2017-02-21 Jakub Jelinek <jakub@redhat.com>
251 PR tree-optimization/61441
252 * simplify-rtx.c (simplify_const_unary_operation): For
253 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
256 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
258 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
259 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
260 instead of SYSTEM_HEADER_DIR.
262 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
263 Martin Liška <mliska@suse.cz>
265 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
266 Fix typos and grammar, use active voice, and clarify.
268 2017-02-20 Marek Polacek <polacek@redhat.com>
271 * gimplify.c (gimplify_expr): Handle unused *&&L;.
274 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
276 2017-02-20 Jakub Jelinek <jakub@redhat.com>
279 * config/i386/i386.c (ix86_expand_builtin): Handle
280 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
281 ix86_builtins_isa[fcode].isa as a requirement of those
282 flags and any other flag in the bitmask.
283 (ix86_init_mmx_sse_builtins): Use 0 instead of
284 ~OPTION_MASK_ISA_64BIT as mask.
285 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
286 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
287 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
288 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
290 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
293 * lra-constraints.c (split_reg): Check requested split mode
294 is supported by the register.
296 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
298 * lra-constraints.c (simplify_operand_subreg): Remove early
301 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
304 * lra-constraints.c (curr_insn_transform): Tighten condition
305 for converting SUBREG reloads from OP_OUT to OP_INOUT.
307 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
310 * lra-constraints.c (curr_insn_transform): Handle
311 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
313 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
316 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
318 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
320 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
323 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
326 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
328 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
331 2017-02-18 Jakub Jelinek <jakub@redhat.com>
333 * final.c (last_columnnum, override_columnnum): New variables.
334 (final_start_function): Set last_columnnum, pass it to begin_prologue
335 hook and pass 0 to dwarf2out_begin_prologue.
336 (final_scan_insn): Update override_columnnum. Pass last_columnnum
337 to source_line debug hook.
338 (notice_source_line): Compute last_columnnum and for debug_column_info
339 return true on column changes.
340 * debug.h (struct gcc_debug_hooks): Add column argument to
341 source_line and begin_prologue hooks.
342 (debug_nothing_int_charstar_int_bool): Remove prototype.
343 (debug_nothing_int_int_charstar,
344 debug_nothing_int_int_charstar_int_bool): New prototypes.
345 (dwarf2out_begin_prologue): Add column argument.
346 * debug.c (do_nothing_debug_hooks): Adjust source_line and
347 begin_prologue hooks.
348 (debug_nothing_int_charstar_int_bool): Remove.
349 (debug_nothing_int_int_charstar,
350 debug_nothing_int_int_charstar_int_bool): New functions.
351 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
352 through to dwarf2out_source_line.
353 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
354 (dwarf2out_source_line): Add column argument, emit it if requested.
355 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
357 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
358 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
359 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
360 through to dwarf2out_begin_prologue.
361 (vmsdbgout_source_line): Add column argument, pass it through to
362 dwarf2out_source_line.
363 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
364 dbxout_source_line caller.
365 (dbxout_source_line): Add column argument.
367 * common.opt (gno-column-info, gcolumn-info): New options.
368 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
369 (check_die): Also test for multiple DW_AT_decl_column attributes.
370 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
371 DW_AT_decl_column if requested.
372 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
374 (gen_variable_die): Likewise.
375 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
376 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
379 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
380 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
381 (ix86_handle_option): Handle OPT_m3dnowa.
382 * doc/invoke.texi (-m3dnowa): Document.
383 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
384 -m3dnowa instead of -m3dnow -march=athlon.
387 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
388 instead of gcc_assert for K, r and R code checks. Formatting fixes.
390 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
393 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
394 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
395 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
396 generator for vsx_xxpermdi_<mode>_be.
397 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
398 force big-endian semantics.
399 (vsx_xxpermdi_<mode>_be): New define_expand with same
400 implementation as previous version of vsx_xxpermdi_<mode>.
402 2017-02-17 Jakub Jelinek <jakub@redhat.com>
404 PR tree-optimization/79327
405 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
406 variable, its initialization and use.
408 2017-02-17 Julia Koval <julia.koval@intel.com>
410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
411 (OPTION_MASK_ISA_PKU_UNSET): New.
412 (ix86_handle_option): Handle -mrdpid.
413 * config/i386/cpuid.h (bit_RDPID): New.
414 * config/i386/driver-i386.c (host_detect_local_cpu):
415 Detect RDPID feature.
416 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
417 * config/i386/i386-c.c (ix86_target_macros_internal):
419 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
420 (ix86_valid_target_attribute_inner_p): Add "rdpid".
421 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
422 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
423 * config/i386/i386.md (define_insn "rdpid"): New.
424 * config/i386/i386.opt Add -mrdpid.
425 * config/i386/immintrin.h (_rdpid_u32): New.
427 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
429 PR rtl-optimization/79541
430 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
431 instead of transforming it into USE.
433 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
435 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
436 If HONOR_SNANS (SFmode) force the input to a register.
437 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
438 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
439 an frsp or similar insn.
441 2017-02-17 Martin Liska <mliska@suse.cz>
443 PR rtl-optimization/79577
444 * params.def (selsched-max-sched-times): Increase minimum to 1.
446 2017-02-17 Martin Liska <mliska@suse.cz>
448 PR rtl-optimization/79574
449 * gcse.c (want_to_gcse_p): Prevent integer overflow.
451 2017-02-17 Martin Liska <mliska@suse.cz>
453 PR tree-optimization/79529
454 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
455 ssa_defined_default_def_p to handle cases which are implicitly
457 * tree-ssa.c (ssa_defined_default_def_p): New function.
458 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
459 which are implicitly defined.
460 * tree-ssa.h (ssa_defined_default_def_p): Declare.
462 2017-02-17 Richard Biener <rguenther@suse.de>
465 * params.def (max-ssa-name-query-depth): Limit to 10.
467 2017-02-17 Richard Biener <rguenther@suse.de>
469 PR tree-optimization/79552
470 * tree-ssa-structalias.c (visit_loadstore): Properly verify
473 2017-02-17 Richard Biener <rguenther@suse.de>
476 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
478 2017-02-17 Marek Polacek <polacek@redhat.com>
481 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
482 (fold_negate_expr): New wrapper.
484 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
486 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
487 Correct terminology and de-emphasize pre-standard behavior.
489 2017-02-16 Alan Modra <amodra@gmail.com>
491 PR rtl-optimization/79286
492 * ira.c (def_dominates_uses): New function.
493 (update_equiv_regs): Don't create an equivalence for insns that
494 may trap where the register def does not dominate the use.
496 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
498 PR rtl-optimization/78127
499 * lra.c (lra): Call lra_eliminate before finish the loop after
502 2017-02-16 Richard Biener <rguenther@suse.de>
504 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
506 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
507 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
508 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
509 (isl_val_int_from_wi): New function.
510 (extract_affine_gmp): Rename to ...
511 (extract_affine_wi): ... this, take a widest_int.
512 (extract_affine_int): Just wrap extract_affine_wi.
513 (add_param_constraints): Use isl_val_int_from_wi.
514 (add_loop_constraints): Likewise, and extract_affine_wi.
516 2017-02-15 Jeff Law <law@redhat.com>
519 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
520 ira_init_register_move_cost_if_necessary.
522 2017-02-15 Martin Sebor <msebor@redhat.com>
525 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
526 removed in a prior commit.
528 2017-02-15 Bin Cheng <bin.cheng@arm.com>
530 PR tree-optimization/79347
531 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
532 counters during peeling.
534 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
536 * Makefile.in (site.exp): Remove "set ISLVER".
538 2017-02-15 Jakub Jelinek <jakub@redhat.com>
541 * real.c (real_from_integer): Call real_convert even for decimal.
543 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
546 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
548 2017-02-14 Andrew Pinski <apinski@cavium.com>
550 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
551 cores and change the partno/implementer to be correct.
552 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
553 the 'B" as the implementer.
554 * config/aarch64/aarch64-tune.md: Regenerate.
556 2017-02-14 Carl Love <cel@us.ibm.com>
558 * config/rs6000/rs6000.c: Add case statement entry to make the
559 xvcvuxdsp built-in argument unsigned.
560 * config/rs6000/vsx.md: Fix the source and return operand types so they
561 match the instruction definitions from the ISA document. Fix typo
562 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
565 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
568 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
569 member early_clobber_alts.
570 * lra-lives.c (reg_early_clobber_p): New.
571 (process_bb_lives): Use it.
572 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
573 (debug_operand_data): Initialize early_clobber_alts.
574 (setup_operand_alternative): Set up early_clobber_alts.
575 (collect_non_operand_hard_regs): Ditto. Pass early clobber
576 alternatives to new_insn_reg.
577 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
579 (lra_update_insn_regno_info): Pass the new arg.
581 2017-02-14 Jakub Jelinek <jakub@redhat.com>
584 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
585 (new_oacc_loop_raw): Don't clear already cleared fields.
588 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
589 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
590 _mm512_prefetch_i64gather_ps): New inline functions and macros.
592 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
595 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
597 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
600 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
601 the extra instruction to the right place to store 128-bit constant
604 2017-02-14 Martin Sebor <msebor@redhat.com>
607 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
608 warning for strings of unknown length.
610 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
612 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
614 2017-02-14 Jeff Law <law@redhat.com>
617 * ira-costs.c (scan_one_insn): Initialize register move costs
618 for pseudos seen in USE/CLOBBER insns.
620 PR tree-optimization/79095
621 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
622 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
623 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
624 if the operands are known to be not equal, then the resulting range
626 (intersect_ranges): If the new range is ~[0,0] and the old range is
627 wide, then prefer ~[0,0].
628 * tree-vrp.c (overflow_comparison_p_1): New function.
629 (overflow_comparison_p): New function.
630 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
631 if NAME is used in an overflow test.
632 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
633 overflow check that can be expressed as an equality test, then adjust
634 ops to be that equality test.
636 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
638 * config/s390/s390-builtin-types.def: Remove flags argument.
639 * config/s390/s390.c (s390_init_builtins): Likewise.
641 2017-02-14 Martin Liska <mliska@suse.cz>
643 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
644 vector. Fix trailing white spaces.
646 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
648 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
651 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
653 PR rtl-optimization/68664
654 * config/arm/arm.c (arm_sched_can_speculate_insn):
655 New function. Declare prototype.
656 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
658 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
660 PR rtl-optimization/68664
661 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
663 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
665 2017-02-14 Amit Pawar <amit.pawar@amd.com>
667 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
668 max skip bytes for function, loop and jump.
670 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
672 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
673 ABS_EXPR for gimple dump.
675 2017-02-14 Jakub Jelinek <jakub@redhat.com>
678 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
680 PR tree-optimization/79408
681 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
682 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
683 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
684 also if rhs1 is INTEGER_CST.
686 2017-02-14 Richard Biener <rguenther@suse.de>
689 * tree-into-ssa.c (insert_phi_nodes): When the function can
690 have abnormal edges rewrite SSA names with broken use-def
691 dominance out of SSA and register them for PHI insertion.
693 2017-02-13 Martin Sebor <msebor@redhat.com>
696 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
697 clearing info.nowrite flag when snprintf size argument is a range.
699 2017-02-13 Jakub Jelinek <jakub@redhat.com>
701 * cprop.c (cprop_jump): Add missing space in string literal.
702 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
703 (get_constraint_for_component_ref): Likewise.
704 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
705 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
706 * lra-constraints.c (process_alt_operands): Likewise.
707 * ipa-inline.c (inline_small_functions): Likewise.
708 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
709 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
710 * trans-mem.c (diagnose_tm_1_op): Likewise.
711 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
712 (grid_parallel_clauses_gridifiable): Likewise.
714 * config/nvptx/mkoffload.c (process): Add space in between
717 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
718 "MOD4_SSE_REGS" and "ALL_REGS".
720 * spellcheck.c (test_data): Add , in between "foo" and "food".
722 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
725 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
726 boundary crossing check and subsequent code generation agree.
728 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
730 * config/aarch64/aarch64.c (has_memory_op): Delete.
731 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
734 2017-02-13 Jakub Jelinek <jakub@redhat.com>
736 PR rtl-optimization/79388
737 PR rtl-optimization/79450
738 * combine.c (distribute_notes): When removing TEM_INSN for which
739 corresponding dest has last value recorded, invalidate that last
742 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
744 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
745 of explicit '@'. Add missing assembly comment marker on branch costs
748 2017-02-13 Nathan Sidwell <nathan@acm.org>
750 * gengtype-lex.l (<in_struct>): Add '/'.
752 2017-02-13 Martin Liska <mliska@suse.cz>
755 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
757 2017-02-13 Richard Biener <rguenther@suse.de>
759 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
761 * configure: Re-generate.
762 * config.in: Likewise.
763 * graphite-dependences.c: Simplify as if
764 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
765 * graphite-isl-ast-to-gimple.c: Likewise.
766 * graphite-optimize-isl.c: Likewise.
767 * graphite-poly.c: Likewise.
768 * graphite-sese-to-poly.c: Likewise.
769 * graphite.h: Likewise.
770 * toplev.c: Include isl/version.h and use isl_version () for
771 printing the ISL version.
772 * doc/install.texi: Update ISL requirement.
774 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
776 * doc/standards.texi (Standards): Update reference to
779 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
781 * doc/extend.texi (Named Address Spaces): sourceware.org now
783 * doc/install.texi (Binaries): Ditto.
786 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
788 * doc/cpp.texi: Replace "stringify"/"stringification" with C
789 standard terminology "stringize"/"stringizing" throughout.
790 * doc/cppinternals.texi: Likewise.
792 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
794 * doc/extend.texi: Fix some spelling mistakes and typos.
795 * doc/invoke.texi: Likewise.
797 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
800 * params.def (inline-min-speedup) Change from 10 to 8.
802 2017-02-11 Jakub Jelinek <jakub@redhat.com>
804 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
807 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
810 * ipa-inline-analysis.c (get_minimal_bb): New function.
811 (record_modified): Use it.
812 (remap_edge_change_prob): Handle also ancestor functions.
814 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
816 * doc/contrib.texi (Contributors): Remove broken link into
817 the Mauve CVS repository.
819 2017-02-11 Jakub Jelinek <jakub@redhat.com>
822 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
823 result computation whenever lhs doesn't have vector mode, not
824 just when it has BLKmode.
826 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
828 * doc/makefile.texi (profiledbootstrap): Refer to the
829 installation instructions only in textual form.
831 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
834 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
836 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
838 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
839 (Specific): Update mingw-w64 reference.
841 (Specific): Remove broken link to Renesas RX processor.
843 2017-02-10 Richard Biener <rguenther@suse.de>
845 * toplev.c (process_options): Do not mention obsolete graphite
846 options when printing sorry message about missing graphite support.
847 Mention -floop-nest-optimize.
849 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
851 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
852 (vtst_p16): Likewise.
853 (vtstq_p8): Likewise.
854 (vtstq_p16): Likewise.
856 (vtstq_p64): Likewise.
857 * config/arm/arm_neon.h (vgetq_lane_p64): New.
858 (vset_lane_p64): New.
859 (vsetq_lane_p64): New.
861 2017-02-10 Jakub Jelinek <jakub@redhat.com>
863 PR tree-optimization/79411
864 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
865 stmt operands are SSA_NAMEs used in abnormal phis.
866 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
869 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
872 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
875 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
877 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
879 2017-02-09 Jakub Jelinek <jakub@redhat.com>
881 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
885 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
886 not arbitrary TREE_CONSTANT.
889 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
890 "omp declare target link" attribute unless is_global_var.
891 * omp-offload.c (find_link_var_op): Likewise.
893 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
894 Chung-Lin Tang <cltang@codesourcery.com>
896 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
898 (gimplify_adjust_omp_clauses): Don't delete TILE.
899 (gimplify_omp_for): Deal with TILE.
900 * internal-fn.c (expand_GOACC_TILE): New function.
901 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
903 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
904 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
906 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
907 avoid DIV for outermost collapse var.
