1 2018-06-14 Richard Biener <rguenther@suse.de>
4 * tree-vect-generic.c (build_word_mode_vector_type): Remove
5 duplicate and harmful type_hash_canon.
6 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8 2018-06-14 Richard Biener <rguenther@suse.de>
11 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
14 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
16 * config/rtems.h (STDINT_LONG32): Define.
18 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
19 Prachi Godbole <prachi.godbole@imgtec.com>
21 * config/mips/mips-cpus.def: Define P6600.
22 * config/mips/mips-tables.opt: Regenerate.
23 * config/mips/mips.c (mips_ucbranch_type): New enum.
24 (mips_rtx_cost_data): Add support for P6600.
25 (mips_issue_rate): Likewise.
26 (mips_multipass_dfa_lookahead): Likewise.
27 (mips_avoid_hazard): Likewise.
28 (mips_reorg_process_insns): Likewise.
29 (mips_classify_branch_p6600): New function.
30 * config/mips/mips.h (TUNE_P6600): New define.
31 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
32 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
33 * config/mips/mips.md: Include p6600.md.
34 (processor): Add p6600.
35 * config/mips/p6600.md: New file.
36 * doc/invoke.texi: Add p6600 to supported architectures.
38 2018-06-13 Martin Sebor <msebor@redhat.com>
40 PR tree-optimization/86114
41 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
43 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
45 2018-06-13 Richard Biener <rguenther@suse.de>
47 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
48 Properly set vector type of the intermediate stmt.
49 * tree-vect-stmts.c (vectorizable_operation): The destination
50 var always has vectype_out type.
52 2018-06-13 Jeff Law <law@redhat.com>
54 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
55 integer 0 for argument to print_rtl_with_bb.
56 (rl78_reorg): Likewise.
58 2018-06-13 David Malcolm <dmalcolm@redhat.com>
60 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
61 from rtx to rtx_insn *.
62 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
64 (add_sched_insns_for_speculation): Likewise for local "target",
65 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
66 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
67 from rtx_insn ** to rtx_code_label **.
68 (reorg_emit_nops): Likewise.
69 (c6x_reorg): Likewise for local "call_labels".
70 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
72 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
73 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
74 the loops over LABEL_REFS.
75 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
77 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
78 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
79 (split_branches): Strengthen local "olabel" from rtx to
80 rtx_insn *, adding a safe_as_a cast.
81 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
83 (add_insn_after): Likewise for first two params.
84 (add_insn_before): Likewise.
85 (remove_insn): Likewise for param.
86 (emit_pattern_before_noloc): Likewise for second and third params.
87 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
88 (emit_call_insn_before_noloc): Likewise.
89 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
91 (emit_barrier_before): Likewise.
92 (emit_label_before): Strengthen "label" param from "rtx" to
93 "rtx_code_label *". Strengthen "before" param from "rtx" to
95 (emit_insn_after_1): Strengthen "after" param from "rtx" to
97 (emit_pattern_after_noloc): Likewise.
98 (emit_insn_after_noloc): Likewise.
99 (emit_jump_insn_after_noloc): Likewise.
100 (emit_call_insn_after_noloc): Likewise.
101 (emit_debug_insn_after_noloc): Likewise.
102 (emit_barrier_after): Likewise.
103 (emit_label_after): Likewise for both params.
104 (emit_pattern_after_setloc): Likewise for "after" param. Convert
105 "loc" param from "int" to "location_t".
106 (emit_insn_after_setloc): Likewise.
107 (emit_jump_insn_after_setloc): Likewise.
108 (emit_call_insn_after_setloc): Likewise.
109 (emit_debug_insn_after_setloc): Likewise.
110 (emit_pattern_before_setloc): Likewise for "before" param. Convert
111 "loc" param from "int" to "location_t".
112 (emit_pattern_before): Convert NULL_RTX to NULL.
113 (emit_insn_before_setloc): Convert "loc" param from "int" to
115 (emit_jump_insn_before_setloc): Likewise.
116 (emit_call_insn_before_setloc): Likewise.
117 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
118 rtx_insn *. Convert "loc" param from "int" to "location_t".
119 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
120 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
121 Convert 3rd param from "int" to "location_t".
122 (emit_barrier_before, emit_barrier_after, next_real_insn):
123 Strengthen param from rtx to rtx_insn *.
124 (emit_label_before): Strengthen 1st param from "rtx" to
125 "rtx_code_label *". Strengthen 2nd param from "rtx" to
127 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
128 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
129 Strengthen 2nd param from "rtx" to "rtx_insn *".
130 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
131 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
132 Likewise. Convert 3rd param from "int" to "location_t".
133 (emit_label_after): Strengthen 1st param from "rtx" to
135 (next_real_insn, remove_insn): Strengthen param from "rtx" to
137 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
138 from "rtx" to "rtx_insn *".
140 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
142 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
143 bodies streamed in with -Q.
144 * dumpfile.c (dump_files): Add lto-stream-out dump file.
145 * dumpfile.h (tree_dump_index): Add lto_stream_out.
146 * gimple-streamer-out.c: Include gimple-pretty-print.h
147 (output_bb): Dump stmts streamed.
148 * lto-section-out.c: Include print-tree.h
149 (lto_begin_section): Dump sections created.
150 (lto_output_decl_index): Dump decl encoded.
151 * lto-streamer-out.c: Include print-tree.h
152 (create_output_block): Dump output block created.
153 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
154 (output_function): Dump function output.
155 (output_constructor): Dump constructor streamed.
156 (write_global_stream): Output indexes encoded.
157 (produce_asm_for_decls): Dump streams encoded.
158 * lto-streamer.c (streamer_dump_file): New global var.
159 * lto-streamer.h (streamer_dump_file): Declare.
160 * passes.c (ipa_write_summaries): Initialize streamer dump.
161 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
164 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
167 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
168 offsets for register save directives. Emit a second batch of save
169 directives, if need be, when the function accesses prior frames.
171 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
173 * config/arc/fpu.md (fmasf4): Force operand to register.
176 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
178 * config/arc/arc-protos.h (arc_pad_return): Remove.
179 * config/arc/arc.c (machine_function): Remove force_short_suffix
181 (arc_print_operand): Adjust printing of '&'.
182 (arc_verify_short): Remove conditional printing of short suffix.
183 (arc_final_prescan_insn): Remove reference to size_reason.
184 (pad_return): New function.
185 (arc_reorg): Call pad_return.
186 (arc_pad_return): Remove.
187 (arc_init_machine_status): Remove reference to force_short_suffix.
188 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
189 (attr length): When attribute iscompact is true force to 2
190 regardless; in the case of maybe check if we want to force the
191 instruction to have 4 bytes length.
192 (nopv): Change it to generate 4 byte long nop as well.
193 (blockage): New pattern.
194 (simple_return): Remove call to arc_pad_return.
195 (p_return_i): Likewise.
197 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
199 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
201 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
203 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
206 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
208 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
209 for ARC700 and ARCv2.
211 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
214 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
215 operands[2] instead of operands[1].
218 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
220 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
221 case, check whether the outer register overlaps an unallocatable
222 register, not just whether it fits the required class.
224 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
226 * poly-int.h (can_div_trunc_p): Add new overload in which all values
228 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
229 (memrefs_conflict_p): Likewise.
230 (init_alias_analysis): Likewise.
231 * cfgexpand.c (expand_debug_expr): Likewise.
232 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
233 * cse.c (fold_rtx): Likewise.
234 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
235 * expr.c (emit_block_move_hints): Likewise.
236 (clear_storage_hints, push_block, emit_push_insn): Likewise.
237 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
238 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
239 (emit_group_store): Likewise.
240 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
241 to read the PRE/POST_MODIFY increment.
242 * calls.c (store_one_arg): Use strip_offset.
243 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
245 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
247 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
248 (simplify_binary_operation_1): Extend CONST_INT handling to
250 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
251 than a HOST_WIDE_INT.
252 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
254 (adjust_mems, add_stores): Update accodingly.
255 (vt_canonicalize_addr): Track polynomial offsets.
256 (emit_note_insn_var_location): Likewise.
257 (vt_add_function_parameter): Likewise.
258 (vt_initialize): Likewise.
260 2018-06-12 Jeff Law <law@redhat.com>
262 * config.gcc (alpha*-*-freebsd*): Remove.
263 * config/alpha/freebsd.h: Remove.
265 2018-06-12 David Malcolm <dmalcolm@redhat.com>
268 * spellcheck-tree.c (levenshtein_distance): Rename to...
269 (get_edit_distance): ...this, and update for underlying renaming.
270 * spellcheck-tree.h (levenshtein_distance): Rename to...
271 (get_edit_distance): ...this.
272 * spellcheck.c (levenshtein_distance): Rename to...
273 (get_edit_distance): ...this. Convert from Levenshtein distance
274 to Damerau-Levenshtein distance by supporting transpositions of
275 adjacent characters. Rename "v1" to "v_next" and "v0" to
277 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
278 (selftest::test_edit_distance_unit_test_oneway): ...this, and
279 update for underlying renaming.
280 (selftest::levenshtein_distance_unit_test): Rename to...
281 (selftest::test_get_edit_distance_unit): ...this, and update for
283 (selftest::test_find_closest_string): Add example from PR 69968
284 where transposition helps
285 (selftest::test_metric_conditions): Update for renaming.
286 (selftest::test_metric_conditions): Likewise.
287 (selftest::spellcheck_c_tests): Likewise.
288 * spellcheck.h (levenshtein_distance): Rename both overloads to...
289 (get_edit_distance): ...this.
290 (best_match::consider): Update for renaming.
292 2018-06-12 Martin Sebor <msebor@redhat.com>
294 PR tree-optimization/85259
295 * builtins.c (compute_objsize): Handle constant offsets.
296 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
297 true iff a warning has been issued.
298 * gimple.h (gimple_nonartificial_location): New function.
299 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
300 gimple_nonartificial_location and handle -Wno-system-headers.
301 (handle_builtin_stxncpy): Same.
303 2018-06-12 Martin Sebor <msebor@redhat.com>
306 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
308 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
310 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
311 BUILTIN_VEC_XST entries for pointer to double and long long.
313 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
316 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
318 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
321 2018-06-12 Martin Liska <mliska@suse.cz>
323 * doc/options.texi: Document IntegerRange.
325 2018-06-12 Martin Liska <mliska@suse.cz>
327 * config/i386/i386.opt: Make MPX-related options as Deprecated.
328 * opt-functions.awk: Handle Deprecated flag.
329 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
331 (read_cmdline_option): Report warning for a deprecated option.
332 * opts.h (struct cl_option): Add new field cl_deprecated.
333 (CL_ERR_DEPRECATED): New.
335 2018-06-12 Martin Liska <mliska@suse.cz>
337 * doc/options.texi: Document Deprecated option flag.
339 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
341 * config/arc/arc-arch.h (arc_extras): New enum.
342 (arc_cpu_t):Add field extra.
343 (arc_cpu_types): Consider the extras.
344 * config/arc/arc-cpus.def: Add extras info.
345 * config/arc/arc-opts.h (processor_type): Consider extra field.
346 * config/arc/arc.c (arc_override_options): Handle extra field.
348 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
350 * config/arc/arc-arch.h: Update ARC_OPTX macro.
351 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
353 * config/arc/arc.c (arc_init): Update pic warning.
354 (irq_range): Update irq range parsing warnings.
355 (arc_override_options): Update various warning messages.
356 (arc_handle_aux_attribute): Likewise.
358 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
360 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
362 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
364 * doc/sourcebuild.texi: Document usage of line number 0 in verify
365 compiler messages directives.
367 2018-06-12 Matthew Fortune <mfortune@gmail.com>
369 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
370 * config/mips/mips-tables.opt: Regenerate.
371 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
373 * doc/invoke.texi: Document -march=i6500.
375 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
377 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
378 (i6400_gpmul): Add cpu_unit.
379 (i6400_gpdiv): Likewise.
380 (i6400_msa_add_d): Update reservations.
381 (i6400_msa_int_add) Likewise.
382 (i6400_msa_short_logic3) Likewise.
383 (i6400_msa_short_logic2) Likewise.
384 (i6400_msa_short_logic) Likewise.
385 (i6400_msa_move) Likewise.
386 (i6400_msa_cmp) Likewise.
387 (i6400_msa_short_float2) Likewise.
388 (i6400_msa_div_d) Likewise.
389 (i6400_msa_long_logic1) Likewise.
390 (i6400_msa_long_logic2) Likewise.
391 (i6400_msa_mult) Likewise.
392 (i6400_msa_long_float2) Likewise.
393 (i6400_msa_long_float4) Likewise.
394 (i6400_msa_long_float5) Likewise.
395 (i6400_msa_long_float8) Likewise.
396 (i6400_fpu_fadd): Include frint type.
397 (i6400_fpu_store): New define_insn_reservation.
398 (i6400_fpu_load): Likewise.
399 (i6400_fpu_move): Likewise.
400 (i6400_fpu_fcmp): Likewise.
401 (i6400_fpu_fmadd): Likewise.
402 (i6400_int_mult): Include imul3nc type and update reservation.
403 (i6400_int_div): Include idiv3 type and update reservation.
404 (i6400_int_load): Update to check type not move_type.
405 (i6400_int_store): Likewise.
406 (i6400_int_prefetch): Set zero latency.
408 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
410 * gcc.c: Document new %@{...} sequence.
411 (LINK_COMMAND_SPEC): Use it for the -L switches.
412 (cpp_unique_options): Use it for the -I switches.
413 (at_file_argbuf): New global variable.
414 (in_at_file): Likewise.
415 (alloc_args): Create at_file_argbuf.
416 (clear_args): Truncate at_file_argbuf.
417 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
418 (open_at_file): New function.
419 (close_at_file): Likewise.
420 (create_at_file): Delete.
421 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
424 (validate_switches_from_spec): Deal with %@{...} sequence.
425 (validate_switches): Likewise.
426 (driver::finalize): Call clear_args.
428 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
430 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
432 2018-06-11 Martin Sebor <msebor@redhat.com>
434 * doc/invoke.texi (-Wall): List -Wc++17-compat.
435 (Wno-class-memaccess): Add @opindex.
436 (Wno-templates, Wno-multiple-inheritance): Same.
437 (Wno-virtual-inheritance, Wno-namespaces): Same.
438 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
439 (Wno-format-overflow, Wno-format-truncation): Same.
440 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
441 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
442 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
443 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
444 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
445 (Wno-misspelled-isr): Same.
447 2018-06-11 Martin Sebor <msebor@redhat.com>
449 * PR tree-optimization/86083
450 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
452 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
454 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
456 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
459 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
460 on the correct operand.
461 (*movdi_internal64): Ditto.
463 2018-06-11 Martin Liska <mliska@suse.cz>
465 PR tree-optimization/86089
466 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
468 2018-06-11 Julia Koval <julia.koval@intel.com>
470 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
471 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
472 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
474 2018-06-11 Olivier Hainque <hainque@adacore.com>
476 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
477 for Ada with strict dwarf2.
479 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
482 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
485 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
487 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
489 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
491 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
494 2018-06-08 Martin Liska <mliska@suse.cz>
496 * tree-cfg.h (debug_function): Fix argument type to match
499 2018-06-08 Martin Liska <mliska@suse.cz>
501 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
502 Remove usage of MPX-related (and removed) fields.
503 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
505 2018-06-08 David Malcolm <dmalcolm@redhat.com>
507 * cfg.c (debug): Use TDF_NONE rather than 0.
508 * cfghooks.c (debug): Likewise.
509 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
510 (struct dump_option_value_info): Convert to...
511 (struct kv_pair): ...this template type.
512 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
514 (optinfo_verbosity_options): Likewise.
515 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
517 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
518 than int for "optgroup_flags" param.
519 (dump_generic_expr_loc): Use dump_flags_t rather than int for
521 (dump_dec): Likewise.
522 (dump_finish): Use TDF_NONE rather than 0.
523 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
524 rather than int for "optgroup_flags" param. Use TDF_NONE rather
525 than 0. Update for change to option_ptr.
526 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
527 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
528 0. Update for changes to optinfo_verbosity_options and
530 (opt_info_switch_p): Convert optgroup_flags from int to
532 (dump_basic_block): Use dump_flags_t rather than int
533 for "dump_kind" param.
534 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
535 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
536 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
537 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
538 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
539 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
540 TDF_NONE): Convert from macros to...
541 (enum dump_flag): ...this new enum.
542 (dump_flags_t): Update to use enum.
543 (operator|, operator&, operator~, operator|=, operator&=):
544 Implement for dump_flags_t.
545 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
546 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
547 Convert from macros to...
548 (enum optgroup_flag): ...this new enum.
549 (optgroup_flags_t): New typedef.
550 (operator|, operator|=): Implement for optgroup_flags_t.
551 (struct dump_file_info): Convert field "alt_flags" to
552 dump_flags_t. Convert field "optgroup_flags" to
554 (dump_basic_block): Use dump_flags_t rather than int for param.
555 (dump_generic_expr_loc): Likewise.
556 (dump_dec): Likewise.
557 (dump_register): Convert param "optgroup_flags" to
559 (opt_info_enable_passes): Likewise.
560 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
562 * gimple-pretty-print.c (debug): Likewise.
563 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
564 (merged_store_group::apply_stores): Likewise.
565 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
566 * gimple.c (verify_gimple_pp): Likewise.
567 * graphite-poly.c (print_pbb_body): Likewise.
568 * passes.c (pass_manager::register_one_dump_file): Convert
569 local "optgroup_flags" to optgroup_flags_t.
570 * print-tree.c (print_node): Use TDF_NONE rather than 0.
572 (debug_body): Likewise.
573 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
575 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
577 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
578 (convert_mult_to_fma): Likewise.
579 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
580 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
581 * tree-vect-data-refs.c (dump_lower_bound): Convert param
582 "dump_kind" to dump_flags_t.
584 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
586 * config/rs6000/rs6000.c (min, max): Delete.
588 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
590 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
591 -mabi=spe and -mabi=no-spe.
593 2018-06-08 Martin Liska <mliska@suse.cz>
595 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
596 where we expect an existing summary.
598 2018-06-08 Martin Liska <mliska@suse.cz>
600 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
601 * ipa-inline.h (estimate_edge_growth): Likewise.
603 2018-06-08 Martin Liska <mliska@suse.cz>
605 * cgraph.c (function_version_hasher::hash): Use
606 cgraph_node::get_uid ().
607 (function_version_hasher::equal):
608 * cgraph.h (cgraph_node::get_uid): New method.
609 * ipa-inline.c (update_caller_keys): Use
610 cgraph_node::get_uid ().
611 (update_callee_keys): Likewise.
612 * ipa-utils.c (searchc): Likewise.
613 (ipa_reduced_postorder): Likewise.
614 * lto-cgraph.c (input_node): Likewise.
615 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
616 * symbol-summary.h (symtab_insertion): Likewise.
617 (symtab_removal): Likewise.
618 (symtab_duplication): Likewise.
619 * tree-pretty-print.c (dump_function_header): Likewise.
620 * tree-sra.c (convert_callers_for_node): Likewise.
622 2018-06-08 Martin Liska <mliska@suse.cz>
624 * cgraph.c (symbol_table::create_edge): Always assign a new
626 (symbol_table::free_edge): Do not recycle numbers.
627 * cgraph.h (cgraph_edge::get): New method.
628 * symbol-summary.h (symtab_removal): Use it.
629 (symtab_duplication): Likewise.
630 (call_summary::hashable_uid): Remove.
632 2018-06-08 Martin Liska <mliska@suse.cz>
634 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
635 (initialize_growth_caches): Remove.
636 (free_growth_caches): Likewise.
637 (do_estimate_edge_time): Use edge_growth_cache.
638 (do_estimate_edge_size): Likewise.
639 (do_estimate_edge_hints): Likewise.
640 * ipa-inline.c (reset_edge_caches): Likewise.
641 (recursive_inlining): Likewise.
642 (inline_small_functions): Likewise.
643 * ipa-inline.h (initialize_growth_caches): Remove.
644 (estimate_edge_size): Likewise.
645 (estimate_edge_time): Likewise.
646 (estimate_edge_hints): Likewise.
647 (reset_edge_growth_cache): Likewise.
648 * symbol-summary.h (call_summary::remove): New method.
650 2018-06-08 Martin Liska <mliska@suse.cz>
652 * ipa-cp.c (class edge_clone_summary): New summary.
653 (grow_edge_clone_vectors): Remove.
654 (ipcp_edge_duplication_hook): Remove.
655 (class edge_clone_summary_t): New call_summary class.
656 (ipcp_edge_removal_hook): Remove.
657 (edge_clone_summary_t::duplicate): New function.
658 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
659 (create_specialized_node): Likewise.
660 (ipcp_driver): Initialize edge_clone_summaries and do not
663 2018-06-08 Martin Liska <mliska@suse.cz>
665 * symbol-summary.h (get): New function.
666 (call_summary::m_initialize_when_cloning): New class member.
668 2018-06-08 Martin Liska <mliska@suse.cz>
670 * cgraph.c (cgraph_node::remove): Do not recycle uid.
671 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
672 (symbol_table::allocate_cgraph_symbol): Do not set uid.
673 * passes.c (uid_hash_t): Record removed_nodes by their uids.
674 (remove_cgraph_node_from_order): Use the removed_nodes set.
675 (do_per_function_toporder): Likwise.
676 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
677 instead of summary_uid.
678 (symtab_removal): Likewise.
679 (symtab_duplication): Likewise.
681 2018-06-08 Martin Liska <mliska@suse.cz>
683 * ipa-cp.c (ipcp_store_bits_results): Use
684 ipcp_transformation_sum.
685 (ipcp_store_vr_results): Likewise.
686 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
688 (ipcp_transformation_initialize): ... this.
689 (ipa_set_node_agg_value_chain):
690 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
691 (write_ipcp_transformation_info): Likewise.
692 (read_ipcp_transformation_info): Likewise.
693 (ipcp_update_bits): Likewise.
694 (ipcp_update_vr): Likewise.
695 (ipcp_transform_function): Likewise.
696 * ipa-prop.h: Rename ipcp_transformation_summary to
698 (class ipcp_transformation_t): New function summary.
699 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
700 (ipa_get_agg_replacements_for_node): Likewise.
702 2018-06-08 Martin Liska <mliska@suse.cz>
704 * ipa-pure-const.c (struct funct_state_d): Do it class instead
706 (class funct_state_summary_t): New function_summary class.
707 (has_function_state): Remove.
708 (get_function_state): Likewise.
709 (set_function_state): Likewise.
710 (add_new_function): Likewise.
711 (funct_state_summary_t::insert): New function.
712 (duplicate_node_data): Remove.
713 (remove_node_data): Remove.
714 (funct_state_summary_t::duplicate): New function.
715 (register_hooks): Create new funct_state_summaries.
716 (pure_const_generate_summary): Use it.
717 (pure_const_write_summary): Likewise.
718 (pure_const_read_summary): Likewise.
719 (propagate_pure_const): Likewise.
720 (propagate_nothrow): Likewise.
721 (dump_malloc_lattice): Likewise.
722 (propagate_malloc): Likewise.
723 (execute): Do not register hooks, just remove summary
725 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
728 2018-06-08 Martin Liska <mliska@suse.cz>
730 * ipa-reference.c (remove_node_data): Remove.
731 (duplicate_node_data): Likewise.
732 (class ipa_ref_var_info_summary_t): New class.
733 (class ipa_ref_opt_summary_t): Likewise.
734 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
735 (get_reference_optimization_summary): Use
736 ipa_ref_opt_sum_summaries.
737 (set_reference_vars_info): Remove.
738 (set_reference_optimization_summary): Likewise.
739 (ipa_init): Create summaries.
740 (init_function_info): Use function summary.
741 (ipa_ref_opt_summary_t::duplicate): New function.
742 (ipa_ref_opt_summary_t::remove): New function.
743 (get_read_write_all_from_node): Fix GNU coding style.
744 (propagate): Use function summary.
745 (write_node_summary_p): Fix GNU coding style.
746 (stream_out_bitmap): Likewise.