908 (expand_oacc_for): Insert tile element loop as needed. Adjust.
909 Remove out of date comments, fix whitespace.
910 * omp-general.c (omp_extract_for_data): Deal with tiling.
911 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
912 adjust OLF_DIM_BASE value.
913 (struct omp_for_data): Add tiling field.
914 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
915 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
916 for auto loops. Remove default auto determining, moved to
917 oacc_loop_fixed_partitions.
918 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
919 stmts, add e_mask field.
920 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
921 (oacc_thread_numbers): Use oacc_dim_call.
922 (oacc_xform_tile): New.
923 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
924 (finish_oacc_loop): Adjust for ifns vector.
925 (oacc_loop_discover_walk): Append loop abstraction sites to list,
926 add case for GOACC_TILE fns.
927 (oacc_loop_xform_loop): Delete.
928 (oacc_loop_process): Iterate over call list directly, and add
929 handling for GOACC_TILE fns.
930 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
932 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
933 vector partitioning to outer loops. Assign 2 partitions to loops
934 when available. Add TILE handling.
935 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
936 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
937 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
938 * tree.c (omp_clause_num_ops): Adjust TILE ops.
939 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
941 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
943 * configure.ac (ACX_BUGURL): Update.
944 * configure: Regenerate.
946 2017-02-09 Richard Biener <rguenther@suse.de>
948 PR tree-optimization/69823
949 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
950 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
952 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
954 * config/arc/arc-c.def: Add __NPS400__ definition.
955 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
956 (TARGET_NPS400): Define.
958 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
960 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
962 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
964 (arc_cpu_types): Fill the arch_info field with a pointer into the
965 arc_arch_types table.
966 (arc_selected_cpu): Declare.
967 * config/arc/arc.c (arc_selected_cpu): Make global.
968 (arc_selected_arch): Delete.
969 (arc_base_cpu): Delete.
970 (arc_override_options): Remove references to deleted variables,
971 update access to arch information.
972 (ARC_OPT): Update access to arch information.
973 (ARC_OPTX): Likewise.
974 * config/arc/arc.h (arc_base_cpu): Remove declaration.
975 (TARGET_ARC600): Update access to arch information.
976 (TARGET_ARC601): Likewise.
977 (TARGET_ARC700): Likewise.
978 (TARGET_EM): Likewise.
979 (TARGET_HS): Likewise.
980 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
983 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
986 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
987 condition/operands for integer GE/LE/GEU/LEU operations.
989 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
992 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
995 2017-02-08 Martin Jambor <mjambor@suse.cz>
998 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
999 whether allocation happened.
1000 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
1001 nothing was allocated.
1003 2017-02-08 Jakub Jelinek <jakub@redhat.com>
1005 PR tree-optimization/79408
1006 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
1007 constant, but SSA_NAME with a known integer range, use the minimum
1008 of that range instead of op1 to determine if modulo can be replaced
1009 with its first operand.
1011 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1013 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
1015 2017-02-08 Richard Biener <rguenther@suse.de>
1017 PR tree-optimization/71824
1018 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
1019 Check all loops contained in the merged region.
1021 2017-02-07 Andrew Pinski <apinski@cavium.com>
1023 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
1025 2017-02-07 Andrew Pinski <apinski@cavium.com>
1027 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
1028 (thunderxt88): Likewise.
1029 (thunderxt81): Disable LSE and change v8.1 to v8.
1030 (thunderxt83): Likewise.
1032 2017-02-07 Jakub Jelinek <jakub@redhat.com>
1033 Richard Biener <rguenther@suse.de>
1036 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
1037 type from int to size_t.
1038 * ira-costs.c (struct_costs_size): Change type from int to size_t.
1040 2017-02-07 Jakub Jelinek <jakub@redhat.com>
1042 PR rtl-optimization/79386
1043 * cprop.c (bypass_conditional_jumps): Initialize
1044 bypass_last_basic_block already before splitting bbs after
1045 unconditional traps...
1046 (bypass_conditional_jumps): ... rather than here.
1049 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
1050 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
1051 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
1052 fix -masm=intel patterns.
1054 2017-02-07 Richard Biener <rguenther@suse.de>
1056 PR tree-optimization/79256
1058 * builtins.c (get_object_alignment_2): Use min_align_of_type
1059 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
1060 and ADJUST_FIELD_ALIGN.
1062 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
1064 * doc/tm.texi: Regenerate.
1065 * stor-layout.c (layout_decl): Adjust.
1066 (update_alignment_for_field): Likewise.
1067 (place_field): Likewise.
1068 (min_align_of_type): Likewise.
1069 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
1070 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
1071 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
1072 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
1073 * config/frv/frv.c (frv_adjust_field_align): Likewise.
1074 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
1075 * config/i386/i386.c (x86_field_alignment): Likewise.
1076 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
1077 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
1078 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
1079 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
1080 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
1081 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
1085 2017-01-30 Richard Biener <rguenther@suse.de>
1087 PR tree-optimization/79256
1088 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
1089 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
1092 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
1094 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
1095 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
1096 builtins to SImode and emit a zero-extend, if necessary.
1098 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1100 * docs/invoke.texi (RISC-V Options): Alphabetize.
1102 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1104 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
1107 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
1109 * config/riscv/riscv.c: New file.
1110 * gcc/common/config/riscv/riscv-common.c: Likewise.
1111 * config.gcc: Likewise.
1112 * config/riscv/constraints.md: Likewise.
1113 * config/riscv/elf.h: Likewise.
1114 * config/riscv/generic.md: Likewise.
1115 * config/riscv/linux.h: Likewise.
1116 * config/riscv/multilib-generator: Likewise.
1117 * config/riscv/peephole.md: Likewise.
1118 * config/riscv/pic.md: Likewise.
1119 * config/riscv/predicates.md: Likewise.
1120 * config/riscv/riscv-builtins.c: Likewise.
1121 * config/riscv/riscv-c.c: Likewise.
1122 * config/riscv/riscv-ftypes.def: Likewise.
1123 * config/riscv/riscv-modes.def: Likewise.
1124 * config/riscv/riscv-opts.h: Likewise.
1125 * config/riscv/riscv-protos.h: Likewise.
1126 * config/riscv/riscv.h: Likewise.
1127 * config/riscv/riscv.md: Likewise.
1128 * config/riscv/riscv.opt: Likewise.
1129 * config/riscv/sync.md: Likewise.
1130 * config/riscv/t-elf-multilib: Likewise.
1131 * config/riscv/t-linux: Likewise.
1132 * config/riscv/t-linux-multilib: Likewise.
1133 * config/riscv/t-riscv: Likewise.
1134 * configure.ac: Likewise.
1135 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
1136 Waterman as RISC-V maintainers.
1137 * doc/install.texi: Add RISC-V entries.
1138 * doc/invoke.texi: Add RISC-V options section.
1139 * doc/md.texi: Add RISC-V constraints section.
1140 * configure: Regenerated.
1142 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1145 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
1146 false values to be constant vectors with all 0 or all 1 bits set.
1147 (vcondu<mode><mode>): Likewise.
1148 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
1150 (fpmask_comparison_operator): Update comment.
1151 (vecint_comparison_operator): New predicate.
1152 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
1153 vector conditionals when the true and false values are constant
1154 vectors with all 0 bits or all 1 bits set.
1156 2017-02-06 Martin Sebor <msebor@redhat.com>
1158 PR tree-optimization/79376
1159 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
1161 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
1163 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
1164 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
1165 to simplify split condition.
1167 2017-02-06 Jakub Jelinek <jakub@redhat.com>
1169 * omp-expand.c (oxpand_omp_atomic_fetch_op,
1170 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
1173 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
1175 PR rtl-optimization/68664
1176 * target.def (can_speculate_insn): New hook.
1177 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
1178 * doc/tm.texi: Regenerate.
1179 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
1180 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
1181 (rs6000_sched_can_speculate_insn): New function.
1183 2017-02-06 Jakub Jelinek <jakub@redhat.com>
1185 PR tree-optimization/79284
1186 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
1187 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
1188 vectorizable_mask_load_store, vectorizable_operation,
1189 vect_is_simple_cond, get_same_sized_vectype): Use it instead
1190 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
1191 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
1192 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
1193 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1194 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
1195 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
1196 is_gimple_assign (stmt). Replace another such test with
1197 is_gimple_assign (stmt).
1199 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
1202 * config/avr/avr.c (rtl-iter.h): Include it.
1203 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
1204 (avr_legitimate_combined_insn): ...and implementation.
1206 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
1208 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
1209 * config/s390/s390.c (s390_const_operand_ok)
1210 (s390_canonicalize_comparison, s390_extract_part)
1211 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
1212 (s390_contiguous_bitmask_p, s390_rtx_costs)
1213 (legitimize_pic_address): Likewise.
1214 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
1215 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
1216 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
1217 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
1218 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
1220 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
1222 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
1223 REGNO($0) == REGNO($1).
1225 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1227 * config/s390/linux.h(SIZE_TYPE): Add comment.
1229 2017-02-06 Julian Brown <julian@codesourcery.com>
1230 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1231 Virendra Pathak <virendra.pathak@broadcom.com>
1233 * config/aarch64/aarch64-cores.def: Change the scheduler
1235 * config/aarch64/aarch64.md: Include thunderx2t99.md.
1236 * config/aarch64/thunderx2t99.md: New file.
1238 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1240 * doc/standards.texi (Go Language): Update link to language
1243 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
1245 * tree-eh.c (lower_resx): Sanitize profile.
1246 (cleanup_empty_eh_move_lp): Likewise.
1248 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
1251 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
1253 * cfgloopmanip.h (loop_version): Update prototype.
1254 * modulo-sched.c (sms_schedule): Update call of loop_version.
1255 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
1256 * tree-parloops.c (gen_parallel_loop): Likewise.
1257 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
1258 * tree-ssa-loop-split.c (split_loop): Likewise.
1259 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
1260 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1262 2017-02-05 Martin Liska <mliska@suse.cz>
1265 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
1267 (print_operand_address): Initialize a struct to zero.
1269 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1271 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
1272 garbage collector only in textual form.
1274 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
1276 * doc/extend.texi (x86 specific memory model extensions for
1277 transactional memory): Simplify a phrase.
1279 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
1282 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
1283 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
1284 (atomic_storedi_1): Likewise.
1286 2017-02-04 Jakub Jelinek <jakub@redhat.com>
1288 PR tree-optimization/79338
1289 * tree-parloops.c (gather_scalar_reductions): Don't call
1290 vect_analyze_loop_form for loop->inner before destroying loop's
1293 2017-02-03 Martin Sebor <msebor@redhat.com>
1295 PR tree-optimization/79327
1296 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
1297 when precision has resulted in leading zeros.
1298 (format_integer): Adjust the likely counter to assume an unknown
1299 argument that may be zero is non-zero.
1301 2017-02-03 Jason Merrill <jason@redhat.com>
1304 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
1305 avoid copying non-taken branch.
1307 2017-02-03 Jakub Jelinek <jakub@redhat.com>
1309 PR tree-optimization/79340
1310 * tree-vect-loop.c (vectorizable_reduction): Release
1311 vec_defs elements after safe_splicing them into other vectors.
1314 PR tree-optimization/79327
1315 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
1316 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
1318 (format_integer): Use wide_int_to_tree instead of build_int_cst
1319 + to_?hwi. If argmin is NULL, just set argmin and argmax to
1320 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
1321 of shortest and longest sequence.
1323 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
1325 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
1326 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
1328 2017-02-03 Walter Lee <walt@tilera.com>
1331 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
1332 after initial stackframe link reg save.
1333 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
1335 2017-02-03 Jakub Jelinek <jakub@redhat.com>
1338 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
1339 wu for stxssp alternative.
1341 2017-02-03 Martin Sebor <msebor@redhat.com>
1343 PR tree-optimization/79352
1344 * gimple-fold.c (get_range_strlen): Add argument.
1345 (get_range_strlen): Change return type to bool.
1346 (get_maxval_strlen): Pass in a dummy argument.
1347 * gimple-fold.h (get_range_strlen): Change return type to bool.
1348 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
1349 * tree.h (array_at_struct_end_p): Add argument.
1350 * tree.c (array_at_struct_end_p): Handle it.
1352 2017-02-03 Martin Liska <mliska@suse.cz>
1355 * multiple_target.c (create_dispatcher_calls): Redirect edge
1356 from a caller of a dispatcher.
1357 (expand_target_clones): Make the clones local.
1358 (ipa_target_clone): Do both target clones and resolvers.
1359 (ipa_dispatcher_calls): Remove the pass.
1360 (pass_dispatcher_calls::gate): Likewise.
1361 (make_pass_dispatcher_calls): Likewise.
1362 * passes.def (pass_target_clone): Put as very first IPA early
1365 2017-02-03 Martin Liska <mliska@suse.cz>
1367 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
1368 in case of a function with ifunc attribute.
1370 2017-02-03 Martin Liska <mliska@suse.cz>
1372 * cgraph.c (cgraph_node::dump): Dump function version info.
1373 * symtab.c (symtab_node::dump_base): Add missing new line.
1375 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1377 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
1378 (ifcombine_ifandif): Use it.
1380 2017-02-03 Martin Liska <mliska@suse.cz>
1382 * doc/invoke.texi: Document default value for
1383 use-after-scope-direct-emission-threshold.
1385 2017-02-03 Martin Liska <mliska@suse.cz>
1387 PR tree-optimization/79339
1388 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
1389 (format_floating): Likewise.
1391 2017-02-03 Martin Liska <mliska@suse.cz>
1394 * ipa-prop.c (ipa_node_params_t::insert): Remove current
1396 (ipa_node_params_t::remove): Likewise.
1397 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
1398 initialization from removed ipa_node_params_t::insert.
1399 (ipa_node_params::~ipa_node_params): Move from removed
1400 ipa_node_params_t::release.
1401 * symbol-summary.h (symbol_summary::m_released): New member.
1402 Do not release a summary twice. Do not allow to call finalizer
1403 for types of a summary that live in GGC memory.
1405 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1407 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
1410 2017-02-02 Martin Sebor <msebor@redhat.com>
1413 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
1414 (format_string): Tighten up the range of output for non-constant
1415 strings and correct the expected range for wide non-constant strings.
1417 2017-02-02 Martin Sebor <msebor@redhat.com>
1419 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
1422 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
1424 (-fdump-tree-@var): Add to index and document how to come up
1425 with pass-specific option and dump file names.
1426 (-fdump-passes): Clarify where to look for output.
1428 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
1431 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
1432 statistics of the analyzed path; allow threading for speed when
1433 any of BBs along the path are optimized for speed.
1435 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
1438 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
1439 settings of the virtual registers.
1442 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
1444 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1445 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
1448 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1450 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
1451 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
1453 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1455 * config/s390/s390.md: Add missing comments with the expanded
1457 * config/s390/vector.md: Likewise.
1458 * config/s390/vx-builtins.md: Likewise.
1460 2017-02-02 Jakub Jelinek <jakub@redhat.com>
1463 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
1464 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
1465 conditions on a single line.
1467 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1469 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1470 __S390_VX__ to __VX__.
1472 2017-02-01 Andrew Pinski <apinski@cavium.com>
1474 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
1475 stmt_info to record_stmt_cost.
1476 (vect_get_known_peeling_cost): Pass stmt_info if known to
1478 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
1479 cpu_vector_cost field into
1480 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
1481 field into vec_int_stmt_cost and vec_fp_stmt_cost.
1482 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
1483 splitting of scalar_stmt_cost and vec_stmt_cost.
1484 (thunderx_vector_cost): Likewise.
1485 (cortexa57_vector_cost): LIkewise.
1486 (exynosm1_vector_cost): Likewise.
1487 (xgene1_vector_cost): Likewise.
1488 (thunderx2t99_vector_cost): Improve after the splitting of the two
1490 (aarch64_builtin_vectorization_cost): Update for the splitting of
1491 scalar_stmt_cost and vec_stmt_cost.