747 (ipa_reference_read_optimization_summary): Use function summary.
748 (ipa_reference_c_finalize): Do not release hooks.
750 2018-06-08 Martin Liska <mliska@suse.cz>
752 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
753 (analyze_function_body): Extract multiple calls of get_create.
754 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
755 * ipa-inline.c (recursive_inlining): Use ::get method.
756 * ipa-inline.h (estimate_edge_growth): Likewise.
758 2018-06-08 Martin Liska <mliska@suse.cz>
760 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
762 (hsa_function_summary::hsa_function_summary): Use the new enum
764 (hsa_gpu_implementation_p): Use hsa_summaries::get.
765 * hsa-gen.c (hsa_get_host_function): Likewise.
766 (get_brig_function_name): Likewise.
767 * ipa-hsa.c (process_hsa_functions): Likewise.
768 (ipa_hsa_write_summary): Likewise.
769 * symbol-summary.h (symtab_duplication): Use ::get function/
772 2018-06-08 Martin Liska <mliska@suse.cz>
774 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
776 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
777 (hsa_register_kernel): Likewise.
778 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
779 * hsa-gen.c (hsa_get_host_function): Likewise.
780 (get_brig_function_name): Likewise.
781 (generate_hsa): Likewise.
782 (pass_gen_hsail::execute): Likewise.
783 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
784 (devirtualization_time_bonus): Likewise.
785 (ipcp_propagate_stage): Likewise.
786 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
787 (edge_set_predicate): Likewise.
788 (evaluate_conditions_for_known_args): Likewise.
789 (evaluate_properties_for_edge): Likewise.
790 (ipa_fn_summary::reset): Likewise.
791 (ipa_fn_summary_t::duplicate): Likewise.
792 (dump_ipa_call_summary): Likewise.
793 (ipa_dump_fn_summary): Likewise.
794 (analyze_function_body): Likewise.
795 (compute_fn_summary): Likewise.
796 (estimate_edge_devirt_benefit): Likewise.
797 (estimate_edge_size_and_time): Likewise.
798 (estimate_calls_size_and_time): Likewise.
799 (estimate_node_size_and_time): Likewise.
800 (inline_update_callee_summaries): Likewise.
801 (remap_edge_change_prob): Likewise.
802 (remap_edge_summaries): Likewise.
803 (ipa_merge_fn_summary_after_inlining): Likewise.
804 (ipa_update_overall_fn_summary): Likewise.
805 (read_ipa_call_summary): Likewise.
806 (inline_read_section): Likewise.
807 (write_ipa_call_summary): Likewise.
808 (ipa_fn_summary_write): Likewise.
809 (ipa_free_fn_summary): Likewise.
810 * ipa-hsa.c (process_hsa_functions): Likewise.
811 (ipa_hsa_write_summary): Likewise.
812 (ipa_hsa_read_section): Likewise.
813 * ipa-icf.c (sem_function::merge): Likewise.
814 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
815 (do_estimate_edge_time): Likewise.
816 (estimate_size_after_inlining): Likewise.
817 (estimate_growth): Likewise.
818 (growth_likely_positive): Likewise.
819 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
820 (inline_call): Likewise.
821 * ipa-inline.c (caller_growth_limits): Likewise.
822 (can_inline_edge_p): Likewise.
823 (can_inline_edge_by_limits_p): Likewise.
824 (compute_uninlined_call_time): Likewise.
825 (compute_inlined_call_time): Likewise.
826 (want_inline_small_function_p): Likewise.
827 (edge_badness): Likewise.
828 (update_caller_keys): Likewise.
829 (update_callee_keys): Likewise.
830 (recursive_inlining): Likewise.
831 (inline_small_functions): Likewise.
832 (inline_to_all_callers_1): Likewise.
833 (dump_overall_stats): Likewise.
834 (early_inline_small_functions): Likewise.
835 (early_inliner): Likewise.
836 * ipa-inline.h (estimate_edge_growth): Likewise.
837 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
838 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
839 * ipa-prop.h (IPA_NODE_REF): Likewise.
840 (IPA_EDGE_REF): Likewise.
841 * ipa-pure-const.c (malloc_candidate_p): Likewise.
842 (propagate_malloc): Likewise.
843 * ipa-split.c (execute_split_functions): Likewise.
844 * symbol-summary.h: Rename get to get_create.
846 (get_create): Likewise.
847 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
849 2018-06-08 Martin Liska <mliska@suse.cz>
851 * symbol-summary.h (release): Move definition out of class
853 (symtab_removal): Likewise.
854 (symtab_duplication): Likewise.
856 2018-06-08 Martin Liska <mliska@suse.cz>
858 * symbol-summary.h (function_summary): Move constructor
859 implementation out of class declaration.
861 (symtab_insertion): Likewise.
862 (symtab_removal): Likewise.
863 (symtab_duplication): Likewise.
866 2018-06-08 Martin Liska <mliska@suse.cz>
868 * Makefile.in: Remove support for MPX (macros, related functions,
869 fields in cgraph_node, ...).
870 * builtin-types.def (BT_BND): Likewise.
871 (BT_FN_BND_CONST_PTR): Likewise.
872 (BT_FN_CONST_PTR_BND): Likewise.
873 (BT_FN_VOID_PTR_BND): Likewise.
874 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
875 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
876 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
877 (expand_builtin_mempcpy_with_bounds): Likewise.
878 (expand_builtin_memset_with_bounds): Likewise.
879 (expand_builtin_memset_args): Likewise.
880 (std_expand_builtin_va_start): Likewise.
881 (expand_builtin): Likewise.
882 (expand_builtin_with_bounds): Likewise.
883 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
884 (DEF_LIB_BUILTIN_CHKP): Likewise.
885 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
886 (DEF_CHKP_BUILTIN): Likewise.
887 (BUILT_IN_MEMCPY): Likewise.
888 (BUILT_IN_MEMMOVE): Likewise.
889 (BUILT_IN_MEMPCPY): Likewise.
890 (BUILT_IN_MEMSET): Likewise.
891 (BUILT_IN_STPCPY): Likewise.
892 (BUILT_IN_STRCAT): Likewise.
893 (BUILT_IN_STRCHR): Likewise.
894 (BUILT_IN_STRCPY): Likewise.
895 (BUILT_IN_STRLEN): Likewise.
896 (BUILT_IN_MEMCPY_CHK): Likewise.
897 (BUILT_IN_MEMMOVE_CHK): Likewise.
898 (BUILT_IN_MEMPCPY_CHK): Likewise.
899 (BUILT_IN_MEMSET_CHK): Likewise.
900 (BUILT_IN_STPCPY_CHK): Likewise.
901 (BUILT_IN_STRCAT_CHK): Likewise.
902 (BUILT_IN_STRCPY_CHK): Likewise.
903 * calls.c (store_bounds): Likewise.
904 (emit_call_1): Likewise.
905 (special_function_p): Likewise.
906 (maybe_warn_nonstring_arg): Likewise.
907 (initialize_argument_information): Likewise.
908 (finalize_must_preallocate): Likewise.
909 (compute_argument_addresses): Likewise.
910 (expand_call): Likewise.
911 * cfgexpand.c (expand_call_stmt): Likewise.
912 (expand_return): Likewise.
913 (expand_gimple_stmt_1): Likewise.
914 (pass_expand::execute): Likewise.
915 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
916 (cgraph_node::remove): Likewise.
917 (cgraph_node::dump): Likewise.
918 (cgraph_node::verify_node): Likewise.
919 * cgraph.h (chkp_function_instrumented_p): Likewise.
920 (symtab_node::get_alias_target): Likewise.
921 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
922 (cgraph_local_p): Likewise.
923 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
924 (cgraph_edge::rebuild_references): Likewise.
925 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
926 (walk_polymorphic_call_targets): Likewise.
927 (cgraph_node::expand_thunk): Likewise.
928 (symbol_table::output_weakrefs): Likewise.
929 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
930 (ix86_handle_option): Likewise.
931 * config/i386/constraints.md: Likewise.
932 * config/i386/i386-builtin-types.def (BND): Likewise.
936 * config/i386/i386-builtin.def (BDESC_END): Likewise.
937 (BDESC_FIRST): Likewise.
939 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
940 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
941 * config/i386/i386.c (enum reg_class): Likewise.
942 (ix86_target_string): Likewise.
943 (ix86_option_override_internal): Likewise.
944 (ix86_conditional_register_usage): Likewise.
945 (ix86_valid_target_attribute_inner_p): Likewise.
946 (ix86_set_indirect_branch_type): Likewise.
947 (ix86_set_current_function): Likewise.
948 (ix86_function_arg_regno_p): Likewise.
949 (init_cumulative_args): Likewise.
950 (ix86_function_arg_advance): Likewise.
951 (ix86_function_arg): Likewise.
952 (ix86_pass_by_reference): Likewise.
953 (ix86_function_value_regno_p): Likewise.
954 (ix86_function_value_1): Likewise.
955 (ix86_function_value_bounds): Likewise.
956 (ix86_return_in_memory): Likewise.
957 (ix86_setup_incoming_vararg_bounds): Likewise.
958 (ix86_va_start): Likewise.
959 (indirect_thunk_need_prefix): Likewise.
960 (print_reg): Likewise.
961 (ix86_print_operand): Likewise.
962 (ix86_expand_call): Likewise.
963 (ix86_output_function_return): Likewise.
964 (reg_encoded_number): Likewise.
965 (BDESC_VERIFYS): Likewise.
966 (ix86_init_mpx_builtins): Likewise.
967 (ix86_init_builtins): Likewise.
968 (ix86_emit_cmove): Likewise.
969 (ix86_emit_move_max): Likewise.
970 (ix86_expand_builtin): Likewise.
971 (ix86_builtin_mpx_function): Likewise.
972 (ix86_get_arg_address_for_bt): Likewise.
973 (ix86_load_bounds): Likewise.
974 (ix86_store_bounds): Likewise.
975 (ix86_load_returned_bounds): Likewise.
976 (ix86_store_returned_bounds): Likewise.
977 (ix86_class_likely_spilled_p): Likewise.
978 (ix86_hard_regno_mode_ok): Likewise.
979 (x86_order_regs_for_local_alloc): Likewise.
980 (ix86_mitigate_rop): Likewise.
981 (ix86_bnd_prefixed_insn_p): Likewise.
982 (ix86_mpx_bound_mode): Likewise.
983 (ix86_make_bounds_constant): Likewise.
984 (ix86_initialize_bounds): Likewise.
985 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
986 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
987 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
988 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
989 (TARGET_CHKP_BOUND_MODE): Likewise.
990 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
991 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
992 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
993 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
994 * config/i386/i386.h (TARGET_MPX): Likewise.
995 (TARGET_MPX_P): Likewise.
996 (VALID_BND_REG_MODE): Likewise.
997 (FIRST_BND_REG): Likewise.
998 (LAST_BND_REG): Likewise.
999 (enum reg_class): Likewise.
1000 (BND_REG_P): Likewise.
1001 (BND_REGNO_P): Likewise.
1002 (BNDmode): Likewise.
1003 (ADJUST_INSN_LENGTH): Likewise.
1004 * config/i386/i386.md: Likewise.
1005 * config/i386/i386.opt: Likewise.
1006 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
1007 (defined): Likewise.
1008 (LINK_MPX): Likewise.
1009 (MPX_SPEC): Likewise.
1010 (LIBMPX_SPEC): Likewise.
1011 (LIBMPXWRAPPERS_SPEC): Likewise.
1012 (CHKP_SPEC): Likewise.
1013 * config/i386/predicates.md: Likewise.
1014 * dbxout.c (dbxout_type): Likewise.
1015 * doc/extend.texi: Likewise.
1016 * doc/invoke.texi: Likewise.
1017 * doc/md.texi: Likewise.
1018 * doc/tm.texi: Likewise.
1019 * doc/tm.texi.in: Likewise.
1020 * dwarf2out.c (is_base_type): Likewise.
1021 (gen_formal_types_die): Likewise.
1022 (gen_subprogram_die): Likewise.
1023 (gen_type_die_with_usage): Likewise.
1024 (gen_decl_die): Likewise.
1025 (dwarf2out_late_global_decl): Likewise.
1026 * expr.c (expand_assignment): Likewise.
1027 (emit_storent_insn): Likewise.
1028 (store_expr_with_bounds): Likewise.
1029 (store_expr): Likewise.
1030 (expand_expr_real_1): Likewise.
1031 * expr.h (store_expr_with_bounds): Likewise.
1032 * function.c (use_register_for_decl): Likewise.
1033 (struct bounds_parm_data): Likewise.
1034 (assign_parms_augmented_arg_list): Likewise.
1035 (assign_parm_find_entry_rtl): Likewise.
1036 (assign_parm_is_stack_parm): Likewise.
1037 (assign_parm_load_bounds): Likewise.
1038 (assign_bounds): Likewise.
1039 (assign_parms): Likewise.
1040 (expand_function_start): Likewise.
1041 * gcc.c (CHKP_SPEC): Likewise.
1042 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1043 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
1044 (wrestrict_dom_walker::check_call): Likewise.
1045 * gimple.c (gimple_build_call_from_tree): Likewise.
1046 * gimple.h (enum gf_mask): Likewise.
1047 (gimple_call_with_bounds_p): Likewise.
1048 (gimple_call_set_with_bounds): Likewise.
1049 * gimplify.c (gimplify_init_constructor): Likewise.
1050 * ipa-cp.c (initialize_node_lattices): Likewise.
1051 (propagate_constants_across_call): Likewise.
1052 (find_more_scalar_values_for_callers_subset): Likewise.
1053 * ipa-hsa.c (process_hsa_functions): Likewise.
1054 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
1055 * ipa-icf.c (sem_function::merge): Likewise.
1056 * ipa-inline.c (early_inliner): Likewise.
1057 * ipa-pure-const.c (warn_function_noreturn): Likewise.
1058 (warn_function_cold): Likewise.
1059 (propagate_pure_const): Likewise.
1060 * ipa-ref.h (enum GTY): Likewise.
1061 * ipa-split.c (find_retbnd): Likewise.
1062 (consider_split): Likewise.
1063 (split_function): Likewise.
1064 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1065 * ipa.c (walk_polymorphic_call_targets): Likewise.
1066 (symbol_table::remove_unreachable_nodes): Likewise.
1067 (process_references): Likewise.
1068 (cgraph_build_static_cdtor_1): Likewise.
1069 * lto-cgraph.c (lto_output_node): Likewise.
1070 (output_refs): Likewise.
1071 (compute_ltrans_boundary): Likewise.
1072 (input_overwrite_node): Likewise.
1073 (input_node): Likewise.
1074 (input_cgraph_1): Likewise.
1075 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
1076 * passes.c (pass_manager::execute_early_local_passes): Likewise.
1077 (class pass_chkp_instrumentation_passes): Likewise.
1078 (make_pass_chkp_instrumentation_passes): Likewise.
1079 * passes.def: Likewise.
1080 * rtl.h (struct GTY): Likewise.
1081 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
1082 * stor-layout.c (layout_type): Likewise.
1083 * symtab.c: Likewise.
1084 * target.def: Likewise.
1085 * targhooks.c (default_chkp_bound_type): Likewise.
1086 (default_chkp_bound_mode): Likewise.
1087 (default_builtin_chkp_function): Likewise.
1088 (default_chkp_function_value_bounds): Likewise.
1089 (default_chkp_make_bounds_constant): Likewise.
1090 (default_chkp_initialize_bounds): Likewise.
1091 * targhooks.h (default_chkp_bound_type): Likewise.
1092 (default_chkp_bound_mode): Likewise.
1093 (default_builtin_chkp_function): Likewise.
1094 (default_chkp_function_value_bounds): Likewise.
1095 (default_chkp_make_bounds_constant): Likewise.
1096 (default_chkp_initialize_bounds): Likewise.
1097 * toplev.c (compile_file): Likewise.
1098 (process_options): Likewise.
1099 * tree-core.h (DEF_BUILTIN): Likewise.
1100 (DEF_BUILTIN_CHKP): Likewise.
1101 * tree-inline.c (declare_return_variable): Likewise.
1102 (remap_gimple_stmt): Likewise.
1103 (copy_bb): Likewise.
1104 (initialize_inlined_parameters): Likewise.
1105 (expand_call_inline): Likewise.
1106 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
1107 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
1108 (make_pass_ipa_chkp_produce_thunks): Likewise.
1109 (make_pass_chkp): Likewise.
1110 (make_pass_chkp_opt): Likewise.
1111 (make_pass_chkp_instrumentation_passes): Likewise.
1112 * tree-pretty-print.c (dump_generic_node): Likewise.
1113 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
1114 * tree-ssa-dce.c (propagate_necessity): Likewise.
1115 (eliminate_unnecessary_stmts): Likewise.
1116 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
1117 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1118 * tree-ssa-sccvn.h: Likewise.
1119 * tree-ssa-strlen.c (get_string_length): Likewise.
1120 (valid_builtin_call): Likewise.
1121 (adjust_last_stmt): Likewise.
1122 (handle_builtin_strchr): Likewise.
1123 (handle_builtin_strcpy): Likewise.
1124 (handle_builtin_stxncpy): Likewise.
1125 (handle_builtin_memcpy): Likewise.
1126 (handle_builtin_strcat): Likewise.
1127 (strlen_check_and_optimize_stmt): Likewise.
1128 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
1129 * tree-streamer-in.c: Likewise.
1130 * tree-streamer.c (record_common_node): Likewise.
1131 * tree.c (tree_code_size): Likewise.
1132 (wide_int_to_tree_1): Likewise.
1133 (type_contains_placeholder_1): Likewise.
1134 (build_common_tree_nodes): Likewise.
1135 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
1136 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
1137 (POINTER_BOUNDS_P): Likewise.
1138 (BOUNDED_TYPE_P): Likewise.
1139 (BOUNDED_P): Likewise.
1140 (CALL_WITH_BOUNDS_P): Likewise.
1141 (pointer_bounds_type_node): Likewise.
1142 * value-prof.c (gimple_ic): Likewise.
1143 * var-tracking.c (vt_add_function_parameters): Likewise.
1144 * varasm.c (make_decl_rtl): Likewise.
1145 (assemble_start_function): Likewise.
1146 (output_constant): Likewise.
1147 (maybe_assemble_visibility): Likewise.
1148 * varpool.c (ctor_for_folding): Likewise.
1149 * chkp-builtins.def: Remove.
1150 * ipa-chkp.c: Remove.
1151 * ipa-chkp.h: Remove.
1152 * rtl-chkp.c: Remove.
1153 * rtl-chkp.h: Remove.
1154 * tree-chkp-opt.c: Remove.
1155 * tree-chkp.c: Remove.
1156 * tree-chkp.h: Remove.
1158 2018-06-07 Carl Love <cel@us.ibm.com>
1160 * gcc/config/rs6000/vsx.md (vextract_fp_from_shorth,
1161 vextract_fp_from_shortl): Add BE support.
1163 2018-06-07 Paul Koning <ni1d@arrl.net>
1165 * compare-elim.c (try_merge_compare): Don't merge compare if
1166 address contains a side effect.
1167 (try_eliminate_compare): Likewise.
1169 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
1171 * config.gcc: Support "tremont".
1172 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
1173 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1175 * config/i386/i386.c (m_TREMONT): Define.
1176 (processor_target_table): Add "tremont".
1177 (PTA_TREMONT): Define.
1178 (ix86_lea_outperforms): Add TARGET_TREMONT.
1179 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
1180 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
1181 and M_INTEL_GOLDMONT_PLUS.
1182 (fold_builtin_cpu): Add "tremont".
1183 (ix86_add_stmt_cost): Add TARGET_TREMONT.
1184 (ix86_option_override_internal): Add "tremont".
1185 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
1186 (processor_type): Add PROCESSOR_TREMONT.
1187 * config/i386/x86-tune.def: Add m_TREMONT.
1188 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
1190 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1192 * gcc/config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
1193 symbol defined for msp430i* devices to be lower case.
1195 gcc/testsuite/gcc.target/msp430/
1196 * msp430i-device-symbol.c: New test.
1197 * msp430f-device-symbol.c: New test.
1198 * msp430.h: New test header file.
1200 2018-06-07 Richard Biener <rguenther@suse.de>
1202 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
1203 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
1204 Properly wrap signed arithmetic if overflow wraps.
1206 2018-06-07 Jakub Jelinek <jakub@redhat.com>
1208 PR tree-optimization/69615
1209 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
1210 of a cast from a same precision integral SSA_NAME in a bb dominated
1211 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
1212 cast to utype if rhs2 has already a compatible type.
1214 2018-06-07 Richard Biener <rguenther@suse.de>
1216 PR tree-optimization/85935
1217 * graphite-scop-detection.c (find_params_in_bb): Analyze
1218 condition operands with respect to the correct loop. Assert
1219 the analysis doesn't fail.
1221 2018-06-04 Carl Love <cel@us.ibm.com>
1223 * gcc/config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
1224 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
1225 as it is slightly cheaper.
1226 (first_match_or_eos_index_<mode>):
1227 Calculate index using natural element order.
1228 (first_match_index_<mode>):
1229 Calculate index using natural element order.
1230 (first_match_or_eos_index_<mode>):
1231 Calculate index using natural order.
1232 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
1233 for BE and LE modes.
1234 * gcc/config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
1235 P9V_BUILTIN_VCLZLSBB_V16QI.
1236 * gcc/config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
1239 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
1241 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
1242 indentation and line wrap for many prototypes. Add missing
1243 @smallexample directives around block of prototypes for vec_xl and
1246 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
1248 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
1249 track if we pass or return IEEE 128-bit floating point.
1250 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
1251 C++ mangling that is compatible with GCC 8.1.
1252 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
1253 (init_cumulative_args): Note if we pass or return IEEE 128-bit
1254 floating point types.
1255 (rs6000_function_arg_advance_1): Likewise.
1256 (rs6000_mangle_type): Optionally generate mangled names that match
1257 what GCC 8.1 generated for IEEE 128-bit floating point types.
1258 (rs6000_globalize_decl_name): If we have an external function that
1259 passes or returns IEEE 128-bit types, generate a weak reference
1260 from the mangled name used in GCC 8.1 to the current mangled
1262 (rs6000_init_builtins): Make __ibm128 use the long double type if
1263 long double is IBM extended double. Make __float128 use the long
1264 double type if long double is IEEE 128-bit.
1267 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
1268 macro for __ibm128 built-in functions.
1269 (PACK_IF): Add __ibm128 pack/unpack functions.
1270 (UNPACK_IF): Likewise.
1271 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
1272 enable long double built-in functions if long double is IEEE
1273 128-bit floating point.
1274 (rs6000_invalid_builtin): Update long double built-in function
1276 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
1277 functions, adjust the built-in function to use the long double
1278 built-in function if __ibm128 and long double are the same type.
1279 * doc/extend.texi (PowerPC builtins): Update documention for
1280 __builtin_{,un}pack_longdouble. Add documentation for
1281 __builtin_{,un}pack_ibm128.
1283 2018-06-06 Jim Wilson <jimw@sifive.com>
1285 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
1286 (struct machine_function): New field interrupt_mode.
1287 (riscv_handle_type_attribute): New function. Add forward declaration.
1288 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
1289 (riscv_expand_epilogue): Check interrupt_mode field.
1290 (riscv_set_current_function): Check interrupt attribute args and
1291 set interrupt_mode field.
1292 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
1293 (riscv_sret, riscv_uret): New.
1294 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
1295 new arguments to interrupt attribute.
1297 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
1300 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
1301 Don't handle -mcpu=power8 if -mpower9-vector is also used.
1303 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
1305 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
1306 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
1307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1308 several redundant entries.
1310 2018-06-06 David Malcolm <dmalcolm@redhat.com>
1312 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
1313 type from "rtx" to "rtx_insn *".
1314 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
1315 for local "call_insn", removing cast.
1316 (ix86_expand_call): Likewise, introducing a "call_insn" local.
1318 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
1320 PR tree-optimization/86066
1321 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
1322 for BIT_INSERT_EXPR stores.
1324 2018-06-06 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/86062
1327 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
1328 component refs ontop
1329 of to be offsetted base.
1331 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1333 * gcc/config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
1334 to be static and remove check on interrupt attribute name.
1336 gcc/testsuite/gcc.target/msp430/
1337 * function-attributes-4.c: New test.
1338 * static-interrupts.c: New test.
1340 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
1342 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
1343 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
1345 2018-06-05 Steve Ellcey <sellcey@cavium.com>
1348 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
1350 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
1351 Remove second argument, change how error is called.
1352 (aarch64_layout_arg): Remove second argument from
1353 aarch64_err_no_fpadvsimd call.
1354 (aarch64_init_cumulative_args): Ditto.
1355 (aarch64_gimplify_va_arg_expr): Ditto.
1356 * config/aarch64/aarch64.md (mov<mode>): Ditto.
1358 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
1360 * config/i386/i386.md (simple_return_indirect_internal): New expander.
1361 (*simple_return_indirect_internal<mode>): Rename from
1362 simple_return_indirect_internal. Use W mode iterator.
1363 (rstorssp): New expander.
1364 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
1365 (clrssbsy): New expander.
1366 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
1368 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
1370 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
1372 (cmse_check_pointed_object): Likewise.
1374 2018-06-05 Martin Liska <mliska@suse.cz>
1376 PR gcov-profile/47618
1377 * doc/invoke.texi: Document how -fprofile-dir format
1380 2018-06-05 Richard Biener <rguenther@suse.de>
1382 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
1383 removal pretend DOM info isn't available so we do not update
1384 it and only remove edges, not dominated blocks. Actually free
1385 DOM info in case we removed something. Remove unreachable blocks.
1386 (mfb_keep_latches): Work with either DOM info or marked backedges.
1387 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
1388 first. Mark backedges if DOM info isn't available.
1389 (Re-)compute DOM info after cleanup_control_flow_pre.
1391 2018-06-05 Richard Biener <rguenther@suse.de>
1393 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
1394 (locus_discrim_hasher::hash): Adjust.
1395 (locus_discrim_hasher::equal): Likewise.
1396 (next_discriminator_for_locus): Work on line directly.
1397 (same_line_p): Pass in expanded locus1 as well.
1398 (assign_discriminators): Avoid redundant location expansions.
1400 2018-06-05 Richard Biener <rguenther@suse.de>
1402 PR tree-optimization/86046
1403 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
1404 if required after clearing TREE_ADDRESSABLE.
1406 2018-06-05 Richard Biener <rguenther@suse.de>
1408 PR tree-optimization/86047
1409 * tree-ssa-loop.c (for_each_index): Glob handling of all
1410 decls and constants and really handle all of them.
1412 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1415 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
1416 qualifier_void_pointer and qualifier_const_void_pointer.
1417 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
1418 (arm_init_builtins): Handle the above.
1419 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
1420 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
1421 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
1424 2018-06-05 Martin Liska <mliska@suse.cz>
1426 * auto-profile.c (read_autofdo_file): Do not use
1427 gcov_ctr_summary struct.
1428 (afdo_callsite_hot_enough_for_early_inline): Likewise.
1429 * coverage.c (struct counts_entry): Likewise.
1430 (read_counts_file): Read just single summary entry.
1431 (get_coverage_counts): Use gcov_summary struct.
1432 * coverage.h (get_coverage_counts): Likewise.
1433 * gcov-dump.c (dump_working_sets): Likewise.
1434 (tag_summary): Dump just single summary.
1435 * gcov-io.c (gcov_write_summary): Write just histogram
1437 (gcov_read_summary): Read just single summary.
1438 (compute_working_sets): Use gcov_summary struct.
1439 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
1440 of GCOV_COUNTERS_SUMMABLE.
1441 (GCOV_COUNTERS_SUMMABLE): Remove.
1442 (GCOV_FIRST_VALUE_COUNTER): Replace with
1443 GCOV_COUNTER_V_INTERVAL.
1444 (struct gcov_ctr_summary): Remove.
1445 (struct gcov_summary): Directly use fields of former
1447 (compute_working_sets): Use gcov_summary struct.
1448 * gcov.c (read_count_file): Do not use ctrs fields.
1449 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
1451 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
1453 * profile.c: Likewise.
1454 * profile.h: Likewise.
1456 2018-06-05 Martin Liska <mliska@suse.cz>
1458 PR gcov-profile/84846
1459 * gcov.c (output_lines): Print working directory only
1460 in intermediate format.
1462 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
1464 * config/s390/s390-builtin-types.def: Add void function type.
1465 * config/s390/s390-builtins.def: Use the function type for the
1468 2018-06-04 Jim Wilson <jimw@sifive.com>
1470 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
1472 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
1473 and maybe_eh_return. Change regno to unsigned int. Use new args to
1474 handle EH_RETURN_DATA_REGNO registers properly.
1475 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
1476 (riscv_expand_epilogue): Update comment. Change argument name and
1477 type. Update code to use new name and type. Pass new args to
1478 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
1480 * config/riscv/riscv.md (NORMAL_RETURN): New.
1481 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
1482 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
1483 (eh_return): Call gen_eh_return_internal and emit barrier.
1484 (eh_return_internal): Call riscv_expand_epilogue.
1486 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
1488 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
1489 bit_insertion field and declare can_be_merged_into method.
1490 (merged_store_group::can_be_merged_into): New method.
1491 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
1492 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
1493 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
1495 2018-06-04 Richard Biener <rguenther@suse.de>
1497 PR tree-optimization/85955
1498 * builtins.c (fold_builtin_sincos): Convert pointers to
1499 destination to appropriate type before dereferencing.
1501 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
1503 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
1505 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
1507 * expr.c (expand_expr_real_1): Force the operand into memory if
1508 its TYPE_MODE is BLKmode and if there is no integer mode for
1509 the number of bits being extracted.
1511 2018-06-04 Jakub Jelinek <jakub@redhat.com>
1515 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
1516 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
1518 2018-06-04 Richard Biener <rguenther@suse.de>
1520 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
1521 (cleanup_tree_cfg_noloop): ... single caller. Do
1522 start_recording_case_labels later.
1524 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
1526 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
1527 to _IMMINTRIN_H_INCLUDED.
1528 * config/i386/pconfigintrin.h: Ditto.
1529 * config/i386/waitpkgintrin.h: Ditto.
1530 * config/i386/immintrin.h: Add includes for sgxintrin.h,
1531 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
1532 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
1533 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
1534 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
1535 waitpkgintrin.h and cldemoteintrin.h.
1537 2018-06-04 Richard Biener <rguenther@suse.de>
1539 PR tree-optimization/86038
1540 * tracer.c (find_best_successor): Check probability for
1541 being initialized, bail out if not.
1543 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
1546 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
1547 of bits to ignore when comparing architectures.
1549 2018-06-04 Jakub Jelinek <jakub@redhat.com>
1551 PR tree-optimization/69615
1552 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
1553 maximum or minimum of the type, try to merge it also as if
1554 range1 is + [-, x - 1] or + [x + 1, -].
1557 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
1559 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
1561 PR tree-optimization/86034
1562 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
1563 the unsigned bitfield type in a bit insertion sequence if it does not
1564 have a larger precision than the bitfield size.
1565 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
1567 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
1569 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
1571 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
1573 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
1574 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
1575 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
1576 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
1578 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
1580 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
1581 Disable -fdelete-null-pointer-checks for ELF toolchain.
1583 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
1584 Kito Cheng <kito.cheng@gmail.com>
1586 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
1587 (nds32le-*-*, nds32be-*-*): Integrate checking process.
1588 (nds32*-*-*): Add glibc and uclibc conditions.
1589 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
1590 (TARGET_EXCEPT_UNWIND_INFO): Define.
1591 * config/nds32/elf.h: New file.
1592 * config/nds32/linux.h: New file.
1593 * config/nds32/nds32-elf.opt: New file.
1594 * config/nds32/nds32-linux.opt: New file.
1595 * config/nds32/nds32-fp-as-gp.c
1596 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
1597 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
1599 (nds32_asm_file_end): Ditto.
1600 (nds32_print_operand): Ditto.
1601 (nds32_insert_attributes): Ditto.
1602 (nds32_init_libfuncs): New function.
1603 (TARGET_HAVE_TLS): Define.
1604 (TARGET_INIT_LIBFUNCS): Define.
1605 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
1607 (TARGET_ELF): Apply different mcmodel setting.
1608 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
1609 been migrated into elf.h and linux.h files.
1610 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
1611 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
1612 (mcmodel): The content has been migrated into nds32-elf.opt and
1613 nds32-linux.opt files.
1614 * config/nds32/t-elf: New file.
1615 * config/nds32/t-linux: New file.
1617 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
1618 Shiva Chen <shiva0217@gmail.com>
1620 * config/nds32/constants.md (unspec_volatile_element): Add
1621 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
1622 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
1624 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
1625 (make_pass_nds32_fp_as_gp): Declare.
1626 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
1628 (nds32_asm_function_end_prologue): Remove unused asm output.
1629 (nds32_asm_function_begin_epilogue): Remove unused asm output.
1630 (nds32_asm_file_start): Output necessary fp_as_gp information.
1631 (nds32_option_override): Adjust register usage.
1632 (nds32_expand_prologue): Consider fp_as_gp situation.
1633 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
1634 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
1637 (simple_return): Ditto.
1638 (omit_fp_begin): Output special directive for fp_as_gp.
1639 (omit_fp_end): Output special directive for fp_as_gp.
1640 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
1641 mforbid-fp-as-gp): New options.
1643 2018-06-01 Mark Wielaard <mark@klomp.org>
1645 * dwarf2out.c (dwarf2out_finish): Remove generation of
1646 DW_AT_loclists_base.
1648 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
1650 * gimple-ssa-store-merging.c: Include gimple-fold.h.
1651 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
1652 (struct merged_store_group): Add bit_insertion field.
1653 (dump_char_array): Use standard hexadecimal format.
1654 (merged_store_group::merged_store_group): Set bit_insertion to false.
1655 (merged_store_group::apply_stores): Use optimal buffer size. Deal
1656 with BIT_INSERT_EXPR stores. Move up code updating the mask and
1657 also print the mask in the dump file.
1658 (pass_store_merging::gate): Minor tweak.
1659 (imm_store_chain_info::coalesce_immediate): Fix wrong association
1660 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
1661 stores with INTEGER_CST stores.
1662 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
1663 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
1664 and use it throughout. Generate bit insertion sequences if need be.
1665 (pass_store_merging::process_store): Remove redundant condition.
1666 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
1668 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
1670 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
1671 the 128-bit floating point types. Fix function comment.
1673 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1675 * config/aarch64/aarch64-simd.md
1676 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
1678 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
1681 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
1683 PR tree-optimization/85989
1684 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
1686 (backprop::intersect_uses): Check it when deciding whether this
1687 is a backedge reference.
1688 (backprop::process_block): Add each phi to m_visited_phis
1689 after visiting it, then clear it at the end.
1691 2018-06-01 Richard Biener <rguenther@suse.de>
1693 * tree-vectorizer.h (vect_dr_stmt): New function.
1694 (vect_get_load_cost): Adjust.
1695 (vect_get_store_cost): Likewise.
1696 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
1697 Use vect_dr_stmt instead of DR_SMTT.
1698 (vect_record_base_alignments): Likewise.
1699 (vect_calculate_target_alignment): Likewise.
1700 (vect_compute_data_ref_alignment): Likewise and make static.
1701 (vect_update_misalignment_for_peel): Likewise.
1702 (vect_verify_datarefs_alignment): Likewise.
1703 (vector_alignment_reachable_p): Likewise.
1704 (vect_get_data_access_cost): Likewise. Pass down
1705 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
1706 (vect_get_peeling_costs_all_drs): Likewise.
1707 (vect_peeling_hash_get_lowest_cost): Likewise.
1708 (vect_enhance_data_refs_alignment): Likewise.
1709 (vect_find_same_alignment_drs): Likewise.
1710 (vect_analyze_data_refs_alignment): Likewise.
1711 (vect_analyze_group_access_1): Likewise.
1712 (vect_analyze_group_access): Likewise.
1713 (vect_analyze_data_ref_access): Likewise.
1714 (vect_analyze_data_ref_accesses): Likewise.
1715 (vect_vfa_segment_size): Likewise.
1716 (vect_small_gap_p): Likewise.
1717 (vectorizable_with_step_bound_p): Likewise.
1718 (vect_prune_runtime_alias_test_list): Likewise.
1719 (vect_analyze_data_refs): Likewise.
1720 (vect_supportable_dr_alignment): Likewise.
1721 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
1722 (vect_gen_prolog_loop_niters): Likewise.
1723 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
1724 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
1726 (vect_recog_mask_conversion_pattern): Likewise.
1727 (vect_try_gather_scatter_pattern): Likewise.
1728 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
1729 to vect_get_store_cost.
1730 (vect_get_store_cost): Get stmt_info instead of DR.
1731 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
1732 (vect_get_load_cost): Get stmt_info instead of DR.
1734 2018-06-01 Richard Biener <rguenther@suse.de>
1737 * gimple-fold.c (var_decl_component_p): Also allow offsetted
1738 vars wrapped in MEM_REFs.
1740 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
1742 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
1743 Fix subreg tests so that we only return a choice between
1744 GENERAL_REGS and FP_REGS if the original classes included both.
1746 2018-06-01 Richard Biener <rguenther@suse.de>
1749 * tree-ssa-structalias.c (get_function_part_constraint):
1750 Handle NULL fi->decl.
1751 (find_func_aliases_for_call): Properly handle indirect
1752 fi from direct call.
1753 (find_func_clobbers): Likewise.
1754 (ipa_pta_execute): Likewise.
1755 (create_variable_info_for): For functions that are ifunc_resolver
1756 resolve to a varinfo that contains the result of the resolver call.
1757 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
1760 2018-05-31 Michael Collison <michael.collison@arm.com>
1762 * config/aarch64/aarch64.md:
1763 (*fix_to_zero_extenddfdi2): New pattern.
1764 * gcc.target/aarch64/fix_extend1.c: New testcase.
1766 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
1770 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
1771 and BUILT_IN_STRNCMP_EQ.
1772 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
1773 BUILT_IN_STRNCMP_EQ.
1774 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
1775 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
1776 (gimple_fold_builtin): Likewise.
1777 * tree-ssa-strlen.c (compute_string_length): New function.
1778 (determine_min_obsize): New function.
1779 (handle_builtin_string_cmp): New function to handle calls to
1780 string compare functions.
1781 (strlen_optimize_stmt): Add handling to builtin string compare
1783 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1784 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
1785 * tree.c (build_common_builtin_nodes): Add new defines of
1786 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
1788 2018-05-31 Jakub Jelinek <jakub@redhat.com>
1791 * bb-reorder.c (pass_partition_blocks::gate): Return false for
1792 functions with naked attribute.
1794 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
1796 * config/i386/sse.md (avx_vec_concat<mode>):
1797 Substitute concat_tg_mode mode attribute with xtg_mode.
1798 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
1799 (concat_tg_mode): Remove mode attribute.
1801 2018-05-31 Martin Sebor <msebor@redhat.com>
1804 * calls.c (alloc_max_size): Correct a logic error/typo.
1805 Treat excessive arguments as infinite. Warn for invalid arguments.
1806 * doc/invoke.texi (-Walloc-size-larger-than): Update.
1808 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
1811 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
1812 and movx for Haswell.
1814 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
1815 Cesar Philippidis <cesar@codesourcery.com>
1818 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
1819 when emitting error on private/firstprivate reductions.
1820 * omp-low.c (lower_omp_target): Avoid reference-type processing
1821 on pointers for firstprivate clause.
1823 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
1825 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
1828 * config/aarch64/aarch64-simd.md
1829 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
1830 (aarch64_ld1_x3_<mode>): Likewise
1831 (aarch64_st1x2<VALLDIF:mode>): Likewise
1832 (aarch64_st1_x2_<mode>): Likewise
1833 (aarch64_st1x3<VALLDIF:mode>): Likewise
1834 (aarch64_st1_x3_<mode>): Likewise
1835 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
1836 (vld1_s8_x3): Likewise.
1837 (vld1_u16_x3): Likewise.
1838 (vld1_s16_x3): Likewise.
1839 (vld1_u32_x3): Likewise.
1840 (vld1_s32_x3): Likewise.
1841 (vld1_u64_x3): Likewise.
1842 (vld1_s64_x3): Likewise.
1843 (vld1_f16_x3): Likewise.
1844 (vld1_f32_x3): Likewise.
1845 (vld1_f64_x3): Likewise.
1846 (vld1_p8_x3): Likewise.
1847 (vld1_p16_x3): Likewise.
1848 (vld1_p64_x3): Likewise.
1849 (vld1q_u8_x3): Likewise.
1850 (vld1q_s8_x3): Likewise.
1851 (vld1q_u16_x3): Likewise.
1852 (vld1q_s16_x3): Likewise.
1853 (vld1q_u32_x3): Likewise.
1854 (vld1q_s32_x3): Likewise.
1855 (vld1q_u64_x3): Likewise.
1856 (vld1q_s64_x3): Likewise.
1857 (vld1q_f16_x3): Likewise.
1858 (vld1q_f32_x3): Likewise.
1859 (vld1q_f64_x3): Likewise.
1860 (vld1q_p8_x3): Likewise.
1861 (vld1q_p16_x3): Likewise.
1862 (vld1q_p64_x3): Likewise.
1863 (vst1_s64_x2): Likewise.
1864 (vst1_u64_x2): Likewise.
1865 (vst1_f64_x2): Likewise.
1866 (vst1_s8_x2): Likewise.
1867 (vst1_p8_x2): Likewise.
1868 (vst1_s16_x2): Likewise.
1869 (vst1_p16_x2): Likewise.
1870 (vst1_s32_x2): Likewise.
1871 (vst1_u8_x2): Likewise.
1872 (vst1_u16_x2): Likewise.
1873 (vst1_u32_x2): Likewise.
1874 (vst1_f16_x2): Likewise.
1875 (vst1_f32_x2): Likewise.
1876 (vst1_p64_x2): Likewise.
1877 (vst1q_s8_x2): Likewise.
1878 (vst1q_p8_x2): Likewise.
1879 (vst1q_s16_x2): Likewise.
1880 (vst1q_p16_x2): Likewise.
1881 (vst1q_s32_x2): Likewise.
1882 (vst1q_s64_x2): Likewise.
1883 (vst1q_u8_x2): Likewise.
1884 (vst1q_u16_x2): Likewise.
1885 (vst1q_u32_x2): Likewise.
1886 (vst1q_u64_x2): Likewise.
1887 (vst1q_f16_x2): Likewise.
1888 (vst1q_f32_x2): Likewise.
1889 (vst1q_f64_x2): Likewise.
1890 (vst1q_p64_x2): Likewise.
1891 (vst1_s64_x3): Likewise.
1892 (vst1_u64_x3): Likewise.
1893 (vst1_f64_x3): Likewise.
1894 (vst1_s8_x3): Likewise.
1895 (vst1_p8_x3): Likewise.
1896 (vst1_s16_x3): Likewise.
1897 (vst1_p16_x3): Likewise.
1898 (vst1_s32_x3): Likewise.
1899 (vst1_u8_x3): Likewise.
1900 (vst1_u16_x3): Likewise.
1901 (vst1_u32_x3): Likewise.
1902 (vst1_f16_x3): Likewise.
1903 (vst1_f32_x3): Likewise.
1904 (vst1_p64_x3): Likewise.
1905 (vst1q_s8_x3): Likewise.
1906 (vst1q_p8_x3): Likewise.
1907 (vst1q_s16_x3): Likewise.
1908 (vst1q_p16_x3): Likewise.
1909 (vst1q_s32_x3): Likewise.
1910 (vst1q_s64_x3): Likewise.
1911 (vst1q_u8_x3): Likewise.
1912 (vst1q_u16_x3): Likewise.
1913 (vst1q_u32_x3): Likewise.
1914 (vst1q_u64_x3): Likewise.
1915 (vst1q_f16_x3): Likewise.
1916 (vst1q_f32_x3): Likewise.
1917 (vst1q_f64_x3): Likewise.
1918 (vst1q_p64_x3): Likewise.
1920 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1922 * config/msp430/msp430.c (msp430_output_labelref): Prepend
1923 user_label_prefix to name.
1925 * tree-core.h: Update comment about the format of NAME string
1926 passed to handler in attribute_spec.
1928 * config/msp430/msp430.md: Remove erroneous subreg expression from
1929 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
1930 zero_extend{q,h}isi2.
1932 2018-05-30 Borislav Petkov <bp@suse.de>
1934 * doc/extend.texi: Document some architecture specific
1935 constraints and sort entries.
1937 2018-05-30 Martin Sebor <msebor@redhat.com>
1940 * builtins.c (expand_builtin_stpcpy_1): New function.
1941 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
1942 only if the former succeeds.
1944 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
1946 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
1949 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
1951 * doc/invoke.texi (-flinker-output): Document
1953 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
1955 * passes.c (ipa_write_summaries): Only modify statements if body
1957 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
1958 incrementally linking.
1959 (ipa_passes): Likewise.
1960 * lto-cgraph.c (lto_output_node): When incrementally linking do not
1961 pass down resolution info.
1962 * common.opt (flag_incremental_link): Update info.
1963 * gcc.c (plugin specs): Turn flinker-output=* to
1964 -plugin-opt=-linker-output-known
1965 * toplev.c (compile_file): Also cut compilation when doing incremental
1967 * flag-types. (enum lto_partition_model): Add
1968 LTO_LINKER_OUTPUT_NOLTOREL.
1969 (invoke.texi): Add -flinker-output docs.
1970 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
1971 link same way as WPA; do not stream in dead initializers.
1973 * dwarf2out.c (dwarf2out_die_ref_for_decl,
1974 darf2out_register_external_decl): Support incremental link.
1976 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
1978 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
1980 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
1982 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
1983 it down to simple_object_copy_lto_debug_sections.
1984 (run_gcc): Determine incremental LTO link time and configure
1985 lto1 into non-wpa mode, disable renaming of debug sections.
1987 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
1989 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
1990 descriptions of various incorrectly documented functions.
1992 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
1995 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
1996 address check not strict.
1998 2018-05-30 Richard Biener <rguenther@suse.de>
2000 PR tree-optimization/85964
2001 * tracer.c (better_p): Drop initialized count check, we only
2002 call the function with initialized counts now.
2003 (find_best_successor): Do find a best edge if one
2004 has uninitialized count.
2005 (find_best_predecessor): Likewise. Do BB frequency check only
2006 if count is initialized.
2008 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
2010 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
2011 (aarch64_ldrstr_offset_compare): New.
2012 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
2013 load/store orderings.
2014 (aarch64_gen_adjusted_ldpstp): Likewise.
2016 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
2018 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2019 Check for subset of GENERAL_REGS and FP_REGS.
2020 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
2023 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
2025 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
2026 and wi::to_poly_offset. Add the current offset and then check
2027 whether the sum fits, rather than using an unchecked addition of
2028 a checked term. Check for a shwi rather than a uhwi.
2029 * expr.c (get_bit_range): Use tree_to_poly_uint64.
2030 (store_constructor): Use poly_int_tree_p.
2031 (expand_expr_real_1): Likewise.
2032 * function.c (assign_temp): Likewise.
2033 * fold-const.c (const_binop): Use poly_int_tree_p and
2035 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
2037 * ipa-icf-gimple.c (func_checker::compare_operand): Use
2038 to_poly_offset for MEM offsets.
2039 * ipa-icf.c (sem_variable::equals): Likewise.
2040 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
2041 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
2042 wi::to_poly_offset for BIT_FIELD_REF offsets.
2043 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
2045 * var-tracking.c (emit_note_insn_var_location): Use
2046 tree_to_poly_uint64.
2048 2018-05-29 Jim Wilson <jimw@sifive.com>
2050 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
2052 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
2055 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
2056 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
2058 (sse4_1_round<mode>2): Use nonimmediate_operand
2059 for operand 1 predicate.