1493 2017-02-01 Torvald Riegel <triegel@redhat.com>
1494 Richard Henderson <rth@redhat.com>
1496 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
1497 conditional on existance of a fast atomic load.
1498 * optabs-query.c (can_atomic_load_p): New function.
1499 * optabs-query.h (can_atomic_load_p): Declare it.
1500 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
1501 no fast atomic load is available for the particular size of access.
1502 (expand_atomic_compare_and_swap): Likewise.
1503 (expand_atomic_load): Likewise.
1504 (expand_atomic_store): Likewise.
1505 (expand_atomic_fetch_op): Likewise.
1506 * testsuite/lib/target-supports.exp
1507 (check_effective_target_sync_int_128): Remove x86 because it provides
1508 no fast atomic load.
1509 (check_effective_target_sync_int_128_runtime): Likewise.
1511 2017-02-01 Richard Biener <rguenther@suse.de>
1513 * graphite.c: Include tree-vectorizer.h for find_loop_location.
1514 (graphite_transform_loops): Provide opt-info for optimized nests.
1515 * tree-parloop.c (parallelize_loops): Provide opt-info for
1518 2017-02-01 Richard Biener <rguenther@suse.de>
1521 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
1524 2017-02-01 Richard Biener <rguenther@suse.de>
1526 PR tree-optimization/71824
1527 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
1528 Verify the loops are valid in the merged SESE region.
1529 (scop_detection::can_represent_loop_1): Check analyzing the
1530 evolution of the number of iterations in the region succeeds.
1532 2017-01-31 Ian Lance Taylor <iant@golang.org>
1534 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
1535 REG_ARGS_SIZE note to 32-bit push insns and call insn.
1537 2017-01-31 David Malcolm <dmalcolm@redhat.com>
1539 PR preprocessor/79210
1540 * input.c (get_substring_ranges_for_loc): Replace line_width
1541 assertion with error-handling.
1543 2017-01-31 Richard Biener <rguenther@suse.de>
1545 PR tree-optimization/77318
1546 * graphite-sese-to-poly.c (extract_affine): Fix assert.
1547 (create_pw_aff_from_tree): Take loop parameter.
1548 (add_condition_to_pbb): Pass loop of the condition to
1549 create_pw_aff_from_tree.
1551 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1553 * config/s390/s390.c (s390_asan_shadow_offset): New function.
1554 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
1556 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1560 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
1562 (convert_int_to_float128): Likewise.
1563 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
1564 (convert_int_to_float128): Likewise.
1565 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
1566 (UNSPEC_IEEE128_CONVERT): Likewise.
1567 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
1568 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
1569 Use local variables for IBM extended format.
1570 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
1571 (fix_trunc<mode>si2_fprs): Likewise.
1572 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
1573 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
1574 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
1575 to know that we can now have integers of all sizes in vector
1577 (fix<uns>_<mode>di2_hw): Likewise.
1578 (float<uns>_<mode>si2_hw): Likewise.
1579 (fix_<mode>si2_hw): Likewise.
1580 (fixuns_<mode>si2_hw): Likewise.
1581 (float<uns>_<mode>di2_hw): Likewise.
1582 (float_<mode>di2_hw): Likewise.
1583 (float_<mode>si2_hw): Likewise.
1584 (floatuns_<mode>di2_hw): Likewise.
1585 (floatuns_<mode>si2_hw): Likewise.
1586 (xscvqp<su>wz_<mode>): Delete, no longer used.
1587 (xscvqp<su>dz_<mode>): Likewise.
1588 (xscv<su>dqp_<mode>): Likewise.
1589 (ieee128_mfvsrd_64bit): Likewise.
1590 (ieee128_mfvsrd_32bit): Likewise.
1591 (ieee128_mfvsrwz): Likewise.
1592 (ieee128_mtvsrw): Likewise.
1593 (ieee128_mtvsrd_64bit): Likewise.
1594 (ieee128_mtvsrd_32bit): Likewise.
1596 2017-01-31 Martin Liska <mliska@suse.cz>
1599 * ipa-prop.c (ipa_free_all_node_params): Call release method
1600 instead of ~sumbol_summary to not to trigger double times
1603 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
1605 PR tree-optimization/71691
1606 * bitmap.h (class auto_bitmap): New.
1607 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
1608 is_maybe_undefined instead of ssa_undefined_value_p.
1610 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1612 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
1613 __S390_ARCH_LEVEL__ to __ARCH__.
1615 2017-01-31 Jakub Jelinek <jakub@redhat.com>
1617 PR tree-optimization/79267
1618 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
1619 if should_remove_lhs_p is true.
1621 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
1624 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
1625 (add_alignment_attribute): New.
1626 (base_type_die): Add alignment attribute.
1627 (subrange_type_die): Likewise.
1628 (modified_type_die): Likewise.
1629 (gen_array_type_die): Likewise.
1630 (gen_descr_array_type_die: Likewise.
1631 (gen_enumeration_type_die): Likewise.
1632 (gen_subprogram_die): Likewise.
1633 (gen_variable_die): Likewise.
1634 (gen_field_die): Likewise.
1635 (gen_ptr_to_mbr_type_die): Likewise.
1636 (gen_struct_or_union_type_die): Likewise.
1637 (gen_subroutine_type_die): Likewise.
1638 (gen_typedef_die): Likewise.
1639 (base_type_cmp): Compare alignment attribute.
1641 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1644 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
1645 (setb_unsigned) New pattern for setb with CCUNS.
1646 * config/rs6000/rs6000.c (expand_block_compare): Use a different
1647 subfc./subfe sequence to avoid overflow problems. Generate a
1648 shorter sequence with cmpld/setb for power9.
1649 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
1650 for generating subfc. instruction.
1651 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
1652 now uses this instruction.
1654 2017-01-30 Ian Lance Taylor <iant@google.com>
1657 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
1658 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
1660 2017-01-30 Martin Sebor <msebor@redhat.com>
1662 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
1663 Move constant to the right of a relational operator.
1664 (get_mpfr_format_length, format_character, format_string): Ditto.
1665 (should_warn_p, maybe_warn): Same.
1667 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
1669 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1672 * asan.c (get_translation_unit_decl): Remove function.
1673 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
1675 2017-01-30 Martin Liska <mliska@suse.cz>
1677 PR gcov-profile/79259
1678 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
1681 2017-01-30 Martin Liska <mliska@suse.cz>
1684 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
1685 Initialize variables with NULL value.
1687 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
1690 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
1692 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
1694 2017-01-30 Richard Biener <rguenther@suse.de>
1696 PR tree-optimization/79276
1697 * tree-vrp.c (process_assert_insertions): Properly adjust common
1698 when removing a duplicate.
1700 * gcc.dg/torture/pr79276.c: New testcase.
1702 2017-01-30 Richard Biener <rguenther@suse.de>
1704 PR tree-optimization/79256
1705 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
1706 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
1708 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
1710 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
1713 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
1714 ("*r<noxa>sbg_<mode>_sll_bitmask")
1715 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
1716 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
1717 Use contiguous_bitmask_nowrap_operand.
1719 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1722 * config/rs6000/altivec.h (vec_xl): Revise #define.
1723 (vec_xst): Likewise.
1725 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
1727 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
1729 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
1731 PR rtl-optimization/79194
1732 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
1733 traps before call to bypass_conditional_jumps.
1735 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1737 PR tree-optimization/71374
1738 * lra-constraints.c (check_conflict_input_operands): New.
1739 (match_reload): Use it.
1741 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
1744 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
1745 account to calculate conflict_set.
1747 2017-01-27 Bin Cheng <bin.cheng@arm.com>
1749 PR rtl-optimization/78559
1750 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
1751 other_insn in combine.
1753 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
1755 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
1756 uint16_type_node for BT_UINT16.
1758 2017-01-27 David Malcolm <dmalcolm@redhat.com>
1760 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
1761 "RTL Tests" to menu.
1762 (GIMPLE Tests): New node.
1763 (RTL Tests): New node.
1765 2017-01-27 Richard Biener <rguenther@suse.de>
1767 PR tree-optimization/79245
1768 * tree-loop-distribution.c (distribute_loop): Apply cost
1769 modeling also to detected patterns.
1771 2017-01-27 Richard Biener <rguenther@suse.de>
1773 PR tree-optimization/71433
1774 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
1775 (compare_assert_loc): New function.
1776 (process_assert_insertions): Sort and optimize assert locations
1777 to remove duplicates and push down identical assertions on
1778 edges to their destination block.
1780 2017-01-27 Richard Biener <rguenther@suse.de>
1782 PR tree-optimization/79244
1783 * tree-vrp.c (remove_range_assertions): Forcefully propagate
1784 out SSA names even if abnormal.
1786 2017-01-27 Jakub Jelinek <jakub@redhat.com>
1788 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
1789 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
1790 instead of MPFR_RNDN.
1792 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
1795 * arm.c (arm_option_override): Don't call build_target_option_node
1796 until after doing all option overrides.
1797 (arm_valid_target_attribute_tree): Likewise.
1799 2017-01-27 Martin Liska <mliska@suse.cz>
1801 * doc/invoke.texi (-fprofile-arcs): Document profiling support
1802 for {cd}tors and C++ {cd}tors.
1804 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
1806 * config/s390/s390.md ("*setmem_long_and")
1807 ("*setmem_long_and_31z"): Use zero_extend instead of and.
1809 2017-01-26 Martin Sebor <msebor@redhat.com>
1811 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
1814 2017-01-26 Martin Sebor <msebor@redhat.com>
1816 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
1817 HAVE_DFmode before using XFmode or DFmode.
1818 (parse_directive): Avoid using the z length modifier to avoid
1819 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
1822 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
1823 to accept adjustment as an array.
1824 (get_int_range): New function.
1825 (struct directive): Make width and prec arrays.
1826 (directive::set_width, directive::set_precision): Call get_int_range.
1827 (format_integer, format_floating): Handle width and precision ranges.
1828 (format_string, parse_directive): Same.
1830 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1833 * dwarf2out.c (generate_skeleton_bottom_up): For children with
1834 comdat_type_p set, just clone them, but keep the children in the
1838 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
1839 which have direct callers with -fvar-tracking-assignments enabled
1841 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
1842 inside of type units.
1844 2017-01-26 Martin Sebor <msebor@redhat.com>
1847 * gimple-ssa-sprintf.c (struct result_range): Add likely and
1849 (struct format_result): Replace number_chars, number_chars_min,
1850 and number_chars_max with a single member of struct result_range.
1852 (format_result::operator+=): Adjust.
1853 (struct fmtresult): Remove bounded. Handle likely and unlikely
1855 (fmtresult::adjust_for_width_or_precision): New function.
1856 (fmtresult:type_max_digits): New function.
1857 (bytes_remaining): Handle likely and unlikely counters.
1858 (min_bytes_remaining): Remove.
1859 (format_percent): Simplify.
1860 (format_integer, format_floating): Set likely and unlikely counters.
1861 (get_string_length, format_character, format_string): Same.
1862 (format_plain, should_warn_p): New function.
1863 (maybe_warn): Call should_warn_p. Update diagnostic messages
1864 and handle those for all directives, including plain strings.
1865 (format_directive): Handle likely and unlikely counters.
1866 Remove unnecessary quoting from diagnostics. Add an informational
1868 (add_bytes): Remove.
1869 (pass_sprintf_length::compute_format_length): Simplify.
1870 (try_substitute_return_value): Handle likely and unlikely counters.
1872 2017-01-26 Carl Love <cel@us.ibm.com>
1874 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
1875 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
1877 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
1880 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
1881 endianess for subregs into account.
1882 * lra-constraints.c (lra_constraints): Do risky transformations
1883 always on the first iteration.
1884 * lra-lives.c (check_pseudos_live_through_calls): Add arg
1885 last_call_used_reg_set.
1886 (process_bb_lives): Define and use last_call_used_reg_set.
1887 * lra.c (lra): Always continue after lra_constraints on the first
1890 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
1892 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
1894 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
1896 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1898 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
1899 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
1900 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
1901 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
1902 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
1903 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
1904 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
1905 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
1906 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
1908 2017-01-26 Marek Polacek <polacek@redhat.com>
1911 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
1912 for the third operand.
1914 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1917 * omp-low.c (struct omp_context): Add simt_stmt field.
1918 (scan_omp_for): Return omp_context *.
1919 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
1920 context to the _simt_ SIMD stmt.
1921 (lower_omp_for): For combined SIMD with sibling _simt_
1922 SIMD, make sure to use the same decls in _looptemp_
1923 clauses as in the sibling.
1925 2017-01-26 David Sherwood <david.sherwood@arm.com>
1928 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
1931 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1934 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
1935 emit fld b; fld a; if possible.
1937 * brig-builtins.def: Update copyright years.
1938 * config/arm/arm_acle_builtins.def: Update copyright years.
1940 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1943 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
1944 constraint instead of o for the stxsd instruction.
1946 2017-01-25 Carl Love <cel@us.ibm.com>
1948 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
1949 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
1951 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
1953 * doc/invoke.texi (C++ Dialect Options): Fix typo.
1955 2017-01-25 Richard Biener <rguenther@suse.de>
1957 PR tree-optimization/69264
1958 * target.def (vector_alignment_reachable): Improve documentation.
1959 * doc/tm.texi: Regenerate.
1960 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
1962 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
1963 earlier changes with respect to TYPE_USER_ALIGN.
1964 (vector_alignment_reachable_p): Likewise. Improve dumping.
1966 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1969 * config/arm/arm.md (xordi3): Force constant operand into a register
1972 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1974 * doc/invoke.texi (-fstore-merging): Correct default optimization
1975 levels at which it is enabled.
1976 (-O): Move -fstore-merging from list to...
1979 2017-01-25 Richard Biener <rguenther@suse.de>
1982 * omp-expand.c: Include debug.h.
1983 (expand_omp_taskreg): Make sure to generate early debug before
1984 outlining anything from a function.
1985 (expand_omp_target): Likewise.
1986 (grid_expand_target_grid_body): Likewise.
1988 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
1991 * asan.c (get_translation_unit_decl): New function.
1992 (asan_add_global): Extract modules file name from globals
1993 TRANSLATION_UNIT_DECL name.
1995 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
1998 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
1999 for long calls with APCS frame and VFP.
2001 2017-01-24 David Malcolm <dmalcolm@redhat.com>
2003 * cfg.c (original_copy_tables_initialized_p): New function.
2004 * cfg.h (original_copy_tables_initialized_p): New decl.
2005 * cfgrtl.c (relink_block_chain): Guard the call to
2006 free_original_copy_tables with a call to
2007 original_copy_tables_initialized_p.
2008 * cgraph.h (symtab_node::native_rtl_p): New decl.
2009 * cgraphunit.c (symtab_node::native_rtl_p): New function.
2010 (symtab_node::needed_p): Don't assert for early assembly output
2011 for __RTL functions.
2012 (cgraph_node::finalize_function): Set "force_output" for __RTL
2014 (cgraph_node::analyze): Bail out early for __RTL functions.
2015 (analyze_functions): Update assertion to support __RTL functions.
2016 (cgraph_node::expand): Bail out early for __RTL functions.
2017 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
2019 * function.h (struct function): Update comment for field
2021 * gimple-expr.c: Include "tree-pass.h".
2022 (gimple_has_body_p): Return false for __RTL functions.
2023 * Makefile.in (OBJS): Add run-rtl-passes.o.
2024 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
2026 (gcc::pass_manager::get_clean_slate): New accessor.
2027 * passes.c: Include "insn-addr.h".
2028 (should_skip_pass_p): Add logging. Update logic for running
2029 "expand" to be compatible with both __GIMPLE and __RTL. Guard
2030 property-provider override so it is only done for gimple passes.
2032 (skip_pass): New function.
2033 (execute_one_pass): Call skip_pass when skipping passes.
2034 * read-md.c (md_reader::read_char): Support filtering
2035 the input to a subset of line numbers.
2036 (md_reader::md_reader): Initialize fields
2037 m_first_line and m_last_line.