2061 2018-05-29 Martin Sebor <msebor@redhat.com>
2062 Richard Biener <rguenther@suse.de>
2065 * calls.c (get_size_range): Call determine_value_range instead
2066 of get_value_range..
2067 * tree-vrp.h (determine_value_range): Declared new function.
2068 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
2070 2018-05-29 Richard Biener <rguenther@suse.de>
2072 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
2073 sure to use non-pattern stmts for get_earlier_stmt arguments.
2074 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
2075 called on pattern stmts.
2076 (get_later_stmt): Likewise.
2078 2018-05-29 Martin Liska <mliska@suse.cz>
2080 PR gcov-profile/85759
2081 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
2084 2018-05-29 Jakub Jelinek <jakub@redhat.com>
2086 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
2088 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
2092 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
2093 VEC_PACK_FLOAT_EXPR): New tree codes.
2094 * tree-pretty-print.c (op_code_prio): Handle
2095 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
2096 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
2097 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
2098 * tree-inline.c (estimate_operator_cost): Likewise.
2099 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
2100 * fold-const.c (const_binop): Likewise.
2101 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
2102 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
2103 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2104 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
2105 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
2106 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
2107 * expr.c (expand_expr_real_2): Likewise.
2108 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
2109 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
2110 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
2112 * optabs.c (expand_widen_pattern_expr): For
2113 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
2114 sign from result type rather than operand's type.
2115 (expand_binop_directly): For vec_packu_float_optab and
2116 vec_packs_float_optab allow result type to be different from operand's
2118 * optabs-tree.c (optab_for_tree_code): Handle
2119 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
2120 VEC_PACK_FLOAT_EXPR. Formatting fixes.
2121 * tree-vect-generic.c (expand_vector_operations_1): Handle
2122 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
2123 VEC_PACK_FLOAT_EXPR.
2124 * tree-vect-stmts.c (supportable_widening_operation): Handle
2126 (supportable_narrowing_operation): Handle FLOAT_EXPR.
2127 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
2128 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
2129 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
2130 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
2132 (vec_pack<floatprefix>_float_<mode>): New expander.
2133 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
2135 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
2136 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
2137 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
2138 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
2139 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
2141 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
2142 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
2143 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
2144 VEC_PACK_FLOAT_EXPR): Document.
2146 2018-05-29 Richard Biener <rguenther@suse.de>
2148 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
2150 (stmt_vec_info_vec): Make pointer.
2151 (init_stmt_vec_info_vec): Remove.
2152 (free_stmt_vec_info_vec): Likewise.
2153 (set_stmt_vec_info_vec): New function.
2154 (free_stmt_vec_infos): Likewise.
2155 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
2156 (set_vinfo_for_stmt): Likewise.
2157 (get_earlier_stmt): Likewise.
2158 (get_later_stmt): Likewise.
2159 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
2160 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
2161 (vec_info::~vec_info): Free stmt_vec_infos.
2162 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
2163 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
2164 (pass_slp_vectorize::execute): Likewise.
2165 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
2166 (free_stmt_vec_info_vec): Likewise.
2167 (set_stmt_vec_info_vec): New function.
2168 (free_stmt_vec_infos): Likewise.
2169 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
2170 the global stmt_vec_info_vec.
2171 * tree-parloops.c (gather_scalar_reductions): Use
2172 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
2175 2018-05-29 Richard Biener <rguenther@suse.de>
2177 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
2179 2018-05-29 Martin Liska <mliska@suse.cz>
2180 David Malcolm <dmalcolm@redhat.com>
2182 * vec.c (test_reverse): New.
2183 (vec_c_tests): Add new test.
2184 * vec.h (vl_ptr>::reverse): New function.
2186 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
2188 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
2190 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
2193 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
2195 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
2197 2018-05-28 Richard Biener <rguenther@suse.de>
2199 PR tree-optimization/85933
2200 * tree-vect-data-refs.c (vect_record_base_alignments): Only
2201 look at stmts marked as vectorizable.
2203 2018-05-28 Richard Biener <rguenther@suse.de>
2205 PR tree-optimization/85934
2206 * tree-vect-generic.c (expand_vector_operations_1): Hoist
2207 vector boolean check before scalar optimization.
2209 2018-05-28 Jakub Jelinek <jakub@redhat.com>
2211 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
2214 2018-05-28 Mark Wielaard <mark@klomp.org>
2216 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
2217 if it is an expression containing a minus sign.
2219 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
2221 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
2223 2018-05-27 Paul Koning <ni1d@arrl.net>
2225 * config/pdp11/pdp11.md (truncsihi2): Remove.
2227 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
2228 Chung-Ju Wu <jasonwucj@gmail.com>
2230 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
2232 (unaligned_store_dw): Ditto.
2233 * config/nds32/nds32-memory-manipulation.c
2234 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
2235 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
2236 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
2237 (emit_setmem_word_loop): Rename to ...
2238 (emit_setmem_doubleword_loop): ... this.
2239 (nds32_gen_dup_4_byte_to_word_value): New function.
2240 (nds32_gen_dup_8_byte_to_double_word_value): New function.
2241 (nds32_expand_setmem_loop): Refine implementation.
2242 (nds32_expand_setmem_loop_v3m): Ditto.
2243 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
2246 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
2248 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
2250 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
2252 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
2253 (nds32_init_machine_status): Initialize machine->attr_naked_p and
2254 machine->attr_no_prologue_p.
2255 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
2256 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
2257 (nds32_expand_epilogue): Consider attr_naked_p.
2258 (nds32_expand_epilogue_v3pop): Likewise.
2259 (nds32_can_use_return_insn): Likewise.
2260 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
2261 attr_no_prologue_p fields.
2262 * config/nds32/nds32.opt (mret-in-naked-func): New option.
2264 2018-05-27 Jakub Jelinek <jakub@redhat.com>
2267 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
2269 * config/i386/sse.md
2270 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
2272 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
2274 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
2276 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
2278 (*<floatsuffix>floatv2div2sf2): Rename to ...
2279 (*float<floatunssuffix>v2div2sf2): ... this.
2280 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
2281 (float<floatunssuffix>v2div2sf2_mask): ... this.
2282 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
2283 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
2284 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
2286 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
2288 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
2290 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
2292 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
2294 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
2296 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
2297 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
2298 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
2299 gen_ufix_truncv8dfv8si2.
2300 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
2301 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
2302 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
2303 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
2304 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
2305 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
2306 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
2307 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
2309 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
2313 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
2315 2018-05-25 Jim Wilson <jimw@sifive.com>
2317 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
2318 * config/riscv/riscv.c (struct machine_function): Add
2319 interrupt_handler_p and attribute_checked_p fields.
2320 (riscv_attribute_table): Add interrupt.
2321 (riscv_interrupt_type_p): New.
2322 (riscv_save_reg_p): Save extra regs for interrupt handler.
2323 (riscv_use_save_libcall): Return false for interrupt handler.
2324 (riscv_first_stack_step): Add forward declaration.
2325 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
2326 for interrupt handler with large frame. Use it for saved reg list.
2327 (riscv_expand_prologue): Move flag_stack_usage_info support to
2328 eliminate duplication.
2329 (riscv_expand_epilogue): Generate mret for interrupt handler.
2330 (riscv_epilogue_uses): New.
2331 (riscv_can_use_return_insn): Return false for interrupt handler.
2332 (riscv_function_ok_for_sibcall): Likewise.
2333 (riscv_set_current_function): Add interrupt handler support.
2334 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
2335 * config/riscv/riscv.md (UNSPECV_MRET): New.
2337 (riscv_frflags, riscv_fsflags): Use tab after opcode.
2339 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
2341 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
2343 PR tree-optimization/85712
2344 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
2345 this candidate has already been replaced in-situ by a copy.
2347 2018-05-25 Jason Merrill <jason@redhat.com>
2349 PR c++/80485 - inline function non-zero address.
2350 * symtab.c (nonzero_address): Check DECL_COMDAT.
2352 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
2355 * config/alpha/alpha.md (ashlsi3): New insn pattern.
2356 (*ashlsi_se): Rename from *ashldi_se. Define as sign
2357 extension of SImode operation. Use const123_operand predicate.
2358 (*saddsi_1): Remove.
2359 (*saddl_se_1): Ditto.
2361 (*ssubl_se_1): Ditto.
2362 * config/alpha/predicates.md (const123_operand): New predicate.
2363 * config/alpha/constraints.md (P): Use IN_RANGE.
2365 2018-05-25 Richard Biener <rguenther@suse.de>
2367 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
2369 (ref_maybe_used_by_stmt_p): Likewise.
2370 (stmt_may_clobber_ref_p): Likewise.
2371 (stmt_may_clobber_ref_p_1): Likewise.
2372 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
2374 (ref_maybe_used_by_stmt_p): Likewise.
2375 (stmt_may_clobber_ref_p): Likewise.
2376 (stmt_may_clobber_ref_p_1): Likewise.
2377 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
2378 the alias oracle to disambiguate DRs with stmts DR analysis
2380 (vect_analyze_data_refs): Do not give up on not analyzable
2381 DRs for BB vectorization. Remove code truncating the dataref
2384 2018-05-25 Jakub Jelinek <jakub@redhat.com>
2387 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
2388 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
2389 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
2391 2018-05-25 Richard Biener <rguenther@suse.de>
2393 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
2394 function, combining stmt data ref gathering and fatal analysis
2396 (vect_analyze_data_refs): Remove now redudnant code and simplify.
2397 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
2398 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
2399 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
2400 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
2402 2018-05-25 Bin Cheng <bin.cheng@arm.com>
2404 PR tree-optimization/85720
2405 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
2406 SCC if all partitions are builtins.
2407 (version_loop_by_alias_check): New parameter. Generate cancelable
2408 runtime alias check if all partitions are builtins.
2409 (distribute_loop): Update call to above function.
2411 2018-05-25 Bin Cheng <bin.cheng@arm.com>
2413 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
2414 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
2415 (parm_default_def_partition_arg): Ditto.
2416 (set_parm_default_def_partition): Ditto.
2417 (get_parm_default_def_partitions): Ditto and make it static.
2418 (get_undefined_value_partitions): Ditto and make it static.
2419 (remove_ssa_form): Refactor call to init_var_map here.
2420 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
2421 computation for loop region.
2422 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
2423 (register_default_def): Delete.
2424 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
2425 (parm_default_def_partition_arg): Ditto.
2426 (set_parm_default_def_partition): Ditto.
2427 (get_parm_default_def_partitions): Ditto and make it static.
2428 (get_undefined_value_partitions): Ditto and make it static.
2429 (coalesce_with_default, coalesce_with_default): Update comment.
2430 (create_coalesce_list_for_region): New func factored out from
2431 create_outofssa_var_map.
2432 (populate_coalesce_list_for_outofssa): New func factored out from
2433 create_outofssa_var_map and coalesce_ssa_name.
2434 (create_outofssa_var_map): Delete.
2435 (coalesce_ssa_name): Refactor to support live range computation.
2436 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
2437 (get_parm_default_def_partitions): Delete.
2438 (get_undefined_value_partitions): Ditto.
2439 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
2440 computation for loop region.
2441 (new_tree_live_info, loe_visit_block): Ditto.
2442 (live_worklist, set_var_live_on_entry): Ditto.
2443 (calculate_live_on_exit, verify_live_on_entry): Ditto.
2444 * tree-ssa-live.h (struct _var_map): New fields.
2445 (init_var_map): Change decl.
2446 (region_contains_p): New.
2448 2018-05-25 Bin Cheng <bin.cheng@arm.com>
2450 * tree-ssa-live.h (live_merge_and_clear): Delete.
2452 2018-05-25 Richard Biener <rguenther@suse.de>
2455 * tree-dump.c (dequeue_and_dump): Remove access to removed
2456 operand 2 of a SWITCH_EXPR.
2458 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2460 * doc/sourcebuild.texi (vect_double_cond_arith): Include
2461 multiplication and division.
2462 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
2463 (cond_udiv@var{m}, cond_umod@var{m}): Document.
2464 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
2465 (cond_udiv_optab, cond_umod_optab): New optabs.
2466 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
2467 (IFN_COND_RDIV): New internal functions.
2468 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
2469 TRUNC_MOD_EXPR and RDIV_EXPR.
2470 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
2471 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
2473 (SVE_INT_BINARY): Include mult.
2474 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
2475 (optab, sve_int_op): Handle mult.
2476 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
2478 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
2479 for SVE_INT_BINARY_SD.
2481 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2483 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
2484 (optab, sve_int_op): Handle div and udiv.
2485 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
2486 for SVE_INT_BINARY_SD.
2487 (*<optab><mode>3): New insn for the same.
2489 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2491 * tree-vect-patterns.c: Include predict.h.
2492 (vect_recog_divmod_pattern): Restrict check for division support
2493 to when optimizing for size.
2495 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2497 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
2498 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
2499 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
2500 (gimple_match_op::set_op): Likewise.
2501 (gimple_resimplify4): Declare.
2502 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
2503 (expr::gen_transform): Likewise.
2504 (decision_tree::gen): Generate a simplification routine for 4 operands.
2505 * gimple-match-head.c (gimple_simplify): Add an overload for
2506 4 operands. In the top-level function, handle up to 4 call
2507 arguments and call gimple_resimplify4.
2508 (gimple_resimplify4): New function.
2509 (build_call_internal): Pass a fourth operand.
2510 (maybe_push_to_seq): Likewise.
2511 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
2512 Fold VEC_COND_EXPRs of an operation and a default value into
2513 an IFN_COND_* function if possible.
2514 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
2516 (SVE_COND_FP_BINARY): Include them.
2517 (optab, sve_fp_op): Handle them.
2518 (SVE_INT_BINARY_REV): New code iterator.
2519 (SVE_COND_FP_BINARY_REV): New int iterator.
2520 (commutative): New int attribute.
2521 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
2523 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
2525 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
2526 (*cond_<optab><mode>): New patterns for reversed operands.
2528 2018-05-25 Richard Biener <rguenther@suse.de>
2530 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
2531 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
2532 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
2533 (STMT_VINFO_GROUPED_ACCESS): Adjust.
2534 * tree-vect-data-refs.c (everywhere): Adjust users.
2535 * tree-vect-loop.c (everywhere): Likewise.
2536 * tree-vect-slp.c (everywhere): Likewise.
2537 * tree-vect-stmts.c (everywhere): Likewise.
2538 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
2540 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2542 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
2544 (gcc_cv_as_section_exclude): ... this.
2545 Try Solaris as #exclude syntax.
2546 * configure: Regenerate.
2547 * config.in: Regenerate.
2548 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
2550 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
2551 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
2553 * varasm.c (default_elf_asm_named_section): Don't check if
2554 HAVE_GAS_SECTION_EXCLUDE is defined.
2556 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2558 * doc/md.texi: Update the documentation of the cond_* optabs
2559 to mention the new final operand. Fix GET_MODE_NUNITS call.
2560 Describe the scalar case too.
2561 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
2562 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
2564 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
2565 (get_conditional_internal_fn): Update comment.
2566 * tree-vect-loop.c (vectorizable_reduction): Pass the original
2567 accumulator value as a final argument to conditional functions.
2568 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
2569 a define_expand and add an "else" operand. Assert for now that
2570 the else operand is equal to operand 2. Use SVE_INT_BINARY and
2571 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
2572 (*cond_<optab><mode>): New patterns.
2573 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
2574 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
2575 (UNSPEC_COND_EOR): Delete.
2576 (optab): Remove associated mappings.
2577 (SVE_INT_BINARY): New code iterator.
2578 (sve_int_op): Remove int attribute and add "minus" to the code
2580 (SVE_COND_INT_OP): Delete.
2581 (SVE_COND_FP_OP): Rename to...
2582 (SVE_COND_FP_BINARY): ...this.
2584 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
2586 * optabs.c (can_reuse_operands_p): New function.
2587 (maybe_legitimize_operands): Try to reuse the results for
2590 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
2592 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
2593 Add {q} suffix to insn mnemonic.
2595 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2597 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
2598 (msp430_warn_func_return): New.
2600 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
2602 * fold-const.c (tree_nonzero_bits): New function.
2603 * fold-const.h (tree_nonzero_bits): Likewise.
2604 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
2605 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
2607 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
2611 * varasm.c (assemble_alias): Check ifunc_resolver only on
2614 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
2617 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
2618 when memory input operand is handled.
2620 2018-05-24 Luis Machado <luis.machado@linaro.org>
2622 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
2624 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
2626 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
2628 * match.pd: Delay FMA folds until after vectorization.
2630 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
2633 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
2634 address check not strict.
2636 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
2638 * gimple-match.h (gimple_match_op): New class.
2639 (mprts_hook): Replace parameters with a gimple_match_op *.
2640 (maybe_build_generic_op): Likewise.
2641 (gimple_simplified_result_is_gimple_val): Replace parameters with
2642 a const gimple_match_op *.
2643 (gimple_simplify): Replace code_helper * and tree * parameters with
2644 a gimple_match_op * parameter.
2645 (gimple_resimplify1): Replace code_helper *, tree and tree *
2646 parameters with a gimple_match_op * parameter.
2647 (gimple_resimplify2): Likewise.
2648 (gimple_resimplify3): Likewise.
2649 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
2650 parameters with a gimple_match_op * parameter.
2651 * gimple-match-head.c (gimple_simplify): Change prototypes of
2652 auto-generated functions to take a gimple_match_op * instead of
2653 separate code_helper * and tree * parameters. Make the same
2654 change in the top-level overload and update calls to the
2655 gimple_resimplify routines. Update calls to the auto-generated
2656 functions and to maybe_push_res_to_seq in the publicly-facing
2657 operation-specific gimple_simplify overloads.
2658 (gimple_match_op::MAX_NUM_OPS): Define.
2659 (gimple_resimplify1): Replace rcode and ops with a single res_op
2660 parameter. Update call to gimple_simplify.
2661 (gimple_resimplify2): Likewise.
2662 (gimple_resimplify3): Likewise.
2663 (mprts_hook): Replace parameters with a gimple_match_op *.
2664 (maybe_build_generic_op): Likewise.
2665 (build_call_internal): Replace type, nargs and ops with
2666 a gimple_match_op *.
2667 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
2668 with a single gimple_match_op *. Update calls to mprts_hook,
2669 build_call_internal and gimple_simplified_result_is_gimple_val.
2670 Factor out code that is common to the tree_code and combined_fn cases.
2671 * genmatch.c (expr::gen_transform): Replace tem_code and
2672 tem_ops with a gimple_match_op called tem_op. Update calls
2673 to the gimple_resimplify functions and maybe_push_res_to_seq.
2674 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
2675 res_ops. Update call to the gimple_resimplify functions.
2676 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
2677 (decision_tree::gen): Make the functions take a gimple_match_op *
2678 called res_op instead of separate res_code and res_ops parameters.
2679 Update call accordingly.
2680 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
2681 and ops with a single res_op parameter. Update calls to
2682 maybe_build_generic_op and maybe_push_res_to_seq.
2683 (fold_stmt_1): Update calls to gimple_simplify and
2684 replace_stmt_with_simplification.
2685 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
2686 and gimple_simplified_result_is_gimple_val.
2687 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
2689 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
2690 with a gimple_match_op *.
2691 (vn_nary_build_or_lookup): Likewise. Update call to
2692 vn_nary_build_or_lookup_1.
2693 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
2694 gimple_match_op *. Update calls to the gimple_resimplify routines
2695 and to gimple_simplified_result_is_gimple_val.
2696 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
2697 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
2698 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
2699 (visit_nary_op): Likewise.
2700 (visit_reference_op_load): Likewise.
2702 2018-05-23 Luis Machado <luis.machado@linaro.org>
2704 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
2705 modifier for printing the step amount.
2707 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2710 * gcc/tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
2713 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
2715 * doc/sourcebuild.texi (Endianness): New subsubsection.
2717 2018-05-23 Luis Machado <luis.machado@linaro.org>
2719 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
2720 <prefetch_dynamic_strides>: New const bool field.
2721 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
2722 prefetch_dynamic_strides.
2723 (exynosm1_prefetch_tune): Likewise.
2724 (thunderxt88_prefetch_tune): Likewise.
2725 (thunderx_prefetch_tune): Likewise.
2726 (thunderx2t99_prefetch_tune): Likewise.
2727 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
2729 (aarch64_override_options_internal): Update to set
2730 PARAM_PREFETCH_DYNAMIC_STRIDES.
2731 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
2732 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
2733 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
2734 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
2735 prefetch-dynamic-strides setting.
2737 2018-05-23 Luis Machado <luis.machado@linaro.org>
2739 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
2740 <minimum_stride>: New const int field.
2741 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
2742 minimum_stride field defaulting to -1.
2743 (exynosm1_prefetch_tune): Likewise.
2744 (thunderxt88_prefetch_tune): Likewise.
2745 (thunderx_prefetch_tune): Likewise.
2746 (thunderx2t99_prefetch_tune): Likewise.
2747 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
2748 <default_opt_level>: Set to 3.
2749 (aarch64_override_options_internal): Update to set
2750 PARAM_PREFETCH_MINIMUM_STRIDE.
2751 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
2752 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
2753 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
2754 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
2755 stride is constant and is below the minimum stride threshold.
2757 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2759 * config/arm/arm-cpus.in (mode26): Delete.
2760 (armv4): Delete mode26 reference.
2761 * config/arm/arm.c (arm_configure_build_target): Delete use of
2764 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
2766 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
2768 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
2769 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
2771 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
2772 (floatunsdidf2): Ditto.
2774 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
2776 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
2777 (fixuns_trunc<mode>si2_avx512f): Ditto.
2778 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
2779 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
2780 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
2782 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
2784 PR rtl-optimization/79985
2785 * df-scan.c (df_insn_refs_collect): Remove special case for
2786 global registers and asm statements.
2788 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
2790 * extend.texi (Global Register Variables): Rewrite the bullet list.
2791 Note that the register is available for allocation. Note that access
2792 via inline asm must use constraints. Add note about async-signal
2793 handlers. Remove paragraph about automagic register selection.
2795 2018-05-23 Richard Biener <rguenther@suse.de>
2797 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
2798 of fixed offset from memset VN.
2800 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
2802 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
2804 (alloc_cand_and_find_basis): Initialize first_interp field.
2805 (slsr_process_mul): Modify first_interp field.
2806 (slsr_process_add): Likewise.
2807 (slsr_process_cast): Modify first_interp field for each new
2809 (slsr_process_copy): Likewise.
2810 (dump_candidate): Dump first_interp field.
2811 (replace_mult_candidate): Process all interpretations, not just
2813 (replace_rhs_if_not_dup): Likewise.
2814 (replace_one_candidate): Likewise.
2816 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
2818 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
2820 (aarch64_needs_frame_chain): New function.
2821 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
2823 2018-05-23 Sudakshina Das <sudi.das@arm.com>
2826 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
2827 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
2828 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
2829 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
2830 as true for strict-align.
2831 (aarch64_can_inline_p): Perform checks even when callee has no
2832 attributes to check for strict alignment.
2833 * doc/extend.texi (AArch64 Function Attributes): Document
2835 * doc/invoke.texi: (AArch64 Options): Likewise.
2837 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
2839 PR tree-optimization/85853
2840 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
2841 the handling of the root of the node to...
2842 (vect_slp_analyze_node_operations_1): ...this new function,
2843 and run the whole thing with the child nodes' def types
2844 set according to their SLP node's def type.
2846 2018-05-23 Richard Biener <rguenther@suse.de>
2849 * tree-data-ref.c (create_runtime_alias_checks): Defer
2850 and ignore overflow warnings.
2852 2018-05-23 Yury Gribov <tetra2005@gmail.com>
2854 PR tree-optimization/85822
2855 * tree-vrp.c (is_masked_range_test): Fix handling of negative
2858 2018-05-23 Richard Biener <rguenther@suse.de>
2860 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
2861 memset constants via native_interpret_expr.
2863 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
2866 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
2868 (cgraph_node::create_alias): Likewise.
2869 (cgraph_node::get_availability): Check ifunc_resolver instead
2870 of looking up ifunc attribute.
2871 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
2872 * varasm.c (do_assemble_alias): Likewise.
2873 (assemble_alias): Likewise.
2874 (default_binds_local_p_3): Likewise.
2875 * cgraph.h (cgraph_node): Add ifunc_resolver.
2876 (cgraph_node::only_called_directly_or_aliased_p): Return false
2878 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
2880 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
2881 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
2882 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
2883 instead of looking up ifunc attribute.
2885 2018-05-22 Luis Machado <luis.machado@linaro.org>
2887 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
2889 2018-05-22 Martin Sebor <msebor@redhat.com>
2892 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
2893 only when expasion succeeds.
2894 (expand_builtin_strcmp): Same.
2895 (expand_builtin_strncmp): Same.
2897 2018-05-22 Martin Sebor <msebor@redhat.com>
2899 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
2901 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
2902 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2904 * config/aarch64/aarch64-ldpstp.md: Replace uses of
2905 aarch64_mem_pair_operand with memory_operand and delete operand swapping
2907 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
2908 Add check for legitimate_address.
2909 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
2910 (aarch64_swap_ldrstr_operands): New.
2911 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
2914 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
2915 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2917 * config/aarch64/aarch64.md: New patterns to generate stp
2919 (store_pair_sw, store_pair_dw): New patterns to generate stp for
2920 single words and double words.
2921 (load_pair_sw, load_pair_dw): Likewise.
2922 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
2924 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
2926 * config/aarch64/aarch64-ldpstp.md: Modify peephole
2927 for different mode ldpstp and add peephole for merged zero stores.
2929 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
2931 (aarch64_gen_store_pair): Rename calls to match new patterns.
2932 (aarch64_gen_load_pair): Rename calls to match new patterns.
2933 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
2934 (load_pair<DREG:mode><DREG2:mode>): ... This.
2935 (store_pair<mode>): Rename to...
2936 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
2937 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
2939 (V_INT_EQUIV): Handle SImode.
2940 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
2943 2018-05-22 Martin Sebor <msebor@redhat.com>
2946 * calls.c (maybe_warn_nonstring_arg): Use string length to set
2947 or ajust the presumed bound on an operation to avoid unnecessary
2950 2018-05-22 Martin Sebor <msebor@redhat.com>
2952 PR tree-optimization/85826
2953 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
2954 assuming that a DECL necesarily has a constant size.
2956 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
2959 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
2961 2018-05-22 Richard Biener <rguenther@suse.de>
2963 PR tree-optimization/85834
2964 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
2965 non-constant and non-zero memset arguments.
2967 2018-05-22 Martin Liska <mliska@suse.cz>
2970 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
2972 2018-05-22 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/85863
2975 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
2976 comparisons when vectype is specified.
2977 (vectorizable_condition): Do not specify vectype for
2978 vect_is_simple_cond when SLP vectorizing.
2980 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
2983 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
2984 define __ibm128 as long double.
2985 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
2986 as a distinct type when IEEE 128-bit support is enabled.
2987 (init_float128_ieee): Fix up conversions between IFmode and IEEE
2988 128-bit types to use the correct functions.
2989 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
2990 convert between 128-bit floating point types that have different
2991 modes but the same representation, instead of using gen_lowpart to
2993 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
2995 (IFKF_reg): New attributes to give the register constraints for
2997 (extend<mode>tf2_internal): New insns to mark an explicit
2998 conversion between 128-bit floating point types that have a
2999 different mode but share the same representation.
3001 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
3003 PR tree-optimization/85814
3004 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
3005 a null return from get_strinfo when unsharing the next
3006 strinfo in the chain.
3008 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
3011 * varasm.c (weak_finish): Clean up weak_decls.
3013 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3015 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
3016 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
3017 UNSPEC_UADALP values.
3018 * config/aarch64/iterators.md (ABAL): New int iterator.
3021 (sur): Add mappings for the above.
3022 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
3024 (aarch64_<sur>abal<mode>_4): Likewise.
3025 (aarch64_<sur>adalp<mode>_3): Likewise.
3026 (<sur>sadv16qi): New define_expand.
3028 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
3030 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
3031 (*movdf_internal): Ditto.
3032 (*rcpsf2_sse): Ditto.
3033 (*rsqrtsf2_sse): Ditto.
3034 (*sqrt<mode>2_sse): Ditto.
3036 2018-05-21 Tamar Christina <tamar.christina@arm.com>
3038 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
3040 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
3041 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
3042 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
3043 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
3045 * config/aarch64/arm_neon.h: Likewise.
3046 * config/aarch64/iterators.md (VQ_I): New.
3048 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
3050 * config.gcc: Add arc/t-multilib-linux to tmake_file for
3052 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
3055 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
3057 * config/nds32/constraints.md (S): New constraint.
3058 * config/nds32/nds32.md (call_internal): Use constraint S.
3059 (call_value_internal): Likewise.
3060 (sibcall_internal): Likewise.
3061 (sibcall_value_internal): Likewise.
3063 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
3064 Chung-Ju Wu <jasonwucj@gmail.com>
3066 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
3069 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
3070 Chung-Ju Wu <jasonwucj@gmail.com>
3072 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
3073 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
3074 (nds32_rtx_costs_impl): Simplify.
3075 (nds32_address_cost_impl): Simplify.
3076 (nds32_init_rtx_costs): New function.
3077 (nds32_rtx_costs_speed_prefer): Likewise.
3078 (nds32_rtx_costs_size_prefer): Likewise.
3079 (nds32_address_cost_speed_prefer): Likewise.
3080 (nds32_address_cost_speed_fwprop): Likewise.
3081 (nds32_address_cost_size_prefer): Likewise.
3082 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
3083 * config/nds32/nds32.c (nds32_option_override): Use
3084 nds32_init_rtx_costs function.
3086 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
3088 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
3089 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
3090 (TARGET_PIPELINE_N8): Likewise.
3091 (TARGET_PIPELINE_N10): Likewise.
3092 (TARGET_PIPELINE_N13): Likewise.
3093 (TARGET_PIPELINE_GRAYWOLF): Likewise.
3095 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
3097 * config/nds32/nds32-fpu.md: Update copyright year.
3099 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3101 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
3103 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3105 * config/nds32/nds32.c
3106 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
3107 * config/nds32/nds32.opt (minline-asm-r15): New option.
3109 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3111 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
3113 * config/nds32/nds32.md (abssi2): New pattern.
3115 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
3117 * config/i386/i386.md (rex64namesuffix): New mode attribute.
3118 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
3119 Merge insn pattern from sse_cvtsi2ss<round_name> and
3120 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
3121 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
3122 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
3123 using SWI48 mode iterator.
3124 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
3125 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
3126 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
3127 pattern from sse_cvttss2si<round_saeonly_name>
3128 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
3129 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
3130 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
3131 using SWI48 mode iterator.
3132 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
3133 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
3134 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
3135 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
3136 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
3137 using SWI48 mode iterator.
3138 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
3139 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
3140 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
3141 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
3142 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
3143 SWI48 mode iterator.
3144 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
3145 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
3146 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
3147 pattern from sse_cvttsd2si<round_saeonly_name>
3148 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
3150 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3152 * config/nds32/nds32-md-auxiliary.c
3153 (nds32_valid_smw_lwm_base_p): Refine.
3154 (nds32_output_smw_single_word): Refine.
3155 (nds32_output_smw_double_word): New.
3156 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
3158 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3160 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
3161 (nds32_output_stack_pop): Refine.
3162 (nds32_expand_unaligned_load): Refine.
3163 (nds32_expand_unaligned_store): Refine.
3165 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
3166 Chung-Ju Wu <jasonwucj@gmail.com>
3168 * config/nds32/constants.md: Add TP_REGNUM constant.
3169 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
3170 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
3172 * config/nds32/nds32-doubleword.md: Consider flag_pic.
3173 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
3174 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
3175 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
3176 and PIC code generation.
3177 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
3179 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
3181 * config/nds32/nds32.md: Support TLS and PIC.
3182 * config/nds32/nds32.c: Support TLS and PIC.
3183 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
3184 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
3187 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
3189 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
3190 mode with E_ prefix.
3192 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
3193 Chung-Ju Wu <jasonwucj@gmail.com>
3195 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
3196 * config/nds32/nds32-md-auxiliary.c
3197 (symbolic_reference_mentioned_p): New.
3198 (nds32_legitimize_ict_address): New.
3199 (nds32_expand_ict_move): New.
3200 (nds32_indirect_call_referenced_p): New.
3201 (nds32_symbol_binds_local_p): Delete.
3202 (nds32_long_call_p): Modify.
3203 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
3204 * config/nds32/nds32-protos.h
3205 (symbolic_reference_mentioned_p): Declare.
3206 (nds32_legitimize_ict_address): Declare.
3207 (nds32_expand_ict_move): Declare.
3208 (nds32_indirect_call_referenced_p): Declare.
3209 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
3210 (nds32_relax_group): Use nds32_ict_const_p as condition.
3211 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
3212 (nds32_asm_file_start): Output ict_model directive in asm code.
3213 (nds32_legitimate_address_p): Consider indirect call.
3214 (nds32_print_operand): Consider indirect call.
3215 (nds32_print_operand_address): Consider indirect call.
3216 (nds32_insert_attributes): Handle "indirect_call" attribute.
3217 (TARGET_LEGITIMATE_ADDRESS_P): Define.
3218 (TARGET_LEGITIMATE_CONSTANT_P): Define.
3219 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
3220 (TARGET_DELEGITIMIZE_ADDRESS): Define.
3221 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
3222 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
3223 (TARGET_ICT_MODEL_SMALL): Define.
3224 (TARGET_ICT_MODEL_LARGE): Define.
3225 * config/nds32/nds32.md (movsi): Consider ict model.
3226 (call, call_value): Consider ict model.
3227 (sibcall, sibcall_value): Consider ict model.
3228 * config/nds32/nds32.opt (mict-model): New option.
3229 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
3232 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
3233 Monk Chiang <sh.chiang04@gmail.com>
3234 Jim Wilson <jimw@sifive.com>
3236 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
3237 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
3238 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
3239 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
3240 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
3241 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
3242 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
3243 compute save_libcall_adjustment properly.
3244 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
3245 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
3246 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
3247 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
3248 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
3249 (ABI_SPEC): Handle mabi=ilp32e.
3250 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
3251 (RVE): Add RVE mask.
3252 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
3253 <-march>: Add rv32e as an example.
3255 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
3258 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
3259 (intra_create_variable_infos): Handle C++ constructors.
3261 2018-05-18 Martin Liska <mliska@suse.cz>
3263 * passes.def: Remove a redundant pass.
3265 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
3268 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
3270 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3272 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
3274 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
3276 (armv2, armv2a, armv3, armv3m): Delete architectures.
3277 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
3278 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
3279 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
3281 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
3282 (*mulsidi3adddi): Likewise.
3283 (mulsidi3): Likewise.
3284 (*mulsidi3_nov6): Likewise.
3285 (umulsidi3): Likewise.
3286 (umulsidi3_nov6): Likewise.
3287 (umaddsidi4): Likewise.
3288 (*umulsidi3adddi): Likewise.
3289 (smulsi3_highpart): Likewise.
3290 (*smulsi3_highpart_nov6): Likewise.
3291 (umulsi3_highpart): Likewise.
3292 (*umulsi3_highpart_nov6): Likewise.
3293 * config/arm/arm.h (arm_arch3m): Delete.
3294 * config/arm/arm.c (arm_arch3m): Delete.
3295 (arm_option_override_internal): Update armv3-related comment.
3296 (arm_configure_build_target): Delete use of isa_bit_mode32.
3297 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
3298 (arm_rtx_costs_internal): Delete check of arm_arch3m.
3299 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
3301 (mulusa3): Likewise.
3302 * config/arm/arm-protos.h (arm_arch3m): Delete.
3303 * config/arm/arm-tables.opt: Regenerate.
3304 * config/arm/arm-tune.md: Likewise.
3305 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
3306 deleted architectures.
3308 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3310 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
3311 (armv5t, armv5te): New features.
3312 (ARMv5, ARMv5e): Delete fgroups.
3313 (ARMv5t, ARMv5te): Adjust for above changes.
3315 (armv5, armv5e): Delete arches.
3316 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
3318 (*call_reg_arm): Likewise.
3319 (*call_value_reg_armv5): Likewise.
3320 (*call_value_reg_arm): Likewise.
3321 (*call_symbol): Likewise.
3322 (*call_value_symbol): Likewise.
3323 (*sibcall_insn): Likewise.
3324 (*sibcall_value_insn): Likewise.
3326 (prefetch): Likewise.
3327 (define_split and define_peephole2 dependent on arm_arch5):
3329 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
3331 (TARGET_ARM_QBIT): Likewise.
3332 (TARGET_DSP_MULTIPLY): Likewise.
3333 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
3334 (arm_arch5, arm_arch5e): Delete.
3335 (arm_arch5t, arm_arch5te): Declare.
3336 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
3337 (arm_arch5t): Declare.
3338 (arm_option_reconfigure_globals): Update for the above.
3339 (arm_options_perform_arch_sanity_checks): Update comment, replace
3340 use of arm_arch5 with arm_arch5t.
3341 (use_return_insn): Likewise.
3342 (arm_emit_call_insn): Likewise.
3343 (output_return_instruction): Likewise.
3344 (arm_final_prescan_insn): Likewise.
3345 (arm_coproc_builtin_available): Likewise.
3346 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
3347 arm_arch5e with arm_arch5t and arm_arch5te.
3348 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
3349 (arm_arch5t, arm_arch5te): Declare.
3350 * config/arm/arm-tables.opt: Regenerate.
3351 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
3352 * config/arm/t-multilib: Likewise.
3353 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
3354 instead of arm_arch5.
3355 (*call_reg_thumb1): Likewise.
3356 (*call_value_reg_thumb1_v5): Likewise.
3357 (*call_value_reg_thumb1): Likewise.
3358 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
3360 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
3362 2018-05-18 Martin Liska <mliska@suse.cz>
3364 PR gcov-profile/84846
3365 * doc/gcov.texi: Document -t option of gcov tool.
3367 2018-05-18 Martin Liska <mliska@suse.cz>
3369 PR gcov-profile/84846
3370 * gcov.c (print_usage): Add new -t option.
3371 (process_args): Handle the option.
3372 (generate_results): Use stdout as output when requested by
3375 2018-05-18 Martin Liska <mliska@suse.cz>
3377 PR gcov-profile/84846
3378 * coverage.c (coverage_init): Write PWD to .gcno file.
3379 * doc/gcov.texi: Document how working directory is printed.
3380 * gcov-dump.c (dump_gcov_file): Print PWD.
3381 * gcov.c (output_intermediate_file): Likewise.
3382 (read_graph_file): Read PWD string.
3383 (output_lines): Print PWD.
3385 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3388 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
3389 for retval and return false if all args to phi are zero.
3391 2018-05-18 Richard Biener <rguenther@suse.de>
3393 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
3395 (evrp_dom_walker::before_dom_children): Call it.
3397 2018-05-18 Richard Biener <rguenther@suse.de>
3399 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
3400 results when processing array refs with variable index.
3402 2018-05-18 Toon Moene <toon@moene.org>
3404 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
3405 directly after that of -floop-interchange. Indicate that both
3406 options are enabled by default when specifying -O3.
3408 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3410 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
3411 iterator. Delete separate integer-mode vec_set<mode> expander.
3412 (aarch64_simd_vec_setv2di): Delete.
3413 (vec_setv2di): Delete.
3414 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
3415 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
3416 the "w, r" alternative.
3418 2018-05-18 Martin Liska <mliska@suse.cz>
3420 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
3421 * tree-pass.h (make_pass_lower_switch_O0): New function.
3422 * tree-switch-conversion.c (node_has_low_bound): Remove.
3423 (node_has_high_bound): Likewise.
3424 (node_is_bounded): Likewise.
3425 (class pass_lower_switch): Make it a template type and create
3427 (pass_lower_switch::execute): Add template argument.
3428 (make_pass_lower_switch): New function.
3429 (make_pass_lower_switch_O0): New function.
3430 (do_jump_if_equal): Remove.
3431 (emit_case_nodes): Simplify to just handle all 3 cases and leave
3432 all the hard work to tree optimization passes.
3434 2018-05-18 Martin Liska <mliska@suse.cz>
3436 * dbgcnt.c (limit_low): Renamed from limit.
3437 (limit_high): New variable.
3438 (dbg_cnt_is_enabled): Check for upper limit.
3439 (dbg_cnt): Adjust dumping.
3440 (dbg_cnt_set_limit_by_index): Add new argument for high
3442 (dbg_cnt_set_limit_by_name): Likewise.
3443 (dbg_cnt_process_single_pair): Parse new format.
3444 (dbg_cnt_process_opt): Use strtok.
3445 (dbg_cnt_list_all_counters): Remove 'value' and add
3447 * doc/invoke.texi: Document changes.
3449 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
3451 * doc/sourcebuild.texi (scalar_all_fma): Document.
3452 * tree.def (FMA_EXPR): Delete.
3453 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
3454 * internal-fn.c (ternary_direct): New macro.
3455 (expand_ternary_optab_fn): Likewise.
3456 (direct_ternary_optab_supported_p): Likewise.
3457 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
3458 * builtins.c (fold_builtin_fma): Delete.
3459 (fold_builtin_3): Don't call it.
3460 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
3461 * expr.c (expand_expr_real_2): Likewise.
3462 * fold-const.c (operand_equal_p): Likewise.
3463 (fold_ternary_loc): Likewise.
3464 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
3465 * gimple.c (DEFTREECODE): Likewise.
3466 * gimplify.c (gimplify_expr): Likewise.
3467 * optabs-tree.c (optab_for_tree_code): Likewise.
3468 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3469 * tree-eh.c (operation_could_trap_p): Likewise.
3470 (stmt_could_throw_1_p): Likewise.
3471 * tree-inline.c (estimate_operator_cost): Likewise.
3472 * tree-pretty-print.c (dump_generic_node): Likewise.
3473 (op_code_prio): Likewise.
3474 * tree-ssa-loop-im.c (stmt_cost): Likewise.
3475 * tree-ssa-operands.c (get_expr_operands): Likewise.
3476 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
3477 * fold-const-call.h (fold_fma): Delete.
3478 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
3479 CFN_FNMA and CFN_FNMS.
3481 * genmatch.c (combined_fn): New enum.
3482 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
3483 (commutative_op): New function.
3484 (commutate): Use it. Handle more than 2 operands.
3485 (dt_operand::gen_gimple_expr): Use commutative_op.
3486 (parser::parse_expr): Allow :c to be used with non-binary
3487 operators if the commutative operand is known.
3488 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
3489 CFN_FMS, CFN_FNMA and CFN_FNMS.
3490 (backprop::process_assign_use): Remove FMA_EXPR handling.
3491 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
3492 (gen_hsa_fma): New function.
3493 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
3494 IFN_FNMA and IFN_FNMS.
3495 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
3496 * gimple-fold.h (follow_all_ssa_edges): Declare.
3497 * gimple-fold.c (follow_all_ssa_edges): New function.
3498 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
3499 gimple_build interface and use follow_all_ssa_edges to fold the result.
3500 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
3501 instead of checking for optabs directly.
3502 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
3503 rather than FMA_EXPRs.
3504 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
3505 call to IFN_FMA instead of an FMA_EXPR.
3507 2018-05-17 Jim Wilson <jimw@sifive.com>
3509 * expr.c (do_tablejump): When converting index to Pmode, if we have a
3510 sign extended promoted subreg, and the range does not have the sign bit
3511 set, then do a sign extend.
3513 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
3514 test, check for sign extended subreg and/or constant operands, and
3515 do a sign extend in that case.
3517 2018-05-17 Steve Ellcey <sellcey@cavium.com>
3519 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
3520 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
3522 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
3523 Change logics_shift_reg to logics_shift_imm.
3524 (thunderx2t99_fp_loadpair_basic): Delete.
3525 (thunderx2t99_fp_storepair_basic): Delete.
3526 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
3527 (thunderx2t99_asimd_polynomial): Delete.
3528 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
3529 and neon_fp_mul_d_scalar_q.
3530 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
3531 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
3532 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
3533 (thunderx2t99_asimd_lut): Add missing tbl types.
3534 (thunderx2t99_asimd_ext): Delete.
3535 (thunderx2t99_asimd_load1_1_mult): Delete.
3536 (thunderx2t99_asimd_load1_2_mult): Delete.
3537 (thunderx2t99_asimd_load1_ldp): New.
3538 (thunderx2t99_asimd_load1): New.
3539 (thunderx2t99_asimd_load2): Add missing *load2* types.
3540 (thunderx2t99_asimd_load3): New.
3541 (thunderx2t99_asimd_load4): New.
3542 (thunderx2t99_asimd_store1_1_mult): Delete.
3543 (thunderx2t99_asimd_store1_2_mult): Delete.
3544 (thunderx2t99_asimd_store2_mult): Delete.
3545 (thunderx2t99_asimd_store2_onelane): Delete.
3546 (thunderx2t99_asimd_store_stp): New.
3547 (thunderx2t99_asimd_store1): New.
3548 (thunderx2t99_asimd_store2): New.
3549 (thunderx2t99_asimd_store3): New.
3550 (thunderx2t99_asimd_store4): New.
3552 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
3554 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
3555 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
3557 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
3558 Segher Boessenkool <segher@kernel.crashing.org>
3561 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
3564 2018-05-17 Richard Biener <rguenther@suse.de>
3566 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
3567 for pruning loop and prune defs feeding only already visited PHIs.
3569 2018-05-17 Richard Biener <rguenther@suse.de>
3571 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
3573 2018-05-17 Bin Cheng <bin.cheng@arm.com>
3574 Richard Biener <rguenther@suse.de>
3576 PR tree-optimization/85793
3577 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
3578 for VMAT_ELEMENTWISE.
3580 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
3582 * internal-fn.h (lookup_internal_fn): Declare
3583 * internal-fn.c (lookup_internal_fn): New function.
3584 * gimple.c (gimple_build_call_from_tree): Handle calls to
3586 * gimple-pretty-print.c (dump_gimple_call): Print "." before
3587 internal function names.
3588 * tree-pretty-print.c (dump_generic_node): Likewise.
3589 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
3591 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
3593 * gimple-fold.h (gimple_build): Make the function forms take
3594 combined_fn rather than built_in_function.
3595 (gimple_simplify): Likewise.
3596 * gimple-match-head.c (gimple_simplify): Likewise.
3597 * gimple-fold.c (gimple_build): Likewise.
3598 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
3599 rather than gimple_build_call_internal.
3600 (get_initial_defs_for_reduction): Likewise.
3601 (vect_create_epilog_for_reduction): Likewise.
3602 (vectorizable_live_operation): Likewise.
3604 2018-05-17 Martin Liska <mliska@suse.cz>
3606 * gimple-ssa-sprintf.c (format_directive): Do not use
3607 space in between 'G_' and '('.
3609 2018-05-17 Jakub Jelinek <jakub@redhat.com>
3612 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
3613 even if the mask is not all ones.
3616 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
3618 (ix86_gimple_fold_builtin): Likewise.
3621 * config/i386/i386.c: Include tree-vector-builder.h.
3622 (ix86_vector_shift_count): New function.
3623 (ix86_fold_builtin): Fold shift builtins by scalar count.
3624 (ix86_gimple_fold_builtin): Likewise.
3626 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
3627 _mm512_setzero): New intrinsics.
3629 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
3630 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3632 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
3633 code generation for cases where splatting a value is not useful.
3634 * simplify-rtx.c (simplify_ternary_operation): Simplify
3635 vec_merge across a vec_duplicate and a paradoxical subreg forming
3636 a vector mode to a vec_concat.
3638 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
3640 * config.gcc: Support "goldmont-plus".
3641 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3643 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3644 PROCESSOR_GOLDMONT_PLUS.
3645 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
3646 (processor_target_table): Add "goldmont-plus".
3647 (PTA_GOLDMONT_PLUS): Define.
3648 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
3649 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
3650 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
3651 (fold_builtin_cpu): Add "goldmont-plus".