2038 (md_reader::read_file_fragment): New function.
2039 * read-md.h (md_reader::read_file_fragment): New decl.
2040 (md_reader::m_first_line): New field.
2041 (md_reader::m_last_line): New field.
2042 * read-rtl-function.c (function_reader::create_function): Only
2043 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
2044 curr_properties. Set DECL_INITIAL to a dummy block.
2045 (read_rtl_function_body_from_file_range): New function.
2046 * read-rtl-function.h (read_rtl_function_body_from_file_range):
2048 * run-rtl-passes.c: New file.
2049 * run-rtl-passes.h: New file.
2051 2017-01-24 Jeff Law <law@redhat.com>
2053 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
2056 2017-01-24 Bin Cheng <bin.cheng@arm.com>
2058 PR tree-optimization/79159
2059 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
2060 (record_nonwrapping_iv): Improve boundary using above function if no
2061 value range information.
2063 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
2064 Martin Jambor <mjambor@suse.cz>
2066 * brig-builtins.def: New file.
2067 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
2068 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
2069 (DEF_HSAIL_SAT_BUILTIN): Likewise.
2070 (DEF_HSAIL_INTR_BUILTIN): Likewise.
2071 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
2072 * builtin-types.def (BT_INT8): New.
2073 (BT_INT16): Likewise.
2074 (BT_UINT8): Likewise.
2075 (BT_UINT16): Likewise.
2076 (BT_FN_ULONG): Likewise.
2077 (BT_FN_UINT_INT): Likewise.
2078 (BT_FN_UINT_ULONG): Likewise.
2079 (BT_FN_UINT_LONG): Likewise.
2080 (BT_FN_UINT_PTR): Likewise.
2081 (BT_FN_ULONG_PTR): Likewise.
2082 (BT_FN_INT8_FLOAT): Likewise.
2083 (BT_FN_INT16_FLOAT): Likewise.
2084 (BT_FN_UINT32_FLOAT): Likewise.
2085 (BT_FN_UINT16_FLOAT): Likewise.
2086 (BT_FN_UINT8_FLOAT): Likewise.
2087 (BT_FN_UINT64_FLOAT): Likewise.
2088 (BT_FN_UINT16_UINT32): Likewise.
2089 (BT_FN_UINT32_UINT16): Likewise.
2090 (BT_FN_UINT16_UINT16_UINT16): Likewise.
2091 (BT_FN_INT_PTR_INT): Likewise.
2092 (BT_FN_UINT_PTR_UINT): Likewise.
2093 (BT_FN_LONG_PTR_LONG): Likewise.
2094 (BT_FN_ULONG_PTR_ULONG): Likewise.
2095 (BT_FN_VOID_UINT64_UINT64): Likewise.
2096 (BT_FN_UINT8_UINT8_UINT8): Likewise.
2097 (BT_FN_INT8_INT8_INT8): Likewise.
2098 (BT_FN_INT16_INT16_INT16): Likewise.
2099 (BT_FN_INT_INT_INT): Likewise.
2100 (BT_FN_UINT_FLOAT_UINT): Likewise.
2101 (BT_FN_FLOAT_UINT_UINT): Likewise.
2102 (BT_FN_ULONG_UINT_UINT): Likewise.
2103 (BT_FN_ULONG_UINT_PTR): Likewise.
2104 (BT_FN_ULONG_ULONG_ULONG): Likewise.
2105 (BT_FN_UINT_UINT_UINT): Likewise.
2106 (BT_FN_VOID_UINT_PTR): Likewise.
2107 (BT_FN_UINT_UINT_PTR: Likewise.
2108 (BT_FN_UINT32_UINT64_PTR): Likewise.
2109 (BT_FN_INT_INT_UINT_UINT): Likewise.
2110 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
2111 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
2112 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
2113 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
2114 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
2115 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
2116 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
2117 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
2118 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
2119 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
2120 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
2121 * doc/frontends.texi: List BRIG FE.
2122 * doc/install.texi (Testing): Add BRIG tesring requirements.
2123 * doc/invoke.texi (Overall Options): Mention BRIG.
2124 * doc/standards.texi (Standards): Doucment BRIG HSA version.
2126 2017-01-24 Richard Biener <rguenther@suse.de>
2128 PR translation/79208
2129 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
2131 2017-01-24 Martin Jambor <mjambor@suse.cz>
2134 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
2135 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
2138 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
2141 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
2142 casts from signed to unsigned really don't have a range.
2144 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
2146 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
2147 GMP_RNDx for compatiblity.
2149 2017-01-24 Martin Liska <mliska@suse.cz>
2152 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
2153 that would prevent us to call alloca with -1 as argument.
2155 2017-01-24 Jakub Jelinek <jakub@redhat.com>
2157 * dwarf2out.c (output_compilation_unit_header, output_file_names):
2158 Avoid -Wformat-security warning.
2160 2017-01-23 Andrew Pinski <apinski@cavium.com>
2162 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
2165 2017-01-23 Martin Sebor <msebor@redhat.com>
2168 * gimple-ssa-sprintf.c (warn_level): New global.
2169 (format_integer): Use it here and throughout the rest of the file.
2170 Use the same switch to compute sign as base.
2171 (maybe_warn): New function.
2172 (format_directive): Factor out warnings into maybe_warn.
2173 Add debugging output. Use warn_level.
2174 (add_bytes): Use warn_level.
2175 (pass_sprintf_length::compute_format_length): Add debugging output.
2176 (try_substitute_return_value): Same.
2177 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
2180 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
2181 (struct fmtresult, format_integer, format_floating): Adjust.
2182 (fmtresult::fmtresult): Set max correctly in two argument ctor.
2183 (get_string_length, format_string,format_directive): Same.
2184 (pass_sprintf_length::compute_format_length): Same.
2185 (try_substitute_return_value): Simplify slightly.
2188 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
2189 (fmtresult::operator+=): Outlined.
2190 (struct fmtresult): Add ctors.
2191 (struct conversion_spec): Rename...
2192 (struct directive): ...to this. Add and remove data members.
2193 (directive::set_width, directive::set_precision): New functions.
2194 (format_percent): Use fmtresult ctor.
2195 (get_width_and_precision): Remove.
2196 (format_integer): Make naming changes. Avoid computing width and
2198 (format_floating): Same. Adjust indentation.
2199 (format_character, format_none): New functions.
2200 (format_string): Moved character handling to format_character.
2201 (format_directive): Remove arguments, change return type.
2202 (parse_directive): New function.
2203 (pass_sprintf_length::compute_format_length): Move directive
2204 parsing to parse_directive.
2206 2017-01-23 Jakub Jelinek <jakub@redhat.com>
2208 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
2209 (assign_assembler_name_if_needed): ... this.
2210 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
2211 (assign_assembler_name_if_needed): ... this.
2212 (free_lang_data_in_cgraph): Adjust callers.
2213 * cgraphunit.c (cgraph_node::analyze): Likewise.
2214 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
2217 2017-01-23 Richard Biener <rguenther@suse.de>
2219 PR tree-optimization/79088
2220 PR tree-optimization/79188
2221 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
2222 resetting loop bounds after last path deletion. Reset loop
2223 bounds of the target loop, make code match the comments.
2224 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
2225 Make sure loops need no fixups.
2227 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
2229 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
2230 exponent support with double type for first argument.
2231 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
2232 type returned by __builtin_vec_extract_sig,
2233 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
2234 functions from "vector int" to "vector unsigned int" or from
2235 "vector long long int" to "vector unsigned long long int".
2236 Changed type returned by __builtin_vec_extract_exp,
2237 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
2238 functions from "vector int" to "vector unsigned int" or from
2239 "vector long long int" to "vector unsigned long long int".
2240 Changed return type of __builtin_vec_test_data_class,
2241 __builtin_vec_test_data_class_sp, and
2242 __builtin_vec_test_data_class_dp from "vector int" to
2243 "vector bool int" or from "vector long long int" to "vector bool
2244 long long int" and changed second argument type from "unsigned
2245 int" to "int". Added new overloaded function forms "vector float
2246 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
2247 "vector float __builtin_vec_insert_exp_sp (vector float, vector
2248 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
2249 double, vector unsigned long long int)" and "vector double
2250 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
2251 long int)". Changed return type of
2252 __builtin_scalar_test_data_class and
2253 __builtin_scalar_test_data_class_sp and
2254 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
2255 int" and changed second argument from "unsigned int" to "int".
2256 Changed type returned by __builtin_scalar_test_neg,
2257 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
2258 from "int" to "bool int". Added new overloaded function form
2259 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
2260 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
2261 exponent double-precision with floating point first argument.
2262 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
2263 documentation of scalar_test_data_class, scalar_test_neg,
2264 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
2265 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
2266 vec_test_data_class built-in functions to reflect refinements in
2267 their type signatures.
2269 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
2271 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
2273 (aarch64_elf_asm_destructor): Likewise.
2275 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
2277 PR rtl-optimization/78634
2278 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
2279 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
2280 * ifcvt.c (noce_try_cmove): Add missing cost check.
2282 PR rtl-optimization/71724
2283 * combine.c (if_then_else_cond): Look for situations where it is
2284 beneficial to undo the work of one of the recursive calls.
2286 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2288 PR tree-optimization/70754
2289 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2290 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2291 combined stmt before it if not NULL.
2292 (combine_chains): Process refs reversely and compute dominance point
2295 2017-01-23 Martin Liska <mliska@suse.cz>
2297 PR tree-optimization/79196
2298 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
2299 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
2301 (strlen_optimize_stmt): Call the renamed function.
2303 2017-01-23 Michael Matz <matz@suse.de>
2305 PR tree-optimization/78384
2306 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
2308 2017-01-23 Richard Biener <rguenther@suse.de>
2310 PR tree-optimization/79186
2311 * tree-vrp.c (register_new_assert_for): Make sure we've seen
2312 both incoming edges before moving an assert.
2314 2017-01-23 Martin Jambor <mjambor@suse.cz>
2316 * ipa-prop.c (load_from_param_1): Removed.
2317 (load_from_unmodified_param): Bits from load_from_param_1 put back
2319 (load_from_param): Removed.
2320 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
2321 with stmt. Reverted back to use of load_from_unmodified_param.
2323 2017-01-23 Martin Jambor <mjambor@suse.cz>
2326 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
2327 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
2328 field a pointer to garbage collected vector, mark lattices and
2329 ipcp_orig_node with GTY((skip)).
2330 (ipa_get_param_count): Adjust to descriptors being a pointer.
2331 (ipa_get_param): Likewise.
2332 (ipa_get_type): Likewise.
2333 (ipa_get_param_move_cost): Likewise.
2334 (ipa_set_param_used): Likewise.
2335 (ipa_get_controlled_uses): Likewise.
2336 (ipa_set_controlled_uses): Likewise.
2337 (ipa_is_param_used): Likewise.
2338 (ipa_node_params_t): Move into garbage collector. New methods insert
2340 (ipa_node_params_sum): Annotate wth GTY(()).
2341 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
2343 (ipa_load_from_parm_agg): Adjust declaration.
2344 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
2345 * ipa-profile.c (ipa_profile): Likewise.
2346 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
2347 (ipa_populate_param_decls): Make descriptors parameter garbage
2349 (ipa_dump_param): Adjust to descriptors being a pointer.
2350 (ipa_alloc_node_params): Likewise.
2351 (ipa_initialize_node_params): Likewise.
2352 (load_from_param_1): Make descriptors parameter garbage collected.
2353 (load_from_unmodified_param): Likewise.
2354 (load_from_param): Likewise.
2355 (ipa_load_from_parm_agg): Likewise.
2356 (ipa_node_params::~ipa_node_params): Removed.
2357 (ipa_free_all_node_params): Remove call to delete operator.
2358 (ipa_node_params_t::insert): New.
2359 (ipa_node_params_t::remove): Likewise.
2360 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
2361 copy known_csts and known_contexts vectors.
2362 (ipa_read_node_info): Adjust to descriptors being a pointer.
2363 (ipcp_modif_dom_walker): Make m_descriptors field garbage
2365 (ipcp_transform_function): Make descriptors variable garbage
2368 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
2370 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
2371 * config/i386/avx512dqintrin.h: Ditto.
2372 * config/i386/avx512fintrin.h: Ditto.
2373 * gcc/config/i386/i386.c: Handle new builtins.
2374 * config/i386/i386-builtin.def: Add new builtins.
2375 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
2376 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
2378 2017-01-23 Jakub Jelinek <jakub@redhat.com>
2379 Martin Liska <mliska@suse.cz>
2381 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
2382 * asan.c (asan_expand_poison_ifn): Support stores and use
2383 appropriate ASAN report function.
2384 * internal-fn.c (expand_ASAN_POISON_USE): New function.
2385 * internal-fn.def (ASAN_POISON_USE): Declare.
2386 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
2387 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
2388 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
2389 ASAN_POISON calls w/o LHS.
2390 * tree-ssa.c (execute_update_addresses_taken): Create clobber
2391 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
2392 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
2393 * gimplify.c (asan_poison_variables): Add attribute
2394 use_after_scope_memory to variables that really needs to live
2396 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
2397 having the attribute.
2399 2017-01-23 Martin Liska <mliska@suse.cz>
2401 * asan.c (create_asan_shadow_var): New function.
2402 (asan_expand_poison_ifn): Likewise.
2403 * asan.h (asan_expand_poison_ifn): New declaration.
2404 * internal-fn.c (expand_ASAN_POISON): Likewise.
2405 * internal-fn.def (ASAN_POISON): New builtin.
2406 * sanopt.c (pass_sanopt::execute): Expand
2407 asan_expand_poison_ifn.
2408 * tree-inline.c (copy_decl_for_dup_finish): Make function
2410 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
2411 * tree-ssa.c (is_asan_mark_p): New function.
2412 (execute_update_addresses_taken): Rewrite local variables
2413 (identified just by use-after-scope as addressable) into SSA.
2415 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2417 * doc/install.texi (Specific): opensource.apple.com uses https
2418 now. Remove trailing slash.
2420 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
2422 * README.Portability: Remove note on an Irix compatibility issue.
2424 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
2426 * gcov.c (INCLUDE_ALGORITHM): Define.
2427 (INCLUDE_VECTOR): Define.
2428 No longer include <vector> and <algorithm> directly.
2430 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
2432 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
2434 * doc/invoke.texi (Code Gen Options): Ditto.
2436 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
2439 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
2441 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
2443 rtl-optimization/79125
2444 * cprop.c (local_cprop_pass): Handle cases where we make an
2447 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
2451 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
2452 read from, for big endian.
2454 2017-01-20 Jiong Wang <jiong.wang@arm.com>
2456 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
2457 register pauth builtins for LP64 only.
2459 2017-01-20 Marek Polacek <polacek@redhat.com>
2462 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
2465 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2467 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
2469 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
2470 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
2471 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
2473 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2476 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
2477 in big-endian mode when they are not a single duplicated value.
2479 2017-01-20 Richard Biener <rguenther@suse.de>
2481 * BASE-VER: Bump to 7.0.1.
2483 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
2485 * omp-low.c (omplow_simd_context): New struct. Use it...
2486 (lower_rec_simd_input_clauses): ...here and...
2487 (lower_rec_input_clauses): ...here to hold common data. Adjust all
2488 references to idx, lane, max_vf, is_simt.
2490 2017-01-20 Graham Markall <graham.markall@embecosm.com>
2492 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
2495 2017-01-20 Martin Jambor <mjambor@suse.cz>
2497 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
2498 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
2500 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
2501 (GTFILES): Rename hsa.c to hsa-common.c.
2502 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
2503 * hsa-dump.c: Likewise.
2504 * hsa-gen.c: Likewise.
2505 * hsa-regalloc.c: Likewise.
2506 * ipa-hsa.c: Likewise.
2507 * omp-expand.c: Likewise.
2508 * omp-low.c: Likewise.
2509 * toplev.c: Likewise.
2511 2017-01-20 Marek Polacek <polacek@redhat.com>
2514 * doc/invoke.texi: Document -Wduplicated-branches.
2515 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
2516 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
2517 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
2518 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
2519 return 0 only when not OEP_LEXICOGRAPHIC.
2520 (fold_build_cleanup_point_expr): Use the expression
2521 location when building CLEANUP_POINT_EXPR.
2522 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
2523 * tree.c (add_expr): Handle error_mark_node.
2525 2017-01-20 Martin Liska <mliska@suse.cz>
2528 * tree-profile.c (init_ic_make_global_vars): Do not call
2530 (gimple_init_gcov_profiler): Likewise.
2532 2017-01-20 Martin Liska <mliska@suse.cz>
2535 * cgraph.h (maybe_create_reference): Remove argument and
2537 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
2539 * ipa-cp.c (create_specialized_node): Likewise.
2540 * symtab.c (symtab_node::maybe_create_reference): Handle
2541 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
2543 2017-01-20 Martin Liska <mliska@suse.cz>
2545 * read-rtl-function.c (function_reader::create_function): Use
2546 build_decl instread of build_decl_stat.
2548 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
2550 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
2551 * config/i386/avx512dqintrin.h: Ditto.
2552 * config/i386/avx512fintrin.h: Ditto.
2553 * config/i386/i386-builtin-types.def: Add new types.
2554 * gcc/config/i386/i386.c: Handle new types.
2555 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
2556 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
2557 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
2558 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
2559 (__builtin_ia32_kshiftridi): New.
2560 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
2562 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
2566 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
2567 define to rs6000_init_stack_protect_guard.
2568 (rs6000_init_stack_protect_guard): New function.
2570 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2571 Yunqiang Su <yunqiang.su@imgtec.com>
2573 * config.gcc (supported_defaults): Add madd4.
2574 (with_madd4): Add validation.
2575 (all_defaults): Add madd4.
2576 * config/mips/mips.opt (mmadd4): New option.
2577 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2579 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2581 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
2582 (ISA_HAS_FUSED_MADD4): Likewise.
2583 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
2584 * gcc/doc/install.texi (--with-madd4): Document the new option.
2586 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2588 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
2589 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
2590 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
2591 (aarch64_init_pauth_hint_builtins): New.
2592 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
2593 (aarch64_expand_builtin): Expand new builtins.
2595 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2597 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
2598 * combine-stack-adj.c (no_unhandled_cfa): Handle
2599 REG_CFA_TOGGLE_RA_MANGLE.
2600 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
2601 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
2602 info for return address signing.
2603 (aarch64_expand_epilogue): Likewise.
2605 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2607 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
2608 * config/aarch64/aarch64-protos.h
2609 (aarch64_return_address_signing_enabled): New declaration.
2610 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
2612 (aarch64_expand_prologue): Sign return address before it's pushed onto
2614 (aarch64_expand_epilogue): Authenticate return address fetched from
2616 (aarch64_override_options): Sanity check for ILP32 and ISA level.
2617 (aarch64_attributes): New function attributes for "sign-return-address".
2618 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
2619 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
2620 ("*do_return"): Generate combined instructions according to key index.
2621 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
2622 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
2624 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
2625 * config/aarch64/aarch64.opt (msign-return-address=): New.
2626 * doc/extend.texi (AArch64 Function Attributes): Documents
2627 "sign-return-address=".
2628 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
2630 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
2632 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
2633 overall option summary.
2635 2017-01-19 Jiong Wang <jiong.wang@arm.com>
2637 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
2638 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
2639 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
2640 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
2642 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
2644 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
2645 -mpower9-minmax by default for -mcpu=power9.
2646 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
2647 128-bit floating point.
2649 2017-01-20 Alan Modra <amodra@gmail.com>
2651 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
2652 optimizing for size.
2654 2017-01-20 Alan Modra <amodra@gmail.com>
2657 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
2658 for strcmp and strncmp from corresponding builtin decl.
2660 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2662 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
2663 instead of i386/rtems-64.h.
2664 * config/i386/rtems-64.h: Remove.
2666 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
2670 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
2672 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
2674 2017-01-19 Tamar Christina <tamar.christina@arm.com>
2676 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2677 Change int to HOST_WIDE_INT.
2678 * config/aarch64/aarch64-protos.h
2679 (aarch64_simd_gen_const_vector_dup): Likewise.
2680 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
2682 2017-01-19 David Malcolm <dmalcolm@redhat.com>
2684 * langhooks-def.h (lhd_type_for_size): New decl.
2685 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
2686 * langhooks.c (lhd_type_for_size): New function, taken from
2689 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
2691 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
2692 define_bypass for CR latency.
2693 (power9-cracked-alu): Update bypass latency and remove power9-branch.
2694 (power9-alu2): Add define_bypass for CR latency.
2696 (power9-mul): Update insn latency.
2697 (power9-mul-compare): Update insn latency, bypass latency and remove
2700 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2702 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
2704 * config/aarch64/aarch64.md
2705 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
2706 aarch64_nopcrelative_literal_loads.
2707 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
2709 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
2711 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
2713 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
2715 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
2718 * config.gcc (supported_defaults): Add lxc1-sxc1.
2719 (with_lxc1_sxc1): Add validation.
2720 (all_defaults): Add lxc1-sxc1.
2721 * config/mips/mips.opt (mlxc1-sxc1): New option.
2722 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
2724 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
2725 __mips_no_lxc1_sxc1.
2726 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
2727 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
2728 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
2730 2017-01-19 Richard Biener <rguenther@suse.de>
2732 PR tree-optimization/72488
2733 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
2734 sure to restore SSA info.
2735 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
2737 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
2739 PR rtl-optimization/79121
2740 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
2741 of the inner type when shifting an extended value.
2743 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
2746 * symtab.c (symtab_node::equal_address_to): Fix comparing of
2747 interposable aliases.
2749 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
2752 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
2753 Use the evmergelohi instruction.
2754 (mov_si<mode>_e500_subreg4_2_le): Likewise.
2755 (mov_sitf_e500_subreg8_2_be): Likewise.
2756 (mov_sitf_e500_subreg12_2_le): Likewise.
2757 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
2758 (mov_si<mode>_e500_subreg4_2_be): Likewise.
2759 (mov_sitf_e500_subreg8_2_le): Likewise.
2760 (mov_sitf_e500_subreg12_2_be): Likewise.
2762 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2764 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
2765 attribute from vecsimple to vecperm.
2766 (altivec_vbpermq2): Likewise.
2768 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2771 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
2773 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2774 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
2775 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
2776 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
2777 case where N arg is SIZE_MAX.
2778 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
2779 (cmpstrsi): Add pattern.
2781 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2783 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2784 __builtin_vec_revb builtins.
2785 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
2786 built-in functions to support generation of the ISA 3.0 XXBR<x>
2787 vector byte reverse instructions.
2788 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
2789 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
2790 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
2791 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
2792 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
2793 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
2794 (P9V_BUILTIN_VEC_REVB): Likewise.
2795 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
2796 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
2797 (p9_xxbrq_v16qi): Likewise.
2798 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
2799 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
2800 (p9_xxbrh_v8hi): Likewise.
2801 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
2802 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
2803 vec_revb built-in functions.
2805 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
2807 PR rtl-optimization/78952
2808 * config/i386/i386.md (any_extract): New code iterator.
2809 (*insvqi_2): Use any_extract for source operand.
2810 (*insvqi_3): Use any_shiftrt for source operand.
2812 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
2814 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
2816 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
2818 2017-01-18 Matthias Klose <doko@ubuntu.com>
2820 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
2822 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2824 * config/rs6000/altivec.h (vec_bperm): Change #define.
2825 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
2826 (altivec_vbpermq2): New define_insn.
2827 (altivec_vbpermd): Likewise.
2828 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
2830 (VBPERMD): Likewise.
2831 (VBPERM): New polymorphic function interface.
2832 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
2833 Add entries for P9V_BUILTIN_VEC_VBPERM.
2834 * doc/extend.texi: Add interfaces for vec_bperm.
2836 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2838 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
2839 first letter of error messages.
2840 (s390_resolve_overloaded_builtin): Likewise.
2841 * config/s390/s390.c (s390_expand_builtin): Likewise.
2842 (s390_invalid_arg_for_unprototyped_fn): Likewise.
2843 (s390_valid_target_attribute_inner_p): Likewise.
2844 * config/s390/s390.md ("tabort"): Likewise.
2846 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
2848 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
2849 (ISA_AVOID_DIV_HILO): New macro.
2850 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
2851 (ISA_HAS_DDIV): Likewise.
2853 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2855 * doc/invoke.texi (fabi-version): Correct number of occurrences.
2857 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2859 * doc/invoke.texi (fabi-version): Spelling fix.
2861 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2864 * doc/invoke.texi (fabi-version): Mention mangling fix for
2867 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2870 * doc/invoke.texi (fabi-version): Document discriminator mangling.
2872 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
2875 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
2876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
2878 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
2880 (stack_protect_test): Ditto.
2881 * config/rs6000/rs6000.opt (mstack-protector-guard=,
2882 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
2884 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
2885 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
2886 -mstack-protector-guard-offset=.
2887 (RS/6000 and PowerPC Options): Ditto.
2889 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2891 * config/i386/i386.h (MASK_CLASS_P): New define.
2892 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
2893 there are no registers from different register sets also when
2894 mask registers are used. Update function comment.
2895 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
2896 to (*k/*r) and (*k/*km) alternatives.
2898 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
2900 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
2901 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
2902 (EH_RETURN_HANDLER_RTX): New define.
2903 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
2904 Force frame pointer in EH return functions.
2905 (aarch64_expand_epilogue): Add barrier for eh_return.
2906 (aarch64_final_eh_return_addr): Remove.
2907 (aarch64_eh_return_handler_rtx): New function.
2908 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
2910 (aarch64_eh_return_handler_rtx): New prototype.
2912 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2914 * config/rs6000/altivec.h (vec_rlmi): New #define.
2915 (vec_vrlnm): Likewise.
2916 (vec_rlnm): Likewise.
2917 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
2918 (UNSPEC_VRLNM): Likewise.
2919 (VIlong): New mode iterator.
2920 (altivec_vrl<VI_char>mi): New define_insn.
2921 (altivec_vrl<VI_char>nm): Likewise.
2922 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
2925 (RLNM): New polymorphic function entry.
2926 (VRLWMI): New monomorphic function entry.
2928 (RLMI): New polymorphic function entry.
2929 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
2930 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
2931 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
2934 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2937 * dwarf2out.c (field_byte_offset): Restore the
2938 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
2939 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
2940 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
2943 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2946 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
2948 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2951 * dwarf2out.c (add_data_member_location_attribute): For constant
2952 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
2953 instead of DW_AT_data_member_location, DW_AT_bit_offset and
2954 DW_AT_byte_size attributes.
2956 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2958 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
2959 after forcing to constant memory when the code model is medium.
2961 2017-01-17 Julia Koval <julia.koval@intel.com>
2964 * config/i386/avx512fintrin.h
2965 (_mm512_i32gather_ps): Change __addr type to void const*.
2966 (_mm512_mask_i32gather_ps): Ditto.
2967 (_mm512_i32gather_pd): Ditto.
2968 (_mm512_mask_i32gather_pd): Ditto.
2969 (_mm512_i64gather_ps): Ditto.
2970 (_mm512_mask_i64gather_ps): Ditto.
2971 (_mm512_i64gather_pd): Ditto.
2972 (_mm512_mask_i64gather_pd): Ditto.
2973 (_mm512_i32gather_epi32): Ditto.
2974 (_mm512_mask_i32gather_epi32): Ditto.
2975 (_mm512_i32gather_epi64): Ditto.
2976 (_mm512_mask_i32gather_epi64): Ditto.
2977 (_mm512_i64gather_epi32): Ditto.
2978 (_mm512_mask_i64gather_epi32): Ditto.
2979 (_mm512_i64gather_epi64): Ditto.
2980 (_mm512_mask_i64gather_epi64): Ditto.
2981 (_mm512_i32scatter_ps): Change __addr type to void*.
2982 (_mm512_mask_i32scatter_ps): Ditto.
2983 (_mm512_i32scatter_pd): Ditto.
2984 (_mm512_mask_i32scatter_pd): Ditto.
2985 (_mm512_i64scatter_ps): Ditto.
2986 (_mm512_mask_i64scatter_ps): Ditto.
2987 (_mm512_i64scatter_pd): Ditto.
2988 (_mm512_mask_i64scatter_pd): Ditto.
2989 (_mm512_i32scatter_epi32): Ditto.
2990 (_mm512_mask_i32scatter_epi32): Ditto.
2991 (_mm512_i32scatter_epi64): Ditto.
2992 (_mm512_mask_i32scatter_epi64): Ditto.
2993 (_mm512_i64scatter_epi32): Ditto.
2994 (_mm512_mask_i64scatter_epi32): Ditto.
2995 (_mm512_i64scatter_epi64): Ditto.
2996 (_mm512_mask_i64scatter_epi64): Ditto.
2997 * config/i386/avx512pfintrin.h
2998 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
2999 (_mm512_mask_prefetch_i32gather_ps): Ditto.
3000 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3001 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3002 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
3003 (_mm512_prefetch_i32scatter_ps): Ditto.
3004 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3005 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3006 (_mm512_prefetch_i64scatter_pd): Ditto.
3007 (_mm512_prefetch_i64scatter_ps): Ditto.
3008 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3009 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3010 * config/i386/avx512vlintrin.h
3011 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
3012 (_mm_mmask_i32gather_ps): Ditto.
3013 (_mm256_mmask_i32gather_pd): Ditto.
3014 (_mm_mmask_i32gather_pd): Ditto.
3015 (_mm256_mmask_i64gather_ps): Ditto.
3016 (_mm_mmask_i64gather_ps): Ditto.
3017 (_mm256_mmask_i64gather_pd): Ditto.
3018 (_mm_mmask_i64gather_pd): Ditto.
3019 (_mm256_mmask_i32gather_epi32): Ditto.
3020 (_mm_mmask_i32gather_epi32): Ditto.
3021 (_mm256_mmask_i32gather_epi64): Ditto.
3022 (_mm_mmask_i32gather_epi64): Ditto.
3023 (_mm256_mmask_i64gather_epi32): Ditto.
3024 (_mm_mmask_i64gather_epi32): Ditto.
3025 (_mm256_mmask_i64gather_epi64): Ditto.
3026 (_mm_mmask_i64gather_epi64): Ditto.
3027 (_mm256_i32scatter_ps): Change __addr type to void*.
3028 (_mm256_mask_i32scatter_ps): Ditto.
3029 (_mm_i32scatter_ps): Ditto.
3030 (_mm_mask_i32scatter_ps): Ditto.
3031 (_mm256_i32scatter_pd): Ditto.
3032 (_mm256_mask_i32scatter_pd): Ditto.
3033 (_mm_i32scatter_pd): Ditto.
3034 (_mm_mask_i32scatter_pd): Ditto.
3035 (_mm256_i64scatter_ps): Ditto.
3036 (_mm256_mask_i64scatter_ps): Ditto.
3037 (_mm_i64scatter_ps): Ditto.
3038 (_mm_mask_i64scatter_ps): Ditto.
3039 (_mm256_i64scatter_pd): Ditto.
3040 (_mm256_mask_i64scatter_pd): Ditto.
3041 (_mm_i64scatter_pd): Ditto.
3042 (_mm_mask_i64scatter_pd): Ditto.
3043 (_mm256_i32scatter_epi32): Ditto.
3044 (_mm256_mask_i32scatter_epi32): Ditto.
3045 (_mm_i32scatter_epi32): Ditto.
3046 (_mm_mask_i32scatter_epi32): Ditto.
3047 (_mm256_i32scatter_epi64): Ditto.
3048 (_mm256_mask_i32scatter_epi64): Ditto.
3049 (_mm_i32scatter_epi64): Ditto.
3050 (_mm_mask_i32scatter_epi64): Ditto.