3652 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
3653 (ix86_option_override_internal): Add "goldmont-plus".
3654 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
3655 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
3656 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
3657 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
3659 2018-05-17 Richard Biener <rguenther@suse.de>
3661 PR tree-optimization/85757
3662 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
3663 remove defs that only feed that PHI from further processing.
3665 2018-05-16 Jim Wilson <jimw@sifive.com>
3667 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
3669 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
3671 2018-05-16 Mark Wielaard <mark@klomp.org>
3673 * dwarf2out.c (count_index_strings): New function.
3674 (output_indirect_strings): Call count_index_strings and generate
3675 header for dwarf_version >= 5.
3677 2018-05-16 Mark Wielaard <mark@klomp.org>
3679 * dwarf2out.c (dwarf_FORM): New function.
3680 (set_indirect_string): Use dwarf_FORM.
3681 (reset_indirect_string): Likewise.
3682 (size_of_die): Likewise.
3683 (value_format): Likewise.
3684 (output_die): Likewise.
3685 (add_skeleton_AT_string): Likewise.
3686 (output_macinfo_op): Likewise.
3687 (index_string): Likewise.
3688 (output_index_string_offset): Likewise.
3689 (output_index_string): Likewise.
3690 (count_index_strings): Likewise.
3692 2018-05-16 Carl Love <cel@us.ibm.com>
3694 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
3695 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
3697 2018-05-16 Martin Jambor <mjambor@suse.cz>
3699 * ipa-prop.c (ipa_free_all_edge_args): Remove.
3700 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
3702 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
3704 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
3705 (fnma<mode>4): Likewise.
3706 (fms<mode>4): Likewise.
3707 (fnms<mode>4): Likewise.
3708 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
3709 (aarch64_fnma<mode>4): Likewise.
3710 (aarch64_fms<mode>4): Likewise.
3711 (aarch64_fnms<mode>4): Likewise.
3712 (aarch64_fnmadd<mode>4): Likewise.
3714 2018-05-16 Jason Merrill <jason@redhat.com>
3716 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
3718 2018-05-16 Richard Biener <rguenther@suse.de>
3720 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
3721 (dump_stmt_cost): Declare.
3722 (add_stmt_cost): Dump cost we add.
3723 (add_stmt_costs): New function.
3724 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
3726 (vect_analyze_stmt): Adjust prototype.
3727 (vectorizable_condition): Likewise.
3728 (vectorizable_live_operation): Likewise.
3729 (vectorizable_reduction): Likewise.
3730 (vectorizable_induction): Likewise.
3731 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
3732 cost vector to pass to vectorizable_ and record afterwards.
3733 (vect_model_reduction_cost): Take cost vector argument and adjust.
3734 (vect_model_induction_cost): Likewise.
3735 (vectorizable_reduction): Likewise.
3736 (vectorizable_induction): Likewise.
3737 (vectorizable_live_operation): Likewise.
3738 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
3739 SLP_TREE_NUMBER_OF_VEC_STMTS.
3740 (vect_analyze_slp_cost_1): Remove.
3741 (vect_analyze_slp_cost): Likewise.
3742 (vect_slp_analyze_node_operations): Take visited args and
3743 a target cost vector. Avoid processing already visited stmt sets.
3744 (vect_slp_analyze_operations): Use a local cost vector to gather
3745 costs and register those of non-discarded instances.
3746 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
3747 (vect_schedule_slp_instance): Remove copying of
3748 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
3750 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
3751 adding cost. Record cost entry location.
3752 (vect_prologue_cost_for_slp_op): Function to compute cost of
3753 a constant or invariant generated for SLP vect in the prologue,
3754 split out from vect_analyze_slp_cost_1.
3755 (vect_model_simple_cost): Make static. Adjust for SLP costing.
3756 (vect_model_promotion_demotion_cost): Likewise.
3757 (vect_model_store_cost): Likewise, make static.
3758 (vect_model_load_cost): Likewise.
3759 (vectorizable_bswap): Add cost vector arg and adjust.
3760 (vectorizable_call): Likewise.
3761 (vectorizable_simd_clone_call): Likewise.
3762 (vectorizable_conversion): Likewise.
3763 (vectorizable_assignment): Likewise.
3764 (vectorizable_shift): Likewise.
3765 (vectorizable_operation): Likewise.
3766 (vectorizable_store): Likewise.
3767 (vectorizable_load): Likewise.
3768 (vectorizable_condition): Likewise.
3769 (vectorizable_comparison): Likewise.
3770 (can_vectorize_live_stmts): Likewise.
3771 (vect_analyze_stmt): Likewise.
3772 (vect_transform_stmt): Adjust calls to vectorizable_*.
3773 * tree-vectorizer.c: Include gimple-pretty-print.h.
3774 (dump_stmt_cost): New function.
3776 2018-05-16 Richard Biener <rguenther@suse.de>
3778 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
3779 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
3780 * tree-ssa-dse.c: Include tree-ssa-loop.h.
3781 (check_name): New callback.
3782 (dse_classify_store): Track cycles via a visited bitmap of PHI
3783 defs and simplify handling of in-loop and across loop dead stores
3784 and properly fail for loop-variant refs. Handle byte-tracking with
3785 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
3788 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
3790 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
3791 (vect_get_mask_type_for_stmt): Likewise.
3792 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
3794 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
3795 to determine the statement's vector type and the vector type that
3796 should be used for calculating nunits. Deal with cases in which
3797 the type has to be deferred.
3798 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
3799 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
3800 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
3801 (vect_determine_vf_for_stmt): New functions, split out from...
3802 (vect_determine_vectorization_factor): ...here.
3803 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
3804 (vect_get_mask_type_for_stmt): New functions, split out from
3805 vect_determine_vectorization_factor.
3807 2018-05-16 Richard Biener <rguenther@suse.de>
3809 * tree-cfg.c (verify_gimple_assign_ternary): Properly
3810 verify the [VEC_]COND_EXPR embedded comparison.
3812 2018-05-15 Martin Sebor <msebor@redhat.com>
3814 PR tree-optimization/85753
3815 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
3816 RECORD_TYPE in addition to ARRAY_TYPE.
3818 2018-05-15 Martin Sebor <msebor@redhat.com>
3821 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
3823 2018-05-15 Richard Biener <rguenther@suse.de>
3825 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
3826 add by_clobber_p one. Change algorithm to collect all defs
3827 representing uses we need to walk and try reducing them to
3828 a single one before failing.
3829 (dse_dom_walker::dse_optimize_stmt): Adjust.
3831 2018-05-13 Mark Wielaard <mark@klomp.org>
3833 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
3834 (size_of_loc_descr): Likewise.
3835 (output_loc_operands): Likewise.
3836 (output_loc_operands_raw): Likewise.
3837 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
3838 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
3839 (hash_loc_operands): Likewise.
3840 (compare_loc_operands): Likewise.
3842 2018-05-14 Mark Wielaard <mark@klomp.org>
3844 * dwarf2out.c (count_index_addrs): New function.
3845 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
3847 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3849 PR tree-optimization/83648
3850 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
3851 return value as malloc candidate.
3853 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3856 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
3857 param as true in call to suggest_attribute.
3859 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
3861 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
3862 -mreadonly-in-sdata.
3864 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3866 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
3868 (aarch64_crypto_aesd_fused): Likewise.
3870 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
3872 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
3873 (movsi_aarch64): Likewise.
3874 (load_pairsi): Likewise.
3875 (load_pairdi): Likewise.
3876 (store_pairsi): Likewise.
3877 (store_pairdi): Likewise.
3878 (load_pairsf): Likewise.
3879 (load_pairdf): Likewise.
3880 (store_pairsf): Likewise.
3881 (store_pairdf): Likewise.
3882 (zero_extend): Likewise.
3883 (trunc): Swap alternatives.
3884 (fcvt_target): Add '?' to prefer w over r.
3886 2018-05-14 Jakub Jelinek <jakub@redhat.com>
3889 * config/i386/i386.md: Disallow non-commutative arithmetics in
3890 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
3891 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
3892 in the peephole2 before it.
3894 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
3896 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
3897 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
3898 (ix86_handle_option): Handle -mcldemote.
3899 * config.gcc: New header.
3900 * config/i386/cldemoteintrin.h: New file.
3901 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
3902 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3904 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3905 OPTION_MASK_ISA_CLDEMOTE.
3906 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
3907 (ix86_valid_target_attribute_inner_p): Ditto.
3908 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
3909 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
3910 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
3911 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
3912 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
3914 * config/i386/i386.opt: Add -mcldemote.
3915 * config/i386/x86intrin.h: New header.
3916 * doc/invoke.texi: Add -mcldemote.
3918 2018-05-14 Richard Biener <rguenther@suse.de>
3920 * doc/match-and-simplify.texi: Adjust :s documentation.
3922 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
3924 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
3925 intended memcpy size.
3926 (REORDER_45): Likewise.
3928 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
3930 * sort.cc: New file.
3931 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
3932 * vec.c (qsort_chk): Use gcc_qsort.
3933 * Makefile.in (OBJS-libcommon): Add sort.o.
3934 (build/sort.o): New target. Use it...
3935 (BUILD_RTL): ... here, and...
3936 (build/gencfn-macros): ... here, and...
3937 (build/genmatch): ... here.
3939 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
3940 Chung-Ju Wu <jasonwucj@gmail.com>
3942 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
3943 * config/nds32/nds32-graywolf.md: New file.
3944 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
3945 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
3947 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
3948 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
3949 * config/nds32/nds32.md (pipeline_model): Add graywolf.
3950 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
3951 * config/nds32/pipelines.md: Include n15 settings.
3953 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
3954 Chung-Ju Wu <jasonwucj@gmail.com>
3956 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
3957 * config/nds32/nds32-n13.md: New file.
3958 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
3959 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
3961 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
3962 * config/nds32/nds32.md (pipeline_model): Add n13.
3963 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
3964 * config/nds32/pipelines.md: Include n13 settings.
3966 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
3967 Chung-Ju Wu <jasonwucj@gmail.com>
3969 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
3970 * config/nds32/nds32-n10.md: New file.
3971 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
3972 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
3974 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
3975 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
3976 * config/nds32/nds32.md (pipeline_model): Add n10.
3977 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
3978 * config/nds32/pipelines.md: Include n10 settings.
3980 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
3981 Kito Cheng <kito.cheng@gmail.com>
3982 Chung-Ju Wu <jasonwucj@gmail.com>
3984 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
3985 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
3986 Add enum values for DSP extension instructions.
3987 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
3989 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
3990 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
3992 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
3993 * config/nds32/nds32-dspext.md: New file for DSP implementation.
3994 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
3995 * config/nds32/nds32-intrinsic.md: Likewise.
3996 * config/nds32/nds32_intrinsic.h: Likewise.
3997 * config/nds32/nds32-md-auxiliary.c: Likewise.
3998 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
3999 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
4000 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
4001 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
4002 * config/nds32/nds32-protos.h: New declarations for DSP extension.
4003 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
4004 TYPE_DMAC in switch statement.
4005 * config/nds32/nds32.c: New checking and implementation for DSP
4006 extension instructions.
4007 * config/nds32/nds32.h: Likewise.
4008 * config/nds32/nds32.md: Likewise.
4009 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
4010 * config/nds32/predicates.md: Implement new predicates for DSP
4013 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4015 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
4016 Reformat alternatives and attributes so it is easier to identify
4017 which constraints/attributes go with which instruction.
4018 (mov<mode>_hardfloat32, FMOVE64): Likewise.
4019 (mov<mode>_softfloat32, FMOVE64): Likewise.
4020 (mov<mode>_hardfloat64, FMOVE64): Likewise.
4021 (mov<mode>_softfloat64, FMOVE64): Likewise.
4023 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
4025 * doc/extend.texi (PowerPC Built-in Functions): Rename this
4027 (Basic PowerPC Built-in Functions): The new name of the
4028 subsection previously known as "PowerPC Built-in Functions".
4029 (Basic PowerPC Built-in Functions Available on all Configurations):
4031 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
4032 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
4033 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
4034 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
4036 2018-05-11 Martin Jambor <mjambor@suse.cz>
4039 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
4042 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
4045 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
4047 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
4049 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
4050 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
4051 (ix86_handle_option): Handle -mwaitpkg.
4052 * config.gcc: New header.
4053 * config/i386/cpuid.h (bit_WAITPKG): New bit.
4054 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
4055 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
4057 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4058 OPTION_MASK_ISA_WAITPKG.
4059 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
4060 (ix86_option_override_internal): Add PTA_WAITPKG.
4061 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
4062 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
4063 IX86_BUILTIN_TPAUSE.
4064 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
4065 __builtin_ia32_umwait and __builtin_ia32_tpause.
4066 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
4067 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
4068 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
4069 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
4070 UNSPECV_TPAUSE): New.
4071 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
4072 * config/i386/i386.opt: Add -mwaitpkg.
4073 * config/i386/waitpkgintrin.h: New file.
4074 * config/i386/x86intrin.h: New header.
4075 * doc/invoke.texi: Add -mwaitpkg.
4077 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
4080 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
4082 (cortex-m0): Use armv6s-m isa.
4083 (cortex-m0plus): Likewise.
4084 (cortex-m1): Likewise.
4085 (cortex-m0.small-multiply): Likewise.
4086 (cortex-m0plus.small-multiply): Likewise.
4087 (cortex-m1.small-multiply): Likewise.
4089 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
4090 Jakub Jelinek <jakub@redhat.com>
4092 PR tree-optimization/85692
4093 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
4094 source permute as well.
4096 2018-05-11 Martin Liska <mliska@suse.cz>
4099 * doc/extend.texi: Document LLVM style format for no_sanitize
4102 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4104 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
4105 mode_supports_vsx_dform_quad to mode_supports_dq_form.
4106 (mode_supports_vsx_dform_quad): Likewise.
4107 (mode_supports_vmx_dform): Move these functions to be next to the
4108 other mode_supports functions.
4109 (mode_supports_dq_form): Likewise.
4110 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
4111 mode_supports_dq_form.
4112 (reg_offset_addressing_ok_p): Likewise.
4113 (offsettable_ok_by_alignment): Likewise.
4114 (rs6000_legitimate_offset_address_p): Likewise.
4115 (legitimate_lo_sum_address_p): Likewise.
4116 (rs6000_legitimize_address): Likewise.
4117 (rs6000_legitimize_reload_address): Likewise.
4118 (rs6000_secondary_reload_inner): Likewise.
4119 (rs6000_preferred_reload_class): Likewise.
4120 (rs6000_output_move_128bit): Likewise.
4122 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
4124 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
4125 Generate SImode target register for null target.
4126 <case IX86_BUILTIN_XGETBV>: Ditto.
4127 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
4128 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
4130 2018-05-10 Carl Love <cel@us.ibm.com>
4132 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
4133 dcbtt and dcbtstt if operands[2] is 0.
4135 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
4138 * config/i386/sse.md (usadv64qi): New expander.
4140 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
4142 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
4143 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
4144 -maltivec=be support.
4145 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
4146 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
4147 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4148 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
4149 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
4150 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
4151 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
4152 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
4153 altivec_vsumsws): Adjust.
4154 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
4155 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
4156 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
4158 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
4159 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
4160 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
4161 (altivec_lve<VI_char>x): Delete expand.
4162 (*altivec_lve<VI_char>x_internal): Rename to...
4163 (altivec_lve<VI_char>x): ... this.
4164 (altivec_lvxl_<mode>): Delete expand.
4165 (*altivec_lvxl_<mode>_internal): Rename to ...
4166 (altivec_lvxl_<mode>): ... this.
4167 (altivec_stvxl_<mode>): Delete expand.
4168 (*altivec_stvxl_<mode>_internal): Rename to ...
4169 (altivec_stvxl_<mode>): ... this.
4170 (altivec_stve<VI_char>x): Delete expand.
4171 (*altivec_stve<VI_char>x_internal): Rename to ...
4172 (altivec_stve<VI_char>x): ... this.
4173 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
4174 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
4175 reduc_plus_scal_<mode>): Adjust.
4176 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
4178 (rs6000_cpu_cpp_builtins): Adjust.
4179 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
4180 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
4181 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
4182 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
4183 -maltivec=be support.
4184 (rs6000_split_vec_extract_var): Adjust.
4185 (rs6000_split_v4si_init): Adjust.
4186 (swap_selector_for_mode): Delete.
4187 (altivec_expand_lvx_be, altivec_expand_stvx_be,
4188 altivec_expand_stvex_be): Delete.
4189 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
4190 -maltivec=be support.
4191 (rs6000_gimple_fold_builtin): Ditto.
4192 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
4194 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
4195 (TARGET_DIRECT_MOVE_64BIT): Adjust.
4196 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
4197 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
4198 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
4199 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
4200 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
4201 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
4202 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
4203 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
4204 anonymous split): Adjust.
4205 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
4206 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
4208 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
4210 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
4211 when --with-gxx-include-dir is also specified.
4212 * configure: Regenerate.
4214 2018-05-09 Jim Wilson <jimw@sifive.com>
4217 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
4218 * config/riscv/t-withmultilib: New.
4219 * config/riscv/withmultilib.h: New.
4220 * doc/install.texi: Document RISC-V --with-multilib-list support.
4222 2018-05-09 Richard Biener <rguenther@suse.de>
4224 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
4226 (vect_bb_vectorization_profitable_p): Adjust. Compute
4227 actual scalar cost using the cost vector and the add_stmt_cost
4230 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
4232 PR rtl-optimization/85645
4233 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
4234 in the REG_CFA_REGISTER note for LR, don't leave it empty.
4236 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
4238 PR rtl-optimization/85645
4239 * shrink-wrap.c (spread_components): Return a boolean saying if
4240 anything was changed.
4241 (try_shrink_wrapping_separate): Iterate spread_components until
4242 nothing changes anymore.
4244 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
4246 PR rtl-optimization/85645
4247 * regrename.c (build_def_use): Also kill the chains that include the
4248 destination of a REG_CFA_REGISTER note.
4250 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
4252 PR rtl-optimization/85645
4253 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
4254 insn that has a REG_CFA_REGISTER note.
4256 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
4258 * cfgexpand.c (expand_clobber): New function.
4259 (expand_gimple_stmt_1): Use it.
4260 * tree-vect-stmts.c (vect_clobber_variable): New function,
4262 (vectorizable_simd_clone_call): ...here.
4263 (vectorizable_store): Emit a clobber either side of an
4264 IFN_STORE_LANES sequence.
4265 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
4267 2018-05-09 Tom de Vries <tom@codesourcery.com>
4270 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
4271 (define_insn "trap_if_false"): Add exit after trap.
4273 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
4275 PR rtl-optimization/85638
4276 * bb-reorder.c: Include common/common-target.h.
4277 (create_forwarder_block): New function extracted from...
4278 (fix_up_crossing_landing_pad): ...here. Rename into...
4279 (dw2_fix_up_crossing_landing_pad): ...this.
4280 (sjlj_fix_up_crossing_landing_pad): New function.
4281 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
4282 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
4283 from both partitions and exit the loop after one iteration.
4285 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
4288 * doc/extend.texi (PowerPC Built-in Functions): Rename this
4290 (Basic PowerPC Built-in Functions): The new name of the
4291 subsection previously known as "PowerPC Built-in Functions".
4292 (Basic PowerPC Built-in Functions Available on all Configurations):
4294 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
4296 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
4298 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
4300 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
4303 2018-05-08 Jim Wilson <jimw@sifive.com>
4305 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
4306 (LD_EMUL_SUFFIX): New.
4307 (LINK_SPEC): Use it.
4309 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
4311 * doc/extend.texi (PowerPC Built-in Functions): Rename this
4313 (Basic PowerPC Built-in Functions): The new name of the
4314 subsection previously known as "PowerPC Built-in Functions".
4315 (Basic PowerPC Built-in Functions Available on all Configurations):
4317 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
4319 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
4321 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
4323 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
4326 2018-05-08 Jakub Jelinek <jakub@redhat.com>
4329 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
4330 after cmpelim optimization.
4332 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
4334 * config.gcc: Support "goldmont".
4335 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
4336 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4338 * config/i386/i386.c (m_GOLDMONT): Define.
4339 (processor_target_table): Add "goldmont".
4340 (PTA_GOLDMONT): Define.
4341 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
4342 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
4343 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
4344 (fold_builtin_cpu): Add "goldmont".
4345 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
4346 (ix86_option_override_internal): Add "goldmont".
4347 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
4348 (processor_type): Add PROCESSOR_GOLDMONT.
4349 * config/i386/i386.md: Add CPU "glm".
4350 * config/i386/glm.md: New file.
4351 * config/i386/x86-tune.def: Add m_GOLDMONT.
4352 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
4354 2018-05-08 Jakub Jelinek <jakub@redhat.com>
4357 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
4359 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
4360 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
4361 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
4364 * config/i386/i386.c (ix86_fold_builtin): Handle
4365 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
4368 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
4369 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
4371 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
4374 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
4375 (check_arch): Likewise.
4376 (check_fpu): Return the result rather than printing it.
4377 (end arch): Fix operator precedence.
4378 (end cpu): Likewise.
4379 (END): Print the result from check_fpu.
4381 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
4382 Alan Hayward <alan.hayward@arm.com>
4383 David Sherwood <david.sherwood@arm.com>
4385 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
4386 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
4387 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
4388 (*fcmuo<mode>_and): New patterns.
4390 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
4392 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
4393 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
4394 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
4395 (cmp_op, sve_imm_con): New code attributes.
4396 (SVE_COND_INT_CMP, imm_con): Delete.
4397 (cmp_op): Remove above unspecs from int attribute.
4398 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
4400 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
4401 comparison-specific unspecs.
4402 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
4403 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
4404 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
4405 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
4406 (*vec_fcm<cmp_op><mode>): Rename to...
4407 (*fcm<cmp_op><mode>): ...this and adjust likewise.
4408 (*vec_fcmuo<mode>): Rename to...
4409 (*fcmuo<mode>): ...this and adjust likewise.
4410 (*pred_fcm<cmp_op><mode>): New pattern.
4411 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
4412 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
4414 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
4416 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
4417 (aarch64_emit_sve_predicated_cond): New function.
4418 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
4419 (aarch64_emit_unspec_cond_or): Replace with...
4420 (aarch64_emit_sve_or_conds): ...this new function. Use
4421 aarch64_emit_sve_ptrue_op for the individual comparisons and
4422 aarch64_emit_binop to OR them together.
4423 (aarch64_emit_inverted_unspec_cond): Replace with...
4424 (aarch64_emit_sve_inverted_cond): ...this new function. Use
4425 aarch64_emit_sve_ptrue_op for the comparison and
4426 aarch64_emit_unop to invert the result.
4427 (aarch64_expand_sve_vec_cmp_float): Update after the above
4428 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
4430 2018-05-07 Nathan Sidwell <nathan@acm.org>
4432 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
4433 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
4434 (Backwards Compatibility): Likewise.
4436 2018-05-07 Luis Machado <luis.machado@linaro.org>
4440 2018-05-07 Luis Machado <luis.machado@linaro.org>
4442 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4443 <prefetch_dynamic_strides>: New const bool field.
4444 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4445 prefetch_dynamic_strides.
4446 (exynosm1_prefetch_tune): Likewise.
4447 (thunderxt88_prefetch_tune): Likewise.
4448 (thunderx_prefetch_tune): Likewise.
4449 (thunderx2t99_prefetch_tune): Likewise.
4450 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
4452 (aarch64_override_options_internal): Update to set
4453 PARAM_PREFETCH_DYNAMIC_STRIDES.
4454 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
4455 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
4456 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
4457 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
4458 prefetch-dynamic-strides setting.
4460 2018-05-07 Luis Machado <luis.machado@linaro.org>
4462 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4463 <minimum_stride>: New const int field.
4464 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4465 minimum_stride field.
4466 (exynosm1_prefetch_tune): Likewise.
4467 (thunderxt88_prefetch_tune): Likewise.
4468 (thunderx_prefetch_tune): Likewise.
4469 (thunderx2t99_prefetch_tune): Likewise.
4470 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
4471 (aarch64_override_options_internal): Update to set
4472 PARAM_PREFETCH_MINIMUM_STRIDE.
4473 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
4474 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
4475 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
4476 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
4477 stride is constant and is below the minimum stride threshold.
4479 2018-05-07 Luis Machado <luis.machado@linaro.org>
4481 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
4484 2018-05-07 Luis Machado <luis.machado@linaro.org>
4486 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4487 <prefetch_dynamic_strides>: New const bool field.
4488 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4489 prefetch_dynamic_strides.
4490 (exynosm1_prefetch_tune): Likewise.
4491 (thunderxt88_prefetch_tune): Likewise.
4492 (thunderx_prefetch_tune): Likewise.
4493 (thunderx2t99_prefetch_tune): Likewise.
4494 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
4496 (aarch64_override_options_internal): Update to set
4497 PARAM_PREFETCH_DYNAMIC_STRIDES.
4498 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
4499 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
4500 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
4501 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
4502 prefetch-dynamic-strides setting.
4504 2018-05-07 Luis Machado <luis.machado@linaro.org>
4506 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4507 <minimum_stride>: New const int field.
4508 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4509 minimum_stride field.
4510 (exynosm1_prefetch_tune): Likewise.
4511 (thunderxt88_prefetch_tune): Likewise.
4512 (thunderx_prefetch_tune): Likewise.
4513 (thunderx2t99_prefetch_tune): Likewise.
4514 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
4515 (aarch64_override_options_internal): Update to set
4516 PARAM_PREFETCH_MINIMUM_STRIDE.
4517 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
4518 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
4519 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
4520 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
4521 stride is constant and is below the minimum stride threshold.
4523 2018-05-06 Jakub Jelinek <jakub@redhat.com>
4526 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
4527 the type is addressable. Don't force op into register if it has
4530 2018-05-05 Roland McGrath <mcgrathr@google.com>
4533 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
4534 any section for which we don't know a specific type it should have,
4535 regardless of name. Previously this was done only for the exact
4536 names ".init_array", ".fini_array", and ".preinit_array".
4537 (default_elf_asm_named_section): Add comment about
4538 relationship with default_section_type_flags and SECTION_NOTYPE.
4539 (get_section): Don't consider it a type conflict if one side has
4540 SECTION_NOTYPE and the other doesn't, as long as neither has the
4541 SECTION_BSS et al used in the default_section_type_flags logic.
4543 2018-05-05 Tom de Vries <tom@codesourcery.com>
4546 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
4547 (workaround_barsyncs): New function.
4548 (nvptx_reorg): Use workaround_barsyncs.
4549 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
4550 (define_expand "nvptx_membar_cta"): New define_expand.
4551 (define_insn "*nvptx_membar_cta"): New insn.
4553 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
4555 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
4556 To improve optimization opportunities.
4557 * builtin-types.def: The new needed builtin types for the above.
4559 2018-05-04 Richard Biener <rguenther@suse.de>
4561 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
4562 * gimple-ssa-store-merging.c
4563 (imm_store_chain_info::output_merged_store): Remove redundant create,
4564 release split_store vector contents on failure.
4565 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
4566 scalar stmt vector on cache hit.
4568 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
4570 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
4572 * config.gcc (powerpc-xilinx-eabi*): Remove.
4573 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
4575 (fusion_addis_mem_combo_load): Ditto.
4576 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
4578 (rs6000_cpu_cpp_builtins): Ditto.
4579 * config/rs6000/rs6000-linux.c
4580 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
4581 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
4582 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
4584 (rs6000_setup_reg_addr_masks): Ditto.
4585 (rs6000_init_hard_regno_mode_ok): Ditto.
4586 (rs6000_option_override_internal): Ditto.
4587 (legitimate_lo_sum_address_p): Ditto.
4588 (rs6000_legitimize_address): Ditto.
4589 (rs6000_legitimize_reload_address): Ditto.
4590 (rs6000_legitimate_address_p): Ditto.
4591 (abi_v4_pass_in_fpr): Ditto.
4592 (setup_incoming_varargs): Ditto.
4593 (rs6000_gimplify_va_arg): Ditto.
4594 (rs6000_split_multireg_move): Ditto.
4595 (rs6000_savres_strategy): Ditto.
4596 (rs6000_emit_prologue_components): Ditto.
4597 (rs6000_emit_epilogue_components): Ditto.
4598 (rs6000_emit_prologue): Ditto.
4599 (rs6000_emit_epilogue): Ditto.
4600 (rs6000_elf_file_end): Ditto.
4601 (rs6000_function_value): Ditto.
4602 (rs6000_libcall_value): Ditto.
4603 * config/rs6000/rs6000.h: Ditto.
4604 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
4605 (TARGET_MINMAX): ... this. New.
4606 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
4607 * config/rs6000/rs6000.md: Remove Xilinx FP support.
4608 (*movsi_internal1_single): Delete.
4609 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
4610 mfpu=, mxilinx-fpu): Delete.
4611 * config/rs6000/singlefp.h: Delete.
4612 * config/rs6000/sysv4.h: Remove Xilinx FP support.
4613 * config/rs6000/t-rs6000: Ditto.
4614 * config/rs6000/t-xilinx: Delete.
4615 * gcc/config/rs6000/titan.md: Adjust for fp_type removal.
4616 * gcc/config/rs6000/vsx.md: Remove Xilinx FP support.
4617 (VStype_simple): Delete.
4618 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
4619 * config/rs6000/xfpu.h: Delete.
4620 * config/rs6000/xfpu.md: Delete.
4621 * config/rs6000/xilinx.h: Delete.
4622 * config/rs6000/xilinx.opt: Delete.
4623 * gcc/doc/invoke.texi (RS/6000 and PowerPC Options): Remove
4624 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
4626 2018-05-04 Tom de Vries <tom@codesourcery.com>
4629 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
4632 2018-05-04 Richard Biener <rguenther@suse.de>
4635 * tree-complex.c (update_complex_assignment): We are always in SSA form.
4636 (expand_complex_div_wide): Likewise.
4637 (expand_complex_operations_1): Likewise.
4638 (expand_complex_libcall): Preserve EH info of the original stmt.
4639 (tree_lower_complex): Handle removed blocks.
4640 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
4641 on complex multiplication and division libcall builtins.
4643 2018-05-04 Richard Biener <rguenther@suse.de>
4646 * fold-const.c (negate_expr_p): Restrict negation of operand
4647 zero of a division to when we know that can happen without
4649 (fold_negate_expr_1): Likewise.
4651 2018-05-04 Jakub Jelinek <jakub@redhat.com>
4654 * real.h (real_nextafter): Declare.
4655 * real.c (real_nextafter): New function.
4656 * fold-const-call.c (fold_const_nextafter): New function.
4657 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
4658 CASE_CFN_NEXTTOWARD.
4659 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
4660 even when arg1_mode is different from arg0_mode.
4662 2018-05-03 Nathan Sidwell <nathan@acm.org>
4664 * doc/extend.texi (Deprecated Features): Remove
4666 (Backwards Compatibility): Likewise.
4667 * doc/invoke.texi (C++ Language Options): Likewise.
4668 (C++ Dialect Options): Likewise.
4670 2018-05-03 Jakub Jelinek <jakub@redhat.com>
4673 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
4674 _mm512_mask_mullox_epi64): New intrinsics.
4676 2018-05-03 Tom de Vries <tom@codesourcery.com>
4679 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
4680 dump files): Add offload-tree.
4682 2018-05-03 Richard Biener <rguenther@suse.de>
4684 PR tree-optimization/85615
4685 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
4686 to loops not nested in BBs loop father to avoid creating multi-entry
4689 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4691 PR tree-optimization/70291
4692 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
4693 arguments. Change return type to tree. Emit libcall as a new
4694 statement rather than replacing existing one when inplace_p is true.
4695 (expand_complex_multiplication_components): New function.
4696 (expand_complex_multiplication): Expand floating-point complex
4697 multiplication using the above.
4698 (expand_complex_division): Rename inner_type parameter to type.
4699 Update expand_complex_libcall call-site.
4700 (expand_complex_operations_1): Update expand_complex_multiplication
4701 and expand_complex_division call-sites.
4703 2018-05-02 Jakub Jelinek <jakub@redhat.com>
4706 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
4707 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
4708 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
4709 the highest significant bit of the shift count mask is clear. In
4710 check whether and[sq]i3 is needed verify that all significant bits
4711 of the shift count other than the highest are set.
4713 2018-05-02 Tom de Vries <tom@codesourcery.com>
4716 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
4717 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
4718 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
4719 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
4720 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
4721 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
4722 __builtin_goacc_parlevel_size.
4724 2018-05-02 Richard Biener <rguenther@suse.de>
4726 PR tree-optimization/85597
4727 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
4728 do not use split vect_get_vec_defs call but call vect_get_slp_defs
4731 2018-05-02 Tom de Vries <tom@codesourcery.com>
4734 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
4735 dump files): Add ltrans-tree.
4737 2018-05-02 Tom de Vries <tom@codesourcery.com>
4740 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
4741 dump files): Add wpa-ipa.
4743 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
4745 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
4746 powerpc*-*-linux*paired* target.
4747 * config/rs6000/750cl.h: Delete.
4748 * config/rs6000/paired.h: Delete.
4749 * config/rs6000/paired.md: Delete.
4750 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
4752 * config/rs6000/rs6000-builtin.def: Remove paired float support.
4753 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4754 comment. Remove paired float support.
4755 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
4756 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
4758 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
4759 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
4761 * config/rs6000/rs6000.c: Remove paired float support.
4762 (paired_expand_vector_init, paired_expand_vector_move,
4763 paired_emit_vector_compare, paired_emit_vector_cond_expr,
4764 (paired_expand_lv_builtin, paired_expand_stv_builtin,
4765 paired_expand_builtin, paired_expand_predicate_builtin,
4766 paired_init_builtins): Delete.
4767 * config/rs6000/rs6000.h: Remove paired float support.
4768 * config/rs6000/rs6000.md: Remove paired float support.
4769 (move_from_CR_ov_bit): Delete.
4770 * config/rs6000/rs6000.opt (mpaired): Delete.
4771 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
4772 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
4774 2018-05-02 Richard Biener <rguenther@suse.de>
4777 * gimplify.c (gimplify_save_expr): When in SSA form allow
4778 SAVE_EXPRs to compute to SSA vars.
4780 2018-05-02 Jakub Jelinek <jakub@redhat.com>
4783 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
4784 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
4785 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
4786 clobber operands[2], instead use a new pseudo. Formatting fixes.
4788 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
4790 PR tree-optimization/85586
4791 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
4792 exit early for statements in the same group if the accesses are
4795 2018-05-02 Tom de Vries <tom@codesourcery.com>
4798 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
4801 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
4803 PR tree-optimization/85143
4804 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
4806 2018-05-01 Tom de Vries <tom@codesourcery.com>
4809 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
4810 not found" error message.
4812 2018-05-01 Tom de Vries <tom@codesourcery.com>
4815 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
4816 * vec.c (test_ordered_remove_if): New function.
4817 (vec_c_tests): Call test_ordered_remove_if.
4818 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
4819 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
4820 * tree-vect-patterns.c (vect_pattern_recog_1): Use
4821 VEC_ORDERED_REMOVE_IF.
4823 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4825 PR tree-optimization/82665
4826 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
4827 pointer subtraction where arguments come from a memchr call.
4829 2018-05-01 Jakub Jelinek <jakub@redhat.com>
4831 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
4832 --push-state --as-needed and --pop-state instead of --as-needed and
4833 --no-as-needed if ld supports it.
4834 * configure: Regenerated.
4837 * doc/install.texi2html: Replace _002d with - and _002a with * in
4838 generated html files using sed.
4840 2018-04-30 David Malcolm <dmalcolm@redhat.com>
4843 * gcc-rich-location.c (blank_line_before_p): New function.
4844 (use_new_line): New function.
4845 (gcc_rich_location::add_fixit_insert_formatted): New function.
4846 * gcc-rich-location.h
4847 (gcc_rich_location::add_fixit_insert_formatted): New function.
4849 2018-04-30 David Malcolm <dmalcolm@redhat.com>
4851 * selftest.c (assert_streq): Rename "expected" and "actual" to
4852 "val1" and "val2". Extend NULL-handling to cover both inputs
4853 symmetrically, while still requiring both to be non-NULL for a pass.
4854 * selftest.h (assert_streq): Rename "expected" and "actual" to
4856 (ASSERT_EQ): Likewise.
4857 (ASSERT_EQ_AT): Likewise.
4858 (ASSERT_KNOWN_EQ): Likewise.
4859 (ASSERT_KNOWN_EQ_AT): Likewise.
4860 (ASSERT_NE): Likewise.
4861 (ASSERT_MAYBE_NE): Likewise.
4862 (ASSERT_MAYBE_NE_AT): Likewise.
4863 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
4864 the assertion to pass.
4865 (ASSERT_STREQ_AT): Likewise.
4867 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
4869 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
4870 interaction with -pie.
4872 2018-04-30 David Malcolm <dmalcolm@redhat.com>
4874 * selftest.h: Fix alphabetization of per-source-file selftest
4877 2018-04-30 Jason Merrill <jason@redhat.com>
4879 PR c++/61982 - dead stores to destroyed objects.
4880 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
4883 2018-04-30 Jason Merrill <jason@redhat.com>
4885 * tree.c (build_clobber): New.
4886 * tree.h: Declare it.
4887 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
4889 2018-04-30 David Malcolm <dmalcolm@redhat.com>
4891 * diagnostic-show-locus.c (layout::layout): Update for
4892 location_get_source_line returning a char_span.
4893 (struct char_span): Move to input.h.
4894 (struct correction): Update for fields in char_span becoming
4896 (struct source_line): Update for location_get_source_line
4897 returning a char_span.
4898 (layout::print_line): Likewise.
4899 * edit-context.c (edited_file::print_content): Likewise.
4900 (edited_file::print_diff_hunk): Likewise.
4901 (edited_file::print_run_of_changed_lines): Likewise.
4902 (edited_file::get_num_lines): Likewise.
4903 (edited_line::edited_line): Likewise.
4904 * final.c (asm_show_source): Likewise.
4905 * input.c (location_get_source_line): Convert return type
4906 from const char * to char_span, losing the final "line_len"
4908 (dump_location_info): Update for the above.
4909 (get_substring_ranges_for_loc): Likewise. Use a char_span
4910 when handling the literal within the line.
4911 (test_reading_source_line): Update for location_get_source_line
4912 returning a char_span.
4913 * input.h (class char_span): Move here from
4914 diagnostic-show-locus.c, converting from a struct to a class.
4915 Make data members private.
4916 (char_span::operator bool): New.
4917 (char_span::length): New.
4918 (char_span::get_buffer): New.
4919 (char_span::operator[]): New.
4920 (char_span::subspan): Make const.
4921 (char_span::xstrdup): New.
4922 (location_get_source_line): Convert return type from const char *
4923 to char_span, losing the final "line_size" param.
4925 2018-04-30 Jan Hubicka <jh@suse.cz>
4927 * lto-wrapper.c (ltrans_priorities): New static var.
4928 (cmp_priority): New.
4929 (run_gcc): Read priorities and if doing parallel build order
4930 the Makefile by them.
4932 2018-04-30 David Malcolm <dmalcolm@redhat.com>
4934 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
4936 2018-04-30 Richard Biener <rguenther@suse.de>
4938 * tree-cfg.c (verify_address): Remove base argument, add
4939 flag whether to check TREE_ADDRESSABLE and do that.
4940 (verify_expr): Remove.
4941 (verify_types_in_gimple_reference): Add pieces from verify_expr.
4942 (verify_gimple_assign_single): Likewise.
4943 (verify_gimple_switch): Likewise.
4944 (verify_expr_location_1): Dereference tp once. Add (disabled)
4945 piece from verify_expr.
4946 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
4948 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
4950 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
4952 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
4954 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
4955 (small_data_pattern): Likewise.
4956 (arc_rewrite_small_data): Likewise.
4957 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
4958 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
4959 (get_symbol_alignment): New function.
4960 (legitimate_small_data_address_p): Likewise.
4961 (legitimate_scaled_address): Update, call
4962 legitimate_small_data_address_p.
4963 (output_sdata): New static variable.
4964 (arc_print_operand): Update how we handle small data operands.
4965 (arc_print_operand_address): Likewise.
4966 (arc_legitimate_address_p): Update, use
4967 legitimate_small_data_address_p.
4968 (arc_rewrite_small_data_p): Remove.
4969 (arc_rewrite_small_data_1): Likewise.
4970 (arc_rewrite_small_data): Likewise.
4971 (small_data_pattern): Likewise.
4972 (compact_sda_memory_operand): Update to use
4973 legitimate_small_data_address_p and get_symbol_alignment.
4974 (prepare_move_operands): Don't rewite sdata pattern.
4975 (prepare_extend_operands): Remove.
4976 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
4978 (zero_extendqisi2): Likewise.
4979 (zero_extendhisi2): Likewise.
4980 (extendqihi2): Likewise.
4981 (extendqisi2): Likewise.
4982 (extendhisi2): Likewise.
4986 * config/arc/constraints.md (Usd): Change it to memory constraint.
4988 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
4990 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
4991 as source of std instructions.
4992 * config/arc/arc.md (movsi_insn): Update pattern predicate to
4993 allow 6-bit constants as source for store instructions.
4994 (movdi_insn): Update instruction pattern to allow 6-bit constants
4995 as source for store instructions.
4997 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
4999 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
5001 2018-04-30 Nathan Sidwell <nathan@acm.org>
5002 Sandra Loosemore <sandra@codesourcery.com>
5004 * dumpfile.c (dump_open): Allow '-' for stdout.
5005 * doc/invoke.texi (Developer Options): Document dump filename
5006 determination early. Document stdin/stdout selection.
5008 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
5010 Microblaze Target: PIC data text relative
5012 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
5013 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
5015 * gcc/config/microblaze/microblaze.h (microblaze_constant_address_p):
5016 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
5017 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
5018 New addressing mode for data-text relative position indepenedent code.
5019 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
5020 'ADDRESS_SYMBOLIC_TXT_REL'.
5021 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
5022 (microblaze_legitimate_pic_operand): Exclude function calls from
5023 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
5024 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
5026 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
5027 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
5028 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
5029 for 'address + offset'.
5030 (microblaze_expand_prologue): Add new function prologue call for
5032 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
5033 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
5034 table in case of TARGET_PIC_DATA_TEXT_REL.
5035 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
5036 * gcc/config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
5037 Add new macros 'UNSPEC_TEXT',
5038 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
5039 + exclude function calls from 'UNSPEC_PLT' in case of data text
5041 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
5042 new target hook for generating address diff vector tables in case of
5044 * doc/tm.texi : Regenerate.
5045 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
5046 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
5047 of addr diff vector generation.
5048 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
5049 target hook definition.
5050 * targhooks.h, gcc/targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
5051 Add default function for generate_pic_addr_diff_vec -> flag_pic.
5052 * doc/invoke.texi (Add new pic option): Add new microblaze pic
5053 option for data text relative.
5055 2018-04-30 Richard Biener <rguenther@suse.de>
5057 * tree-chrec.h (evolution_function_is_constant_p): Remove
5059 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
5061 2018-04-30 Richard Biener <rguenther@suse.de>
5064 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
5066 2018-04-30 Richard Biener <rguenther@suse.de>
5068 PR tree-optimization/28364
5069 PR tree-optimization/85275
5070 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
5071 copying first exit test.
5073 2018-04-28 Mark Wielaard <mark@klomp.org>
5075 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
5077 (dwarf_AT): Handle DW_AT_addr_base.
5078 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
5080 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
5083 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
5084 (*ashl<dwi>3_doubleword_mask_1): Ditto.
5085 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
5086 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
5088 2018-04-28 Richard Biener <rguenther@suse.de>
5090 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
5091 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
5092 to reflect use. Only add interesting stmts.
5094 2018-04-27 Martin Jambor <mjambor@suse.cz>
5097 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
5098 the jump function allows for passing through aggregate values.
5100 2018-04-27 David Malcolm <dmalcolm@redhat.com>
5102 * input.h (in_system_header_at): Convert from macro to inline
5104 (from_macro_expansion_at): Likewise.
5105 (from_macro_definition_at): Likewise.
5107 2018-04-27 Jeff Law <law@redhat.com>
5109 * config.gcc: Mark tile* targets as deprecated/obsolete.
5111 2018-04-27 Richard Biener <rguenther@suse.de>
5113 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
5116 2018-04-27 Richard Biener <rguenther@suse.de>
5118 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
5120 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
5122 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
5123 with Yd constraint. Set "preferred_for_speed" attribute from
5124 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
5126 (*movdi_internal): Ditto.
5127 (movti_interunit splitters): Remove
5128 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
5129 (movdi_interunit splitters): Ditto.
5130 * config/i386/constraints.md (Ye): Remove.
5131 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
5133 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5136 * config/aarch64/constraints.md (Usg): Limit to 31.
5139 2018-04-27 Jakub Jelinek <jakub@redhat.com>
5141 PR tree-optimization/85529
5142 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
5143 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
5144 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
5145 zero extension or masking of the MSB bit.
5146 (optimize_range_tests): Add FIRST_BB argument, pass it through
5147 to optimize_range_tests_var_bound.
5148 (maybe_optimize_range_tests, reassociate_bb): Adjust
5149 optimize_range_tests callers.
5151 2018-04-26 Richard Biener <rguenther@suse.de>
5152 Jakub Jelinek <jakub@redhat.com>
5154 * cgraph.h (symbol_table): Just declare debug method here.
5155 * symtab.c (symbol_table::debug): Define.
5157 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
5159 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
5161 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
5163 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
5164 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
5165 (*movdi_internal): Substitute Yi and Yj constraint with x
5166 and Ym and Yn constraint with y constraint. Update "isa"
5167 attribute and set "preferred_for_speed" attribute from
5168 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
5169 (*movsi_internal): Ditto.
5170 (*movdf_internal): Ditto.
5171 (*movsf_internal): Ditto.
5172 (*zero_extendsidi2): Ditto.
5173 * config/i386/sse.md (vec_set<mode>_0): Ditto.
5174 (sse2_loadld): Ditto.
5175 (*vec_extract<ssevecmodelower>_0): Ditto.
5176 (*vec_extractv4si_0_zext_sse4): Ditto.
5177 (vec_concatv2di): Ditto.
5178 (*vec_dup<mode>): Ditto.
5179 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
5180 * config/i386/constraints.md (Yi): Remove.
5185 2018-04-26 Nathan Sidwell <nathan@acm.org>
5187 * dumpfile.c (dump_open): New.
5188 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
5189 (dump_finish): Detect stdio/stderr by value not name.
5191 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
5193 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
5195 2018-04-26 Tom de Vries <tom@codesourcery.com>
5198 * config/nvptx/nvptx.c (verify_neutering_jumps)
5199 (verify_neutering_labels): New function
5200 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
5202 2018-04-26 Tom de Vries <tom@codesourcery.com>
5205 * config/nvptx/nvptx.c (needs_neutering_p): New function.
5206 (nvptx_single): Use needs_neutering_p to skip over insns that do not
5209 2018-04-26 Richard Biener <rguenther@suse.de>
5210 Tom de Vries <tom@codesourcery.com>
5213 * lto-streamer-out.c (output_function): Fixup loops if required to match
5214 discovery done in the reader.
5216 2018-04-26 Richard Biener <rguenther@suse.de>
5218 PR tree-optimization/85116
5219 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
5220 have a loop exit from the single latch predecessor. Remove
5221 case of header with just condition.
5222 (ch_base::copy_headers): Exclude infinite loops from any
5224 (pass_ch::execute): Record exits.
5226 2018-04-26 Richard Biener <rguenther@suse.de>
5228 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
5229 prologue cost vector and pass it to vect_get_load_cost.
5230 (vect_get_peeling_costs_all_drs): Likewise.
5231 (vect_peeling_hash_get_lowest_cost): Likewise.
5232 (vect_enhance_data_refs_alignment): Likewise.