3051 (_mm256_i64scatter_epi32): Ditto.
3052 (_mm256_mask_i64scatter_epi32): Ditto.
3053 (_mm_i64scatter_epi32): Ditto.
3054 (_mm_mask_i64scatter_epi32): Ditto.
3055 (_mm256_i64scatter_epi64): Ditto.
3056 (_mm256_mask_i64scatter_epi64): Ditto.
3057 (_mm_i64scatter_epi64): Ditto.
3058 (_mm_mask_i64scatter_epi64): Ditto.
3059 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
3060 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
3061 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
3062 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
3063 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
3064 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
3065 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
3066 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
3067 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
3068 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
3069 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
3070 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
3071 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
3072 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
3073 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
3074 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
3075 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
3076 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
3077 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
3078 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
3079 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
3080 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
3081 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
3082 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
3083 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
3084 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
3085 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
3086 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
3087 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
3088 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
3089 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
3090 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
3091 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
3092 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
3093 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
3094 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
3095 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
3096 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
3097 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
3098 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
3099 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
3100 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
3101 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
3102 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
3103 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
3104 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
3105 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
3106 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
3107 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
3108 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
3109 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
3110 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
3111 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
3112 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
3113 definitions accordingly.
3115 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
3116 Kuan-Lin Chen <kuanlinchentw@gmail.com>
3119 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
3122 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
3125 * ira-conflicts.c (ira_build_conflicts): Update total conflict
3126 hard regs for inner regno.
3128 2017-01-17 Martin Liska <mliska@suse.cz>
3131 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
3132 assumption and add comment.
3134 2017-01-17 Nathan Sidwell <nathan@acm.org>
3136 * ipa-visibility.c (localize_node): New function, broken out of ...
3137 (function_and_variable_visibility): ... here. Call it.
3139 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3142 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
3143 correctly set frequency of oudgoing edge.
3144 (duplicate_thread_path): Fix profile updating.
3146 2017-01-17 Jakub Jelinek <jakub@redhat.com>
3149 * configure.ac: Add GCC_BASE_VER.
3150 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
3151 version from BASE-VER file.
3152 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
3153 (gcc.o): Depend on $(BASEVER).
3154 * common.opt (dumpfullversion): New option.
3155 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
3156 * doc/invoke.texi: Document -dumpfullversion.
3157 * doc/install.texi: Document --with-gcc-major-version-only.
3158 * configure: Regenerated.
3160 2017-01-17 Richard Biener <rguenther@suse.de>
3162 PR tree-optimization/71433
3163 * tree-vrp.c (register_new_assert_for): Merge same asserts
3164 on all incoming edges.
3165 (process_assert_insertions_for): Handle insertions at the
3168 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
3170 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
3171 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
3173 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
3176 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
3179 2017-01-17 Alan Modra <amodra@gmail.com>
3182 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
3183 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
3184 symbolic stack limit when pic.
3186 2017-01-16 Martin Sebor <msebor@redhat.com>
3188 PR tree-optimization/78608
3189 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
3191 2017-01-16 Jeff Law <law@redhat.com>
3194 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
3195 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
3196 for several include directories that may be relative to sysroot.
3197 * config/i386/x-mingw32 (gplus_includedir): Define.
3198 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
3199 (native_system_includedir): Likewise.
3200 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
3201 override if TARGET_SYSTEM_ROOT is defined.
3202 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
3204 PR tree-optimization/79090
3205 PR tree-optimization/33562
3206 PR tree-optimization/61912
3207 PR tree-optimization/77485
3208 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
3209 and computed trims into the dump file.
3211 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
3213 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
3215 2017-01-16 Jakub Jelinek <jakub@redhat.com>
3218 * gimplify.c (gimplify_init_constructor): If want_value and
3219 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
3223 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
3224 sequence. Formatting fixes.
3225 (doloop_optimize): Formatting fixes.
3228 * gcc.c (debug_level_greater_than_spec_func): New function.
3229 (static_spec_functions): Add debug-level-gt spec function.
3230 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
3232 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
3233 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
3234 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
3235 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
3236 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
3237 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
3239 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
3241 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
3242 QImode fixups to general and mask registers only.
3244 2017-01-16 Carl Love <cel@us.ibm.com>
3246 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
3247 for built-in functions
3248 vector signed char vec_nabs (vector signed char)
3249 vector signed short vec_nabs (vector signed short)
3250 vector signed int vec_nabs (vector signed int)
3251 vector signed long long vec_nabs (vector signed long long)
3252 vector float vec_nabs (vector float)
3253 vector double vec_nabs (vector double)
3254 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
3256 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
3257 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
3258 * doc/extend.texi: Update the documentation file for the new built-in
3261 2017-01-16 Martin Sebor <msebor@redhat.com>
3263 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
3266 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3268 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
3269 UNSPEC_VSX__XXSPLTD to require special splat handling.
3271 2017-01-16 David Malcolm <dmalcolm@redhat.com>
3274 * system.h: Poison strndup.
3276 2017-01-16 Alan Modra <amodra@gmail.com>
3279 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
3282 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
3284 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
3286 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
3288 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
3289 call recog here. Assert that INSN_CODE (insn) is non-negative.
3291 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
3294 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
3296 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
3297 in the currently scheduled RTL fragment.
3299 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
3301 PR rtl-optimization/78751
3302 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
3305 2017-01-14 Jeff Law <law@redhat.com>
3307 PR tree-optimization/79090
3308 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
3309 variable length stores.
3310 (compute_trims): Delete dead assignment to *trim_tail.
3311 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
3314 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
3316 PR rtl-optimization/78626
3317 PR rtl-optimization/78727
3318 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
3319 of a block, and split such blocks after everything else is finished.
3321 2017-01-14 Alan Modra <amodra@gmail.com>
3324 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
3325 target legitimate_combined_insn.
3326 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
3327 (rs6000_legitimate_combined_insn): New function.
3328 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
3330 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
3331 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
3332 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
3334 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
3336 * doc/frontends.texi (G++ and GCC): Remove references to Java.
3338 2017-01-13 Jeff Law <law@redhat.com>
3340 PR tree-optimization/33562
3341 PR tree-optimization/61912
3342 PR tree-optimization/77485
3343 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
3345 (delete_dead_assignment): Likewise.
3346 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
3347 statement to delete_dead_call and delete_dead_assignment.
3349 2017-01-13 David Malcolm <dmalcolm@redhat.com>
3352 * substring-locations.c (format_warning_va): Strengthen case 1 so
3353 that both endpoints of the substring must be within the format
3354 range for just the substring to be printed.
3356 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
3358 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
3359 * config/i386/i386.c (ix86_target_string): Add missing options
3360 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
3361 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
3362 flags_other and ix86_target_other to flags2_other. Display unknown
3364 (ix86_valid_target_attribute_inner_p): Add missing options and
3365 reorder options by implied ISAs, as in ix86_target_string.
3367 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3369 * hash-table.h (hash_table::too_empty_p): New function.
3370 (hash_table::expand): Use it.
3371 (hash_table::traverse): Likewise.
3372 (hash_table::empty_slot): Use sizeof (value_type) instead of
3373 sizeof (PTR) to convert bytes to elements. Shrink the table
3374 if the current size is excessive for the current number of
3377 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
3379 * ira-costs.c (record_reg_classes): Break from the inner loop
3380 early once alt_fail is known to be true. Update outer loop
3381 handling accordingly.
3383 2017-01-13 Jeff Law <law@redhat.com>
3385 * tree-ssa-dse.c (decrement_count): New function.
3386 (increment_start_addr, maybe_trim_memstar_call): Likewise.
3387 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
3388 when we know the partially dead statement is a mem* function.
3390 PR tree-optimization/61912
3391 PR tree-optimization/77485
3392 * tree-ssa-dse.c: Include expr.h.
3393 (maybe_trim_constructor_store): New function.
3394 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
3396 PR tree-optimization/33562
3397 PR tree-optimization/61912
3398 PR tree-optimization/77485
3399 * doc/invoke.texi: Document new dse-max-object-size param.
3400 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
3401 * tree-ssa-dse.c: Include params.h.
3402 (dse_store_status): New enum.
3403 (initialize_ao_ref_for_dse): New, partially extracted from
3405 (valid_ao_ref_for_dse, normalize_ref): New.
3406 (setup_live_bytes_from_ref, compute_trims): Likewise.
3407 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
3408 (maybe_trim_partially_dead_store): Likewise.
3409 (maybe_trim_complex_store): Likewise.
3410 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
3411 Track what bytes live from the original store. Return tri-state
3412 for dead, partially dead or live.
3413 (dse_dom_walker): Add constructor, destructor and new private members.
3414 (delete_dead_call, delete_dead_assignment): New extracted from
3416 (dse_optimize_stmt): Make a member of dse_dom_walker.
3417 Use initialize_ao_ref_for_dse.
3419 PR tree-optimization/33562
3420 PR tree-optimization/61912
3421 PR tree-optimization/77485
3422 * sbitmap.h (bitmap_count_bits): Prototype.
3423 (bitmap_clear_range, bitmap_set_range): Likewise.
3424 * sbitmap.c (bitmap_clear_range): New function.
3425 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
3427 2017-01-13 Martin Liska <mliska@suse.cz>
3430 * function.c (set_cfun): Add new argument force.
3431 * function.h (set_cfun): Likewise.
3432 * ipa-inline-transform.c (inline_call): Use the function when
3433 strict alising from is dropped for function we inline to.
3435 2017-01-13 Richard Biener <rguenther@suse.de>
3437 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
3438 for dumping GIMPLE INTEGER_CSTs.
3440 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3442 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
3443 to 201112L since C++17.
3445 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
3448 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
3449 if -fsanitize=kernel-address is present.
3451 2017-01-13 Richard Biener <rguenther@suse.de>
3453 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
3454 as _Literal ( type ) number in case usual suffixes do not
3455 preserve all information.
3457 2017-01-13 Richard Biener <rguenther@suse.de>
3459 PR tree-optimization/77283
3460 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
3461 and ssa-iterators.h.
3462 (is_feasible_trace): Implement a cost model based on joiner
3465 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3468 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
3469 char or short to __float128/_Float128 directly.
3471 2017-01-12 Martin Sebor <msebor@redhat.com>
3473 to -Wformat-overflow.
3474 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
3475 (min_bytes_remaining): Same.
3476 (get_string_length): Same.
3477 (format_string): Same.
3478 (format_directive): Same.
3480 (pass_sprintf_length::handle_gimple_call): Same.
3482 2017-01-12 Jakub Jelinek <jakub@redhat.com>
3484 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
3485 info.nowrite calls with no lhs that can't throw. Return bool
3486 whether gsi_remove has been called or not.
3487 (pass_sprintf_length::handle_gimple_call): Return bool whether
3488 try_substitute_return_value called gsi_remove. Formatting fix.
3489 (pass_sprintf_length::execute): Don't use gsi_remove if
3490 handle_gimple_call returned true.
3493 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
3494 be removed due to side-effects, don't remove following barrier nor
3495 turn the successor edge into fallthru edge.
3497 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3500 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
3501 element-reversing loads and stores as not swappable.
3503 2017-01-12 Nathan Sidwell <nathan@acm.org>
3504 Nicolai Stange <nicstange@gmail.com>
3506 * combine.c (try_combine): Don't ignore result of overlap checking
3507 loop. Combine overlap & asm check into single loop.
3509 2017-01-12 Richard Biener <rguenther@suse.de>
3511 * tree-pretty-print.c (dump_generic_node): Provide -gimple
3512 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
3514 2017-01-12 Richard Biener <rguenther@suse.de>
3516 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
3517 and TS_TARGET_OPTION directly derive from TS_BASE.
3518 * tree-core.h (tree_optimization_option): Derive from tree_base.
3519 (tree_target_option): Likewise.
3521 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
3523 * config/i386/i386.c (memory_address_length): Increase len
3524 only when rip_relative_addr_p returns false.
3526 2017-01-11 Julia Koval <julia.koval@intel.com>
3528 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
3529 (OPTION_MASK_ISA_SGX_SET): New.
3530 (ix86_handle_option): Handle OPT_msgx.
3531 * config.gcc: Added sgxintrin.h.
3532 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
3533 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
3534 * config/i386/i386.c (ix86_target_string): Add -msgx.
3536 (ix86_option_override_internal): Handle new options.
3537 (ix86_valid_target_attribute_inner_p): Add sgx.
3538 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
3539 * config/i386/i386.opt: Add msgx.
3540 * config/i386/sgxintrin.h: New file.
3541 * config/i386/x86intrin.h: Add sgxintrin.h.
3543 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3546 * fold-const.c (maybe_nonzero_address): Return 1 for function
3548 (tree_single_nonzero_warnv_p): Don't handle function local objects
3552 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
3555 2017-01-11 David Malcolm <dmalcolm@redhat.com>
3558 * opts.c: Include "spellcheck.h"
3559 (struct string_fragment): New struct.
3560 (struct edit_distance_traits<const string_fragment &>): New
3562 (get_closest_sanitizer_option): New function.
3563 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
3565 2017-01-11 Jakub Jelinek <jakub@redhat.com>
3567 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
3569 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
3570 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
3571 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
3572 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
3573 for initial die_offset if dwarf_split_debug_info.
3574 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
3575 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
3577 (output_skeleton_debug_sections): Formatting fix. Use
3578 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
3579 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
3581 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
3583 * config/arm/cortex-a53.md: Add bypasses for
3585 (cortex_a53_r2f): Only use for transfers.
3586 (cortex_a53_f2r): Likewise.
3587 (cortex_a53_r2f_cvt): Add reservation for conversions.
3588 (cortex_a53_f2r_cvt): Likewise.
3590 2017-01-11 Tamar Christina <tamar.christina@arm.com>
3592 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
3593 to all inlined functions, change static to extern.
3595 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
3598 * config/arm/arm.c (legitimize_pic_address): Handle reference to
3600 (arm_assemble_integer): Likewise.
3602 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
3604 * config.gcc: Use new awk script to check CPU, FPU and architecture
3605 parameters for --with-... options.
3606 * config/arm/parsecpu.awk: New file
3607 * config/arm/arm-cpus.in: New file.
3608 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
3610 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
3612 * config/arm/t-arm: Update dependency rules.
3613 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
3614 of processing .def files.
3615 * config/arm/genopt.sh: Deleted.
3616 * config/arm/gentune.sh: Deleted.
3617 * config/arm/arm-cores.def: Deleted.
3618 * config/arm/arm-arches.def: Deleted.
3619 * config/arm/arm-fpus.def: Deleted.
3620 * config/arm/arm-tune.md: Regenerated.
3621 * config/arm/arm-tables.opt: Regenerated.
3622 * config/arm/arm-cpu.h: New generated file.
3623 * config/arm/arm-cpu-data.h: New generated file.
3624 * config/arm/arm-cpu-cdata.h: New generated file.
3626 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
3629 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
3631 (input_varpool_node): Unpack dynamically_initialized bit.
3633 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
3635 PR rtl-optimization/79032
3636 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
3637 the alignment of the adjusted memory reference against that of MODE,
3638 instead of the alignment of the original memory reference.
3640 2017-01-11 Martin Jambor <mjambor@suse.cz>
3642 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
3644 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
3645 decorated functions.
3647 2017-01-11 Richard Biener <rguenther@suse.de>
3649 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
3650 set range/nonnull info for PHI results. Do not set it on
3651 stmts marked for removal.
3653 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
3655 * expr.c (store_field): In the bitfield case, fetch the return value
3656 from the registers before applying a single big-endian adjustment.
3657 Always do a final load for a BLKmode value not larger than a word.
3659 2017-01-10 David Malcolm <dmalcolm@redhat.com>
3662 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3663 that we correctly handle column numbers greater than
3664 LINE_MAP_MAX_COLUMN_NUMBER.
3666 2017-01-10 Martin Sebor <msebor@redhat.com>
3669 * gimple-ssa-sprintf.c (get_destination_size): Call
3670 {init,fini}object_sizes.
3671 * tree-object-size.c (addr_object_size): Adjust.
3672 (pass_through_call): Adjust.
3673 (pass_object_sizes::execute): Adjust.
3674 * tree-object-size.h (fini_object_sizes): Declare.
3676 2017-01-10 Martin Sebor <msebor@redhat.com>
3678 PR tree-optimization/78775
3679 * builtins.c (get_size_range): Move...
3680 * calls.c: ...to here.
3681 (alloc_max_size): Accept zero argument.
3682 (operand_signed_p): Remove.
3683 (maybe_warn_alloc_args_overflow): Call get_size_range.
3684 * calls.h (get_size_range): Declare.
3686 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
3688 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
3689 from TI's devices.csv file as of September 2016.
3690 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
3692 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
3694 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
3695 * doc/invoke.texi: Likewise.
3696 * doc/md.texi: Likewise.
3697 * doc/objc.texi: Likewise.
3699 2017-01-10 Joshua Conner <joshconner@google.com>
3701 * config/arm/fuchsia-elf.h: New file.
3702 * config/fuchsia.h: New file.
3703 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
3704 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
3706 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
3708 2016-01-10 Richard Biener <rguenther@suse.de>
3710 PR tree-optimization/79034
3711 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
3712 Propagate out degenerate PHIs in the joiner.
3714 2017-01-10 Martin Liska <mliska@suse.cz>
3716 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
3717 (sort_congruence_classes_by_decl_uid): Likewise.
3718 (sort_congruence_class_groups_by_decl_uid): Likewise.
3719 (sem_item_optimizer::merge_classes): Sort class, groups in these
3720 classes and members in the groups by DECL_UID of declarations.
3721 This would make merge operations stable.
3723 2017-01-10 Martin Liska <mliska@suse.cz>
3725 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
3726 usage of m_classes_vec.
3727 (sem_item_optimizer::~sem_item_optimizer): Likewise.
3728 (sem_item_optimizer::get_group_by_hash): Likewise.
3729 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
3730 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
3731 (sem_item_optimizer::verify_classes): Likewise.
3732 (sem_item_optimizer::process_cong_reduction): Likewise.
3733 (sem_item_optimizer::dump_cong_classes): Likewise.
3734 (sem_item_optimizer::merge_classes): Likewise.
3735 * ipa-icf.h (congruence_class_hash): Rename from
3736 congruence_class_group_hash. Remove declaration of m_classes_vec.
3738 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
3740 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
3741 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
3742 * config.gcc: Add avx512vpopcntdqintrin.h.
3743 * config/i386/avx512vpopcntdqintrin.h: New.
3744 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
3745 * config/i386/i386-builtin-types.def: Add new types.
3746 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
3747 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
3748 __builtin_ia32_vpopcountq_v8di_mask): New.
3749 * config/i386/i386-c.c (ix86_target_macros_internal): Define
3750 __AVX512VPOPCNTDQ__.
3751 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
3752 (PTA_AVX512VPOPCNTDQ): Define.
3753 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
3754 TARGET_AVX512VPOPCNTDQ_P): Define.
3755 * config/i386/i386.opt: Add mavx512vpopcntdq.
3756 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
3757 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
3759 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3762 * predict.def (PRED_CALL): Set to 67.
3764 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
3766 * expr.c (store_field): In the bitfield case, if the value comes from
3767 a function call and is of an aggregate type returned in registers, do
3768 not modify the field mode; extract the value in all cases if the mode
3769 is BLKmode and the size is not larger than a word.
3771 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
3774 * config/i386/cpuid.h: Fix undefined behavior.
3776 2017-01-04 Jeff Law <law@redhat.com>
3778 PR tree-optimization/79007
3779 PR tree-optimization/67955
3780 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
3781 conservative for pt.null when flag_non_call_exceptions is on.
3783 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3785 PR translation/79019
3786 PR translation/79020
3787 * params.def (PARAM_INLINE_MIN_SPEEDUP,
3788 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
3789 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
3791 * config/avr/avr.opt (maccumulate-args): Likewise.
3792 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
3793 * common.opt (freport-bug): Likewise.
3794 * cif-code.def (CIF_FINAL_ERROR): Likewise.
3795 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
3796 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
3797 translatable string.
3798 * config/i386/i386.c (function_value_32): Likewise.
3799 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
3800 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
3802 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
3803 * common/config/msp430/msp430-common.c (msp430_handle_option):
3805 * symtab.c (symtab_node::verify_base): Likewise.
3806 * opts.c (set_debug_level): Likewise.
3807 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
3808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
3809 missing whitespace to translatable strings.
3810 * config/avr/avr.md (bswapsi2): Fix typo in comment.
3811 * config/sh/superh.h: Likewise.
3812 * config/i386/xopintrin.h: Likewise.
3813 * config/i386/znver1.md: Likewise.
3814 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
3815 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
3816 * double-int.h (struct double_int): Likewise.
3817 * double-int.c (div_and_round_double): Likewise.
3818 * wide-int.cc: Likewise.
3819 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
3820 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
3821 * cfgcleanup.c (crossjumps_occured): Renamed to ...
3822 (crossjumps_occurred): ... this.
3823 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
3826 PR tree-optimization/78899
3827 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
3828 returning bool return struct loop *, NULL for failure and the new
3830 (versionable_outer_loop_p): Don't version outer loop if it has
3831 dont_vectorized bit set.
3832 (tree_if_conversion): When versioning outer loop, ensure
3833 tree_if_conversion is performed also on the inner loop of the
3834 non-vectorizable outer loop copy.
3835 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
3836 LOOP_VECTORIZED in inner loop of the scalar outer loop and
3837 prevent vectorization of it.
3838 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
3839 the outer loop vectorization of the non-scalar version is attempted
3840 before vectorization of the inner loop in scalar version. If
3841 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
3842 vectorization of its inner loop.
3843 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
3844 has 2 inner loops, rename also on edges from bb whose single pred
3845 is outer_loop->header. Fix typo in function comment.
3847 2017-01-09 Martin Sebor <msebor@redhat.com>
3850 * asan.c (asan_emit_stack_protection): Increase local buffer size
3851 to avoid snprintf truncation warning.
3853 2017-01-09 Andrew Pinski <apinski@cavium.com>
3855 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
3856 to reference thunderx2t99 for the tuning structure
3857 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
3859 (thunderx2t99_extra_costs): This.
3860 * config/aarch64/aarch64-tune.md: Regenerate.
3861 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
3862 (vulcan_addrcost_table): This.
3863 (vulcan_regmove_cost): Rename to ...
3864 (thunderx2t99_regmove_cost): This.
3865 (vulcan_vector_cost): Rename to ...
3866 (thunderx2t99_vector_cost): this.
3867 (vulcan_branch_cost): Rename to ...
3868 (thunderx2t99_branch_cost): This.
3869 (vulcan_tunings): Rename to ...
3870 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
3871 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
3873 2017-01-09 Martin Jambor <mjambor@suse.cz>
3877 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
3878 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
3879 (propagate_vr_accross_jump_function): Use the above function for all
3880 value range computations for pass-through jump functions and type
3881 converasion from explicit value range values.
3882 (ipcp_propagate_stage): Do not attempt to deduce types of formal
3883 parameters from TYPE_ARG_TYPES.
3884 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
3885 (ipa_write_node_info): Stream type of the actual argument.
3886 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
3888 2017-01-09 Martin Liska <mliska@suse.cz>
3891 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
3892 (lookup_compiler): Do not show error message with have_E.
3894 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3896 PR tree-optimization/78938
3897 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
3898 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
3899 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
3900 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
3903 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3905 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
3908 2017-01-09 Richard Biener <rguenther@suse.de>
3910 PR tree-optimization/78997
3911 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
3912 name condition properly.
3914 2017-01-09 Richard Biener <rguenther@suse.de>
3917 * dwarf2out.c (is_cxx): New overload with context.
3918 (is_naming_typedef_decl): Use it.
3920 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3922 * invoke.texi (Option Summary): Correct spacing in option lists
3923 and add line breaks to fix over-long lines.
3925 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3929 * extend.texi (Common Variable Attributes): Add xref to GCC
3930 Internals manual to explain mode attribute keywords.
3932 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3935 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
3936 and Preprocessor Options.
3937 (Options for Linking): Document -pthread here....
3938 (RS/6000 and PowerPC Options): ...not here.
3939 (Solaris 2 Options): ...or here.
3940 * doc/cppopts.texi: Document -pthread.
3942 2017-01-08 Martin Sebor <msebor@redhat.com>
3945 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
3946 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
3947 New member functions.
3948 (format_directive): Used them.
3950 (pass_sprintf_length::handle_gimple_call): Same.
3951 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
3952 to avoid truncation for any argument.
3953 (extract_affine_mul): Same.
3954 * tree.c (get_file_function_name): Same.
3956 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3959 * predict.def (PRED_INDIR_CALL): Set to 86.
3961 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3963 PR preprocessor/54124
3964 * doc/cppopts.texi: Reformat -d subtable to list the full name
3965 of the options. Add cross-reference to the docs for the general
3966 compiler -d options.
3967 * doc/invoke.texi (Developer Options): Add cross-reference to the
3968 preprocessor-specific -d option documentation.
3970 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3972 PR preprocessor/13498
3973 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
3974 redudant material, and reflect new command-line options.
3975 (System Headers): Likewise.
3977 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3979 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
3980 -isystem, and -idirafter. Copy-edit.
3981 * doc/cppopts.texi: Copy-edit. Remove contradiction about
3982 default for -ftrack-macro-expansion. Delete obsolete and
3983 badly-formatted implementation details about -fdebug-cpp output.
3984 * doc/cppwarnopts.texi: Copy-edit.
3986 2017-01-07 David Malcolm <dmalcolm@redhat.com>
3989 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3990 that the transition from a max line width >= 1<<10 to narrower
3991 lines works correctly.
3993 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
3995 * doc/options.texi (PerFunction): New.
3996 * opt-functions.awk (switch_flags): Map both Optimization and
3997 PerFunction to CL_OPTIMIZATION.
3998 * opth-gen.awk: Test for PerFunction flag along with
4000 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
4001 it only when the latter is present. Skip those that don't in
4002 the hash function generator.
4003 * common.opt (fvar-tracking): Mark as PerFunction instead of
4005 (fvar-tracking-assignments): Likewise.
4006 (fvar-tracking-assignments-toggle): Likewise.
4007 (fvar-tracking-uninit): Likewise.
4009 2017-01-07 Jakub Jelinek <jakub@redhat.com>
4011 PR translation/79018
4012 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
4015 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
4018 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
4019 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
4020 TARGET_LONG_JUMP_TABLE_OFFSETS.
4021 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
4022 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
4023 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
4024 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
4025 * config/m68k/m68k.md (tablejump expander): Likewise.
4026 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
4027 TARGET_LONG_JUMP_TABLE_OFFSETS.
4028 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
4029 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
4031 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
4032 David Holsgrove <david.holsgrove@xilinx.com>
4034 * common/config/microblaze/microblaze-common.c
4035 (TARGET_EXCEPT_UNWIND_INFO): Remove.
4036 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
4038 * config/microblaze/microblaze.c (microblaze_must_save_register)
4039 (microblaze_expand_epilogue, microblaze_return_addr): Handle
4041 (microblaze_eh_return): New function.
4042 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
4043 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
4044 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
4045 * config/microblaze/microblaze.md (eh_return): New pattern.
4047 2017-01-06 Jakub Jelinek <jakub@redhat.com>
4049 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
4050 GCC_DIAGNOSTIC_STRINGIFY): Define.
4052 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
4054 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4056 * config/arm/arm.md (<mcrr>): New.
4058 * config/arm/arm.c (arm_arch5te): New.
4059 (arm_option_override): Set arm_arch5te.
4060 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
4062 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
4063 (arm_mcrr_qualifiers): ... this. New.
4064 (MRRC_QUALIFIERS): Define to...
4065 (arm_mrrc_qualifiers): ... this. New.
4066 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
4068 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
4069 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
4070 (MRRCI, mrrc, MRRC): New.
4071 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
4072 VUNSPEC_MRRC2): New.
4074 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4076 * config/arm/arm.md (<mcr>): New.
4078 * config/arm/arm.c (arm_coproc_builtin_available): Add
4079 support for mcr, mrc, mcr2 and mrc2.
4080 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
4081 (arm_mcr_qualifiers): ... this. New.
4082 (MRC_QUALIFIERS): Define to ...
4083 (arm_mrc_qualifiers): ... this. New.
4084 (MCR_QUALIFIERS): Define to ...
4085 (arm_mcr_qualifiers): ... this. New.
4086 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
4088 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
4089 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
4090 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
4093 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4095 * config/arm/arm.md (*ldc): New.
4099 * config/arm/arm.c (arm_coproc_builtin_available): Add
4100 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
4101 (arm_coproc_ldc_stc_legitimate_address): New.
4102 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
4103 'qualifier_const_pointer'.
4104 (LDC_QUALIFIERS): Define to...
4105 (arm_ldc_qualifiers): ... this. New.
4106 (STC_QUALIFIERS): Define to...
4107 (arm_stc_qualifiers): ... this. New.
4108 * config/arm/arm-protos.h
4109 (arm_coproc_ldc_stc_legitimate_address): New.
4110 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
4111 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
4112 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
4113 stc2, stcl, stc2l): New.
4114 * config/arm/constraints.md (Uz): New.
4115 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
4116 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
4117 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
4118 VUNSPEC_STC2L): New.
4120 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4122 * config/arm/arm.md (<cdp>): New.
4123 * config/arm/arm.c (neon_const_bounds): Rename this ...
4124 (arm_const_bounds): ... this.
4125 (arm_coproc_builtin_available): New.
4126 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
4127 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
4128 (CDP_QUALIFIERS): Define to...
4129 (arm_cdp_qualifiers): ... this. New.
4131 (arm_expand_builtin_args): Add case for 6 arguments.
4132 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
4133 (arm_const_bounds): ... this.
4134 (arm_coproc_builtin_available): New.
4135 * config/arm/arm_acle.h (__arm_cdp): New.
4137 * config/arm/arm_acle_builtins.def (cdp): New.
4139 * config/arm/iterators.md (CDPI,CDP,cdp): New.
4140 * config/arm/neon.md: Rename all 'neon_const_bounds' to
4142 * config/arm/types.md (coproc): New.
4143 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
4144 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
4145 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
4146 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
4148 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4150 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
4151 (UBINOP_QUALIFIERS): New.
4153 (acle_builtin_data): New. Change comment.
4154 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
4155 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
4156 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
4157 arm_acle_builtins.def.
4158 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
4159 (arm_init_acle_builtins): New.
4160 (CRC32_BUILTIN): Remove.
4161 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
4162 crc32cb, crc32ch and crc32cw.
4163 (arm_init_crc32_builtins): Remove.
4164 (arm_init_builtins): Use arm_init_acle_builtins rather
4165 than arm_init_crc32_builtins.
4166 (arm_expand_acle_builtin): New.
4167 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
4168 * config/arm/arm_acle_builtins.def: New.
4170 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
4172 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
4173 (arm_builtin_datum): ... this.
4174 (arm_init_neon_builtin): Rename to ...
4175 (arm_init_builtin): ... this. Add a new parameters PREFIX
4176 and USE_SIG_IN_NAME.
4177 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
4178 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
4179 'arm_builtin_datum'.
4180 (arm_init_vfp_builtins): Likewise.
4181 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
4182 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
4183 (arm_expand_neon_args): Rename to ...
4184 (arm_expand_builtin_args): ... this. Rename builtin_arg
4185 enum values and differentiate between ARG_BUILTIN_MEMORY
4186 and ARG_BUILTIN_NEON_MEMORY.
4187 (arm_expand_neon_builtin_1): Rename to ...
4188 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
4189 values, arm_expand_builtin_args and add bool parameter NEON.
4190 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
4191 (arm_expand_vfp_builtin): Likewise.