5234 2018-04-26 Richard Biener <rguenther@suse.de>
5237 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
5238 checking of integer<->pointer conversions.
5239 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
5240 sign-/zero-extending pointer types.
5241 (expand_omp_for_static_chunk): Likewise.
5243 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
5244 Jean Lee <xiaoyur347@gmail.com>
5246 * config/mips/mips.c (mips_asan_shadow_offset): New function.
5247 (TARGET_ASAN_SHADOW_OFFSET): Define.
5248 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
5249 true for -fsanitize=address.
5251 2018-04-25 Mark Wielaard <mark@klomp.org>
5253 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
5256 2018-04-25 Jakub Jelinek <jakub@redhat.com>
5258 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
5259 than "alu", remove explicit "memory" and "imm_disp" attributes.
5260 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
5263 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
5264 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
5265 gen_lowpart_no_emit.
5267 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
5270 * config/i386/i386.c (ix86_expand_builtin): Change memory
5271 operand to XI, extend p0 to Pmode.
5272 * config/i386/i386.md: Change unspec volatile and operand
5273 1 mode to XI, change operand 0 mode to P.
5275 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
5277 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
5278 GET_MODE_MASK before any checking.
5279 (nds32_can_use_bset_p): Likewise.
5280 (nds32_can_use_btgl_p): Likewise.
5282 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
5284 * config/nds32/nds32-doubleword.md: New define_split pattern for
5285 illegal register number.
5287 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
5289 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
5291 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
5293 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
5295 2018-04-25 Richard Biener <rguenther@suse.de>
5297 * lto-streamer.h (LTO_major_version): Bump to 8.
5299 2018-04-25 Jakub Jelinek <jakub@redhat.com>
5301 * BASE-VER: Set to 9.0.0.
5303 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
5305 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
5306 in __abskf2 and __powikf2.
5308 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5311 * config/aarch64/constraints.md (Usg, Usj): New constraints.
5312 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
5313 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
5314 Use the above on operand 2. Reindent.
5315 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
5317 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
5320 * common/config/i386/i386-common.c (ix86_handle_option): Don't
5322 * config/i386/i386.opt (mcet): Removed.
5323 * doc/install.texi: Remove -mcet documentation.
5324 * doc/invoke.texi: Likewise.
5326 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
5329 * doc/install.texi: Remove -mcet from bootstrap-cet.
5331 2018-04-24 Jakub Jelinek <jakub@redhat.com>
5334 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
5335 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
5339 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
5340 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
5341 containing a CONST_VECTOR.
5343 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
5345 * doc/install.texi: Update newlib dependency for nvptx.
5347 2018-04-24 Jakub Jelinek <jakub@redhat.com>
5350 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
5351 instead of INTVAL when shifting x left.
5353 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
5355 PR tree-optimization/85478
5356 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
5357 vect_grouped_store_supported for single element vectors.
5359 2018-04-24 Richard Biener <rguenther@suse.de>
5362 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
5363 load cost increase to the case of non-constant step.
5365 2018-04-24 Jakub Jelinek <jakub@redhat.com>
5368 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
5369 destination if any_malformed_asm.
5371 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
5374 * expr.c (store_field): In the bitfield case, if the value comes from
5375 a function call and is returned in registers by means of a PARALLEL,
5376 do not change the mode of the temporary unless BLKmode and VOIDmode.
5378 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
5380 PR rtl-optimization/85423
5381 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
5382 dependencies to debug insns when the previous insn is non-debug.
5384 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
5386 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
5387 enums into a single definition.
5388 (fls): Fix predicates and printing.
5391 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
5393 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
5394 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
5395 and short u6 immediate.
5396 (check_if_valid_sleep_operand): Remove.
5397 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
5399 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
5401 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
5402 flag_always_save_lp condition.
5403 * config/nds32/nds32.opt (malways-save-lp): New option.
5405 2018-04-22 Shiva Chen <shiva0217@gmail.com>
5407 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
5408 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
5409 * config/nds32/nds32.h
5410 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
5411 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
5413 2018-04-22 Shiva Chen <shiva0217@gmail.com>
5415 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
5416 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
5418 2018-04-22 Shiva Chen <shiva0217@gmail.com>
5419 Chung-Ju Wu <jasonwucj@gmail.com>
5421 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
5423 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
5424 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
5426 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
5428 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
5430 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
5432 * config/nds32/nds32-protos.h (nds32_data_alignment,
5433 nds32_local_alignment): Declare.
5434 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
5435 nds32_local_alignment): New functions.
5436 (TARGET_CONSTANT_ALIGNMENT): Define.
5437 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
5439 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
5441 * config/nds32/nds32.c
5442 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
5443 (TARGET_MODES_TIEABLE_P): Likewise.
5445 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
5447 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
5450 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
5451 Chung-Ju Wu <jasonwucj@gmail.com>
5453 * config/nds32/constants.md (unspec_volatile_element): Add enum values
5454 for unaligned access.
5455 * config/nds32/nds32-intrinsic.c: Implementation of expanding
5457 * config/nds32/nds32-intrinsic.md: Likewise.
5458 * config/nds32/nds32_intrinsic.h: Likewise.
5459 * config/nds32/nds32.h (nds32_builtins): Likewise.
5460 * config/nds32/nds32.opt (munaligned-access): New option.
5461 * config/nds32/nds32.c (nds32_asm_file_start): Display
5462 flag_unaligned_access status.
5464 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
5466 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
5467 -mno-relax is present.
5468 * config/riscv/linux.h (LINK_SPEC): Ditto.
5470 2018-04-20 Martin Sebor <msebor@redhat.com>
5473 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
5475 (gimple_fold_builtin_stxcpy_chk): Same.
5476 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
5478 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
5481 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
5482 __powikf2 when long double is IEEE 128-bit.
5484 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
5486 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
5487 step to make sure stack always aligned.
5489 2018-04-20 Carl Love <cel@us.ibm.com>
5492 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
5493 size check for arg0.
5495 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
5496 Tom de Vries <tom@codesourcery.com>
5499 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
5500 Emit insns for calls too.
5501 (nvptx_find_par): Always look for worker-level predecessor insn.
5502 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
5504 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
5505 (nvptx_process_pars): Propagate frames for calls.
5507 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
5510 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
5512 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5513 (ix86_handle_option): Don't handle OPT_mibt.
5514 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
5516 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
5518 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
5519 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
5520 (ix86_target_macros): Define __CET__ with flag_cf_protection
5521 for -fcf-protection.
5522 * config/i386/i386.c (isa2_opts): Remove -mibt.
5523 * config/i386/i386.h (TARGET_IBT): Removed.
5524 (TARGET_IBT_P): Likewise.
5525 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
5526 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
5527 * config/i386/i386.opt (mcet): Update help message.
5530 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
5531 -mcet as an alias for -mshstk.
5533 2018-04-20 Richard Biener <rguenther@suse.de>
5536 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
5537 complexity by forcing a single use of the multiply operand.
5539 2018-04-20 Martin Jambor <mjambor@suse.cz>
5542 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
5543 recursion dependency to only apply to non-clones.
5545 2018-04-20 Martin Jambor <mjambor@suse.cz>
5548 * ipa-cp.c (create_specialized_node): Check that clones of
5549 self-recursive edges exist during IPA-CP.
5551 2018-04-19 Toon Moene <toon@moene.org>
5553 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
5556 2018-04-19 Jakub Jelinek <jakub@redhat.com>
5558 PR tree-optimization/85467
5559 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
5560 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
5561 VECTOR_CST element to type.
5563 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
5566 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
5567 * config/i386/i386.md (builtin_setjmp_setup): Removed.
5568 (builtin_longjmp): Likewise.
5569 (save_stack_nonlocal): New pattern.
5570 (restore_stack_nonlocal): Likewise.
5572 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
5575 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
5576 Replace ASM_OUTPUT_LABEL with fprintf.
5578 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
5581 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
5582 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
5583 * config/i386/i386-c.c (ix86_target_macros_internal): Also
5584 define __IBT__ and __SHSTK__ for -fcf-protection.
5585 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
5587 (ix86_trampoline_init): Likewise.
5588 (x86_output_mi_thunk): Likewise.
5589 (ix86_notrack_prefixed_insn_p): Likewise.
5590 (ix86_option_override_internal): Don't disallow -fcf-protection.
5591 * config/i386/i386.md (rdssp<mode>): Also enable for
5593 (incssp<mode>): Likewise.
5594 (nop_endbr): Likewise.
5595 * config/i386/i386.opt (mcet): Change help message to built-in
5599 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
5600 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
5601 enable CET built-in functions.
5603 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
5605 * common/config/i386/i386-common.c
5606 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
5607 OPTION_MASK_ISA_MOVDIRI_UNSET,
5608 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
5609 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
5610 * config.gcc (movdirintrin.h): New header.
5611 * config/i386/cpuid.h (bit_MOVDIRI,
5612 bit_MOVDIR64B): New bits.
5613 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
5615 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
5616 (VOID, PVOID, PCVOID)): New function types.
5617 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
5618 __builtin_ia32_directstoreu_u64,
5619 __builtin_ia32_movdir64b): New builtins.
5620 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
5621 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
5623 (ix86_valid_target_attribute_inner_p): Ditto.
5624 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
5625 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
5626 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
5627 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
5628 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
5629 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
5630 (movdiri<mode>, movdir64b_<mode>): New.
5631 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
5632 * config/i386/immintrin.h: Include movdirintrin.h.
5633 * config/i386/movdirintrin.h: New file.
5634 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
5636 2018-04-19 Richard Biener <rguenther@suse.de>
5639 * cfg.c (clear_bb_flags): When loop state says we have
5640 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
5642 2018-04-19 Richard Biener <rguenther@suse.de>
5644 PR tree-optimization/84737
5645 * tree-vect-data-refs.c (vect_copy_ref_info): New function
5646 copying restrict info.
5647 (vect_setup_realignment): Use it.
5648 * tree-vectorizer.h (vect_copy_ref_info): Declare.
5649 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
5650 the first DR to all generated stores.
5651 (vectorizable_load): Likewise for loads.
5653 2018-04-19 Jakub Jelinek <jakub@redhat.com>
5655 PR tree-optimization/85446
5656 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
5657 the integral and pointer types to have the same precision.
5659 * doc/install.texi: Document --disable-cet being the default and
5662 2018-04-18 Martin Liska <mliska@suse.cz>
5664 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
5667 2018-04-18 Martin Liska <mliska@suse.cz>
5670 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
5673 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
5674 arguments if they are comparable.
5676 2018-04-18 Martin Liska <mliska@suse.cz>
5679 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
5682 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
5685 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
5688 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
5689 ENDBR after calling __morestack.
5691 2018-04-18 David Malcolm <dmalcolm@redhat.com>
5694 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
5695 by using gcc_base_ver to generate a gcc_driver_version, and use
5696 it when generating GCC_DRIVER_NAME.
5697 * configure: Regenerate.
5699 2018-04-18 Jakub Jelinek <jakub@redhat.com>
5702 * config.gcc: Obsolete powerpc*-*-*spe*.
5704 2018-04-17 Jakub Jelinek <jakub@redhat.com>
5707 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
5708 (stabstr_D): Change type of unum from unsigned int to
5709 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
5712 2018-04-17 Jim Wilson <jimw@sifive.com>
5715 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
5716 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
5717 Set arg_pointer_offset after using pretend_args_size.
5719 2018-04-17 Jakub Jelinek <jakub@redhat.com>
5721 PR rtl-optimization/85431
5722 * dse.c (record_store): Ignore zero width stores.
5725 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
5726 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
5727 __builtin_stack_restore rather than after it.
5728 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
5729 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
5730 argument instead of virtual_dynamic_stack_rtx.
5732 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
5734 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
5736 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5737 Add note to error message to explain internal mapping of overloaded
5738 built-in function name to non-overloaded built-in function name.
5739 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
5742 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
5745 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
5746 where the inputs overlap with the output.
5748 2018-04-17 Jakub Jelinek <jakub@redhat.com>
5750 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
5751 (=v, v) alternative and explicit "memory" attribute.
5752 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
5753 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
5755 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
5756 "sselog1" type instead of "sselog".
5757 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
5758 "sselog". Remove explicit "memory" attribute.
5759 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
5760 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
5762 (vec_extract_hi_v32hi): Merge all alternatives into one, use
5763 "sselog1" type instead of "sselog". Remove explicit "memory"
5765 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
5766 use "sselog1" type instead of "sselog". Remove explicit "memory"
5768 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
5769 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
5771 (vec_extract_hi_v64qi): Merge all alternatives into one, use
5772 "sselog1" type instead of "sselog". Remove explicit "memory"
5774 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
5775 use "sselog1" type instead of "sselog". Remove explicit "memory"
5779 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
5782 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
5785 2018-04-17 Martin Jambor <mjambor@suse.cz>
5788 * ipa-cp.c (create_specialized_node): Call
5789 expand_all_artificial_thunks if necessary.
5791 2018-04-17 Martin Liska <mliska@suse.cz>
5794 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
5795 in message, remote space in between '_G' and '('.
5797 2018-04-17 Jakub Jelinek <jakub@redhat.com>
5800 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
5801 avx512f_vmcmp<mode>3<round_saeonly_name>,
5802 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
5803 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
5804 avx512f_rndscale<mode><round_saeonly_name>,
5805 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
5806 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
5807 Use %<iptr>2 instead of %2 for -masm=intel.
5808 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
5809 avx512f_vcvttss2usi<round_saeonly_name>,
5810 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
5812 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
5813 avx512f_vcvttsd2usi<round_saeonly_name>,
5814 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
5815 Use %q1 instead of %1 for -masm=intel.
5816 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
5817 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
5818 of %3 for -masm=intel.
5819 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
5821 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
5823 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
5825 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
5826 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
5828 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
5830 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
5831 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
5832 %g1 and one with %0 and %1.
5833 (avx512er_vmrcp28<mode><round_saeonly_name>,
5834 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
5836 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
5837 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
5838 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
5839 of %0 and %{%4%} for -masm=intel.
5840 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
5841 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
5842 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
5843 order of %0 and %{%5%}%{z%} for -masm=intel.
5845 2018-04-17 Jan Hubicka <jh@suse.cz>
5848 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
5850 2018-04-17 Martin Liska <mliska@suse.cz>
5853 * multiple_target.c (create_dispatcher_calls): Set apostrophes
5854 for target_clone error message. Make default implementation
5855 clone to be a local declaration.
5856 (separate_attrs): Add new argument and check for an empty
5858 (expand_target_clones): Handle it.
5859 (ipa_target_clone): Make redirection just for target_clones
5862 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
5863 Tom de Vries <tom@codesourcery.com>
5866 * omp-expand.c (expand_oacc_for): Add dummy false branch for
5867 tiled basic blocks without omp continue statements.
5869 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
5872 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
5873 vec_extract expression as having side effects to make sure it gets
5876 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
5879 * config/i386/i386.c (get_builtin_code_for_version): Check
5882 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
5885 * gcc/config.gcc: Support "skylake".
5886 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Handle
5888 * gcc/config/i386/i386.c (m_SKYLAKE): Define.
5889 (processor_target_table): Add "skylake".
5890 (ix86_option_override_internal): Add "skylake".
5891 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
5892 PROCESSOR_CANNONLAKE.
5893 (get_builtin_code_for_version): Fix priority for
5894 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
5895 PROCESSOR_SKYLAKE-AVX512.
5896 * gcc/config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
5897 (processor_type): Add PROCESSOR_SKYLAKE.
5899 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
5900 Jason Merrill <jason@redhat.com>
5903 * convert.c (convert_to_integer_1): Use direct recursion for
5904 enumeral types and types with a precision less than the number
5905 of bits in their mode.
5907 2018-04-16 Julia Koval <julia.koval@intel.com>
5910 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
5911 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
5913 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
5916 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
5917 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
5918 and -mno-direct-move.
5920 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
5923 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
5924 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
5925 Ensure negative shifts result in {0}.
5927 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
5929 PR rtl-optimization/79916
5930 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
5931 regs (if any) to define how to gnerate SD moves when LRA is in
5934 2018-04-13 Jakub Jelinek <jakub@redhat.com>
5936 PR rtl-optimization/85393
5937 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
5938 * except.c (expand_dw2_landing_pad_for_region): Make static.
5939 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
5940 a label and unconditional jump to old_bb, rather than
5941 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
5944 PR rtl-optimization/85376
5945 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
5946 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
5947 instead of a specific value.
5949 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
5950 Bin Cheng <bin.cheng@arm.com>
5952 PR tree-optimization/82965
5953 PR tree-optimization/83991
5954 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
5955 by_profile_only parameter.
5956 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
5957 information if the loop was predicted to iterate too many times.
5958 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
5960 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
5963 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
5966 2018-04-13 Martin Liska <mliska@suse.cz>
5967 Jakub Jelinek <jakub@redhat.com>
5970 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
5971 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
5972 * builtins.c (expand_builtin_memory_copy_args): Use
5973 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
5974 handle dest_addr == pc_rtx.
5976 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
5979 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
5980 asked to not generate direct moves.
5981 (fix_trunc<mode>si2_stfiwx): Similar.
5982 (fix_trunc<mode>si2_internal): Similar.
5984 2018-04-12 Jakub Jelinek <jakub@redhat.com>
5987 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
5988 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
5989 lookup if dest in some wider mode is known to be const0_rtx and
5990 if so, record permanent equivalence for it to be ZERO_EXTEND of
5991 the narrower mode destination.
5993 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
5995 * lto-streamer-out.c (output_function): Revert 259346.
5996 * omp-expand.c (expand_oacc_for): Likewise.
5998 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
6000 PR rtl-optimization/85354
6001 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
6002 * sel-sched.c (sel_global_init): ... here.
6004 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
6007 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
6008 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
6009 mode for PE-COFF targets.
6010 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
6011 (i386_pe_asm_lto_end): Likewise.
6012 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
6013 (TARGET_ASM_LTO_END): Likewise.
6014 * config/i386/winnt.c (saved_debug_info_level): New static variable.
6015 (i386_pe_asm_lto_start): New function.
6016 (i386_pe_asm_lto_end): Likewise.
6018 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
6019 Richard Biener <rguenther@suse.de>
6022 * lto-streamer-out.c (output_function): Fix CFG loop state before
6024 * omp-expand.c (expand_oacc_for): Handle calls to internal
6025 functions like regular functions.
6027 2018-04-12 Richard Biener <rguenther@suse.de>
6030 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
6031 for the early LTO debug to properly generate references to it
6032 during DIE emission. Do not re-use that for the skeleton for
6034 (dwarf2out_early_finish): Likewise.
6036 2018-04-12 Jakub Jelinek <jakub@redhat.com>
6039 * config/i386/sse.md
6040 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
6041 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
6042 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
6043 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
6044 and output is a reg, avoid creating invalid lowpart subreg, but
6045 instead split into a 512-bit move. Don't split if not AVX512VL,
6046 input is xmm16+ reg and output is a mem.
6047 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
6048 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
6049 xmm16+ reg and output is a mem.
6051 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6053 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
6054 also for flag_dwarf2_cfi_asm.
6056 2018-04-12 Jakub Jelinek <jakub@redhat.com>
6058 PR rtl-optimization/85342
6059 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
6060 a bool scalar var inside of the loop instead. Don't try to update
6061 recog_data.operand after failed apply_change_group.
6063 2018-04-12 Tom de Vries <tom@codesourcery.com>
6066 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
6067 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
6068 array with flexible array member as array without given dimension.
6069 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
6070 argument for undefined param to true.
6072 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
6075 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
6076 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
6077 from PowerPC section.
6078 * config/rs6000/sysv4.opt (mcall-): Improve help text.
6079 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
6080 help text that is too long.
6081 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
6082 help text that is too long.
6083 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
6084 help text that is too long.
6086 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
6088 * config/alpha/alpha.md (stack_probe_internal): Rename
6089 from "probe_stack". Update all callers.
6091 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
6093 PR rtl-optimization/84566
6094 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
6095 sched_macro_fuse_insns.
6097 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
6100 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
6101 (compute_block_dependences): ... from here.
6103 2018-04-11 Jakub Jelinek <jakub@redhat.com>
6105 PR tree-optimization/85331
6106 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
6107 from int to HOST_WIDE_INT.
6109 2018-04-11 Martin Jambor <mjambor@suse.cz>
6112 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
6113 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
6114 not the same as the source val.
6115 (cgraph_edge_brings_value_p): New parameter.
6116 (gather_edges_for_value): Pass destination value to
6117 cgraph_edge_brings_value_p.
6118 (perhaps_add_new_callers): Likewise.
6119 (get_info_about_necessary_edges): Likewise and exclude values brought
6120 only by self-recursive edges.
6121 (create_specialized_node): Redirect only clones of self-calling edges.
6122 (+self_recursive_pass_through_p): New function.
6123 (find_more_scalar_values_for_callers_subset): Use it.
6124 (find_aggregate_values_for_callers_subset): Likewise.
6125 (known_aggs_to_agg_replacement_list): Removed.
6126 (decide_whether_version_node): Re-calculate known constants for all
6127 remaining context clones.
6129 2018-04-11 Richard Biener <rguenther@suse.de>
6132 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
6133 from early DWARF output.
6134 (dwarf2out_early_finish): Output line info unconditionally into
6135 early DWARF and add reference to it.
6137 2018-04-11 Jakub Jelinek <jakub@redhat.com>
6140 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
6141 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
6142 other than V2DFmode using iptr mode attribute.
6143 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
6145 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
6147 PR rtl-optimization/84659
6148 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
6150 2018-04-11 Jakub Jelinek <jakub@redhat.com>
6153 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
6155 (output_loc_list): Pass address of a dummy size variable even in the
6156 locview handling loop.
6157 (index_location_lists): Add comment on why skip_loc_list_entry can't
6160 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
6163 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
6166 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
6169 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
6170 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
6171 and -mstring-compare-inline-limit.
6173 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
6176 * gcc/config/rs6000/rs6000.md (allocate_stack): Put the residual size
6177 for stack clash protection in a register whenever we need it to be in
6180 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
6182 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
6183 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
6185 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
6188 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
6190 (mlong-double-): Ditto.
6191 * config/rs6000/sysv4.opt (msdata=): Ditto.
6192 (mtls-size=): Ditto.
6194 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
6196 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6197 erroneous entries for
6198 "vector int vec_ldl (int, long int *)", and
6199 "vector unsigned int vec_ldl (int, unsigned long int *)".
6200 Add comments and entries for
6201 "vector bool char vec_ldl (int, bool char *)",
6202 "vector bool short vec_ldl (int, bool short *)",
6203 "vector bool int vec_ldl (int, bool int *)",
6204 "vector bool long long vec_ldl (int, bool long long *)",
6205 "vector pixel vec_ldl (int, pixel *)",
6206 "vector long long vec_ldl (int, long long *)",
6207 "vector unsigned long long vec_ldl (int, unsigned long long *)".
6208 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
6209 type tree bool_long_long_type_node and correct definition of
6210 bool_V2DI_type_node to make reference to this new type tree.
6211 (rs6000_mangle_type): Replace erroneous reference to
6212 bool_long_type_node with bool_long_long_type_node.
6213 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
6214 comments to emphasize sign distinctions for char and int types and
6215 replace RS6000_BTI_bool_long constant with
6216 RS6000_BTI_bool_long_long constant. Also add comment to restrict
6217 use of RS6000_BTI_pixel.
6218 (bool_long_type_node): Remove this macro definition.
6219 (bool_long_long_type_node): New macro definition
6221 2018-04-10 Jakub Jelinek <jakub@redhat.com>
6223 PR rtl-optimization/85300
6224 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
6225 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
6226 simplify_unary_operation fails.
6228 2018-04-10 Martin Liska <mliska@suse.cz>
6230 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
6231 cgraph_edge and ipa_ref.
6233 2018-04-10 Jakub Jelinek <jakub@redhat.com>
6237 * config/i386/sse.md
6238 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
6239 computation of the VEC_MERGE selector from mask.
6240 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
6241 Fix decoding of the VEC_MERGE selector into mask.
6243 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
6245 PR tree