4192 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
4194 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4197 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
4198 * predict.c (tree_estimate_probability_bb): Reverse direction of
4199 polymorphic call predictor.
4201 2017-01-06 David Malcolm <dmalcolm@redhat.com>
4203 * passes.c (execute_one_pass): Split out pass-skipping logic into...
4204 (determine_pass_name_match): ...this new function and...
4205 (should_skip_pass_p): ...this new function.
4207 2017-01-06 Nathan Sidwell <nathan@acm.org>
4209 * ipa-visibility.c (function_and_variable_visibility): Reformat
4210 comments and long lines. Remove extrneous if.
4211 * symtab.c (symtab_node::make_decl_local): Fix code format.
4212 (symtab_node::set_section_for_node): Fix comment typo.
4214 2017-01-06 Martin Liska <mliska@suse.cz>
4217 * lra-constraints.c: Rename invariant to lra_invariant.
4218 * predict.c (set_even_probabilities): Initialize e to NULL.
4220 2017-01-05 Martin Sebor <msebor@redhat.com>
4222 PR tree-optimization/78910
4223 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
4224 (format_integer): Correct off-by-one error in the handling
4225 of precision with negative numbers in signed conversions..
4227 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
4229 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
4231 2017-01-05 Jakub Jelinek <jakub@redhat.com>
4233 PR tree-optimization/71016
4234 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
4235 factor_out_conditional_conversion. Formatting fix.
4236 (factor_out_conditional_conversion): Add cond_stmt argument.
4237 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
4238 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
4241 2017-01-05 David Malcolm <dmalcolm@redhat.com>
4243 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
4244 read-rtl-function.o, and selftest-rtl.o.
4245 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
4246 (selftest::aarch64_test_loading_full_dump): New function.
4247 (selftest::aarch64_run_selftests): New function.
4248 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
4249 selftest::aarch64_run_selftests.
4250 * config/i386/i386.c
4251 (selftest::ix86_test_loading_dump_fragment_1): New function.
4252 (selftest::ix86_test_loading_call_insn): New function.
4253 (selftest::ix86_test_loading_full_dump): New function.
4254 (selftest::ix86_test_loading_unspec): New function.
4255 (selftest::ix86_run_selftests): Call the new functions.
4256 * emit-rtl.c (maybe_set_max_label_num): New function.
4257 * emit-rtl.h (maybe_set_max_label_num): New decl.
4258 * function.c (instantiate_decls): Guard call to
4259 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
4260 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
4262 * gensupport.c (gen_reader::gen_reader): Pass "false"
4263 for new "compact" param of rtx_reader.
4264 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
4265 rather than an empty string for NULL strings.
4266 * read-md.c: Potentially include config.h rather than bconfig.h.
4267 Wrap include of errors.h with #ifdef GENERATOR_FILE.
4268 (have_error): New global, copied from errors.c.
4269 (md_reader::read_name): Rename to...
4270 (md_reader::read_name_1): ...this, adding "out_loc" param,
4271 and converting "missing name or number" to returning false, rather
4273 (md_reader::read_name): Reimplement in terms of read_name_1.
4274 (md_reader::read_name_or_nil): New function.
4275 (md_reader::read_string): Handle "(nil)" by returning NULL.
4276 (md_reader::md_reader): Add new param "compact".
4277 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
4278 (md_reader::read_file): New method.
4279 * read-md.h (md_reader::md_reader): Add new param "compact".
4280 (md_reader::read_file): New method.
4281 (md_reader::is_compact): New accessor.
4282 (md_reader::read_name): Convert return type from void to file_location.
4283 (md_reader::read_name_or_nil): New decl.
4284 (md_reader::read_name_1): New decl.
4285 (md_reader::m_compact): New field.
4286 (noop_reader::noop_reader): Pass "false" for new "compact" param
4288 (rtx_reader::rtx_reader): Add new "compact" param.
4289 (rtx_reader::read_rtx_operand): Make virtual and convert return
4290 type from void to rtx.
4291 (rtx_reader::read_until): New decl.
4292 (rtx_reader::handle_any_trailing_information): New virtual function.
4293 (rtx_reader::postprocess): New virtual function.
4294 (rtx_reader::finalize_string): New virtual function.
4295 (rtx_reader::m_in_call_function_usage): New field.
4296 (rtx_reader::m_reuse_rtx_by_id): New field.
4297 * read-rtl-function.c: New file.
4298 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
4299 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
4300 (selftest::verify_three_block_rtl_cfg): New decl.
4301 * read-rtl-function.h: New file.
4302 * read-rtl.c: Potentially include config.h rather than bconfig.h.
4303 For host, include function.h, memmodel.h, and emit-rtl.h.
4304 (one_time_initialization): New function.
4305 (struct compact_insn_name): New struct.
4306 (compact_insn_names): New array.
4307 (find_code): Handle insn codes in compact dumps.
4308 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
4309 (bind_subst_iter_and_attr): Likewise.
4310 (add_condition_to_string): Likewise.
4311 (add_condition_to_rtx): Likewise.
4312 (apply_attribute_uses): Likewise.
4313 (add_current_iterators): Likewise.
4314 (apply_iterators): Likewise.
4315 (initialize_iterators): Guard usage of apply_subst_iterator with
4316 #ifdef GENERATOR_FILE.
4317 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
4318 (md_reader::read_mapping): Likewise.
4319 (add_define_attr_for_define_subst): Likewise.
4320 (add_define_subst_attr): Likewise.
4321 (read_subst_mapping): Likewise.
4322 (check_code_iterator): Likewise.
4323 (rtx_reader::read_rtx): Likewise. Move one-time initialization
4325 (one_time_initialization): New function.
4326 (rtx_reader::read_until): New method.
4327 (read_flags): New function.
4328 (parse_reg_note_name): New function.
4329 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
4330 Handle reuse_rtx ids.
4331 Wrap iterator lookup within #ifdef GENERATOR_FILE.
4332 Add parsing support for RTL dumps, mirroring the special-cases in
4333 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
4334 values, and calling handle_any_trailing_information.
4335 (rtx_reader::read_rtx_operand): Convert return type from void
4336 to rtx, returning return_rtx. Handle case 'e'. Call
4337 finalize_string on XSTR and XTMPL fields.
4338 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
4339 "(nil)" values were omitted. Call the postprocess vfunc on the
4341 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
4342 class ctor. Initialize m_in_call_function_usage. Call
4343 one_time_initialization.
4344 * rtl-tests.c (selftest::test_uncond_jump): Call
4345 set_new_first_and_last_insn.
4346 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
4347 * selftest-rtl.c: New file.
4348 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
4349 (selftest::get_insn_by_uid): New decl.
4350 * selftest-run-tests.c (selftest::run_tests): Call
4351 read_rtl_function_c_tests.
4352 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
4353 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
4356 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
4358 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
4359 operands in a special way. Assert that pos+len <= mode precision.
4361 2017-01-05 Jakub Jelinek <jakub@redhat.com>
4363 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
4364 3 argument Alias with unlimited for the negative form.
4365 (fno-vect-cost-model): Removed.
4367 2017-01-05 Martin Liska <mliska@suse.cz>
4369 * hsa-gen.c (gen_hsa_divmod): New function.
4370 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
4372 2017-01-05 Martin Liska <mliska@suse.cz>
4375 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
4378 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4380 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
4381 small constant length operands.
4383 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4385 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
4386 between loop iterations.
4388 2017-01-05 Martin Liska <mliska@suse.cz>
4391 * gimplify.c (gimplify_decl_expr): Compare to
4392 asan_poisoned_variables instread of checking flags.
4393 (gimplify_target_expr): Likewise.
4394 (gimplify_expr): Likewise.
4395 (gimplify_function_tree): Conditionally initialize
4396 asan_poisoned_variables.
4398 2017-01-04 Jeff Law <law@redhat.com>
4400 PR tree-optimizatin/78812
4401 * rtl.h (contains_mem_rtx_p): Prototype.
4402 * ifcvt.c (containts_mem_rtx_p): Move from here to...
4403 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
4404 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
4405 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
4406 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
4408 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4410 * input.c (assert_char_at_range): Default-initialize actual_range.
4412 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4414 * df-scan.c (df_ref_create_structure): Make regno unsigned,
4415 to match the caller.
4417 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4419 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
4420 insns after final jump in test to emit dummy move.
4422 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4424 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
4425 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
4427 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
4429 * multiple_target.c (create_dispatcher_calls): Init e_next.
4430 * tree-ssa-loop-split.c (split_loop): Init border.
4431 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
4434 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
4439 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
4440 (altivec_register_operand): Do not return true if the operand
4441 contains a SUBREG mixing SImode and SFmode.
4442 (vsx_register_operand): Likewise.
4443 (vsx_reg_sfsubreg_ok): New predicate.
4444 (vfloat_operand): Do not return true if the operand contains a
4445 SUBREG mixing SImode and SFmode.
4446 (vint_operand): Likewise.
4447 (vlogical_operand): Likewise.
4448 (gpc_reg_operand): Likewise.
4449 (int_reg_operand): Likewise.
4450 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
4451 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
4452 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
4454 (rs6000_emit_move_si_sf_subreg): New helper function.
4455 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
4456 fixup SUBREGs involving SImode and SFmode.
4457 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
4458 numbers for the new peephole2 optimization.
4459 (peephole2 for SFmode unions): New peephole2 to optimize cases in
4460 the GLIBC math library that do AND/IOR/XOR operations on single
4461 precision floating point.
4462 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
4463 target macros to say whether we need to avoid SUBREGs mixing
4465 (TARGET_ALLOW_SF_SUBREG): Likewise.
4466 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
4467 (UNSPEC_SI_FROM_SF): Likewise.
4468 (iorxor): Change spacing.
4469 (and_ior_xor): New iterator for AND, IOR, and XOR.
4470 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
4471 (movdi_from_sf_zero_ext): Likewise.
4472 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
4473 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
4474 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
4475 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
4476 (fms<mode>4): Likewise.
4477 (fnma<mode>4): Likewise.
4478 (fnms<mode>4): Likewise.
4479 (nfma<mode>4): Likewise.
4480 (nfms<mode>4): Likewise.
4482 2017-01-04 Marek Polacek <polacek@redhat.com>
4485 * doc/invoke.texi: Document -Wpointer-compare.
4487 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4489 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
4492 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
4493 descriptions for -gdwarf-5 and emit them as uleb128 instead of
4496 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
4499 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
4500 documentation of the powerpc_popcntb_ok attribute.
4501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4502 code to issue warning messages if a requested CPU configuration is
4503 not supported by the binary (assembler and loader) toolchain.
4504 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
4505 made to define a built-in function that has been disabled.
4506 (paired_init_builtins): Add assertion to prevent ICE if attempt is
4507 made to define a built-in function that has been disabled.
4508 (altivec_init_builtins): Add comment explaining why definition
4509 of the DST built-in functions is not preceded by an assertion
4510 check. Add assertions to prevent ICE if attempts are made to
4511 define an altivec predicate or an abs* built-in function that has
4513 (htm_init_builtins): Add comment explaining why definition of the
4514 htm built-in functions is not preceded by an assertion check.
4516 2017-01-04 Jeff Law <law@redhat.com>
4518 PR tree-optimizatin/67955
4519 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
4520 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
4521 the points-to solution does not include pt_null. Use DECL_PT_UID
4524 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
4526 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
4527 Use gen_int_mode instead of gen_lopwart for const_int operands.
4529 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4531 PR tree-optimization/71563
4532 * match.pd: Simplify X << Y into X if Y is known to be 0 or
4533 out of range value - has low bits known to be zero.
4535 2017-01-04 Alan Modra <amodra@gmail.com>
4537 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
4538 * configure: Regenerate.
4539 * config.in: Regenerate.
4541 2017-01-04 Jakub Jelinek <jakub@redhat.com>
4544 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
4545 a substring of the message, but strcmp with the whole message. Ifdef
4546 ENABLE_NLS, translate the message first using dgettext.
4548 2017-01-03 Jeff Law <law@redhat.com>
4550 PR tree-optimizatin/78856
4551 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
4552 (mark_threaded_blocks): Remove code to truncate thread paths that
4553 cross multiple loop headers. Instead invalidate the cached loop
4554 iteration information and handle case of a thread path walking
4555 into an irreducible region.
4557 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4560 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
4561 assertions. Add support for doing the signbit if the IEEE 128-bit
4562 floating point value is in a GPR.
4563 * config/rs6000/rs6000.md (Fsignbit): Delete.
4564 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
4565 Update the length attribute if the value is in a GPR.
4566 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
4567 the sign or zero extension instruction, since the value is always 0/1.
4568 (signbit<mode>2_dm2): Delete using <Fsignbit>.
4571 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
4572 extracting SImode to a GPR register so that we can generate a
4573 store, limit the vector to be in a traditional Altivec register
4574 for the vextuwrx instruction.
4576 2017-01-03 Ian Lance Taylor <iant@google.com>
4578 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
4580 2017-01-03 Martin Sebor <msebor@redhat.com>
4582 PR tree-optimization/78696
4583 * gimple-ssa-sprintf.c (format_floating): Correct handling of
4584 precision. Use MPFR for %f for greater fidelity. Correct handling
4586 (pass_sprintf_length::compute_format_length): Set width and precision
4587 specified by asrerisk to void_node for vararg functions.
4588 (try_substitute_return_value): Adjust dump output.
4590 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
4592 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
4594 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
4596 * doc/invoke.texi (SPARC options): Document -mlra as the default.
4597 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
4598 -mlra/-mno-lra was passed to the compiler.
4600 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
4602 PR rtl-optimization/65618
4603 * emit-rtl.c (try_split): Move initialization of "before" and
4604 "after" to just before the call to emit_insn_after_setloc.
4606 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
4608 * doc/md.texi (Standard Names): Remove reference to Java frontend.
4610 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
4612 * dwarf2out.c (gen_enumeration_type_die): When
4613 -gno-strict-dwarf, add a DW_AT_encoding attribute.
4615 2017-01-03 Jakub Jelinek <jakub@redhat.com>
4617 PR tree-optimization/78965
4618 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
4619 Change first argument from const call_info & to call_info &. For %n
4620 set info.nowrite to false.
4623 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
4624 possibly throwing calls.
4626 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
4627 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
4628 and fns handling, rather than in a separate case SSA_NAME.
4630 2017-01-02 Jeff Law <law@redhat.com>
4632 * config/darwin-driver.c (darwin_driver_init): Const-correctness
4633 fixes for first_period and second_period variables.
4635 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
4638 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
4639 (*insvqi_1): New insn pattern.
4640 (*insvqi_1_mem_rex64): Ditto.
4642 (*insvqi_3): Rename from *insvqi.
4644 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
4646 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
4648 * doc/cfg.texi (Edges): Remove reference to Java.
4649 (Maintaining the CFG): Ditto.
4651 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4654 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
4655 transparent aliases.
4657 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4660 * predict.def (PRED_CALL): Update hitrate.
4661 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
4662 * predict.c (tree_estimate_probability_bb): Split CALL predictor
4663 into direct/indirect/polymorphic variants.
4665 2017-01-01 Jakub Jelinek <jakub@redhat.com>
4667 Update copyright years.
4669 * gcc.c (process_command): Update copyright notice dates.
4670 * gcov-dump.c (print_version): Ditto.
4671 * gcov.c (print_version): Ditto.
4672 * gcov-tool.c (print_version): Ditto.
4673 * gengtype.c (create_file): Ditto.
4674 * doc/cpp.texi: Bump @copying's copyright year.
4675 * doc/cppinternals.texi: Ditto.
4676 * doc/gcc.texi: Ditto.
4677 * doc/gccint.texi: Ditto.
4678 * doc/gcov.texi: Ditto.
4679 * doc/install.texi: Ditto.
4680 * doc/invoke.texi: Ditto.
4682 Copyright (C) 2017 Free Software Foundation, Inc.
4684 Copying and distribution of this file, with or without modification,
4685 are permitted in any medium without royalty provided the copyright
4686 notice and this notice are preserved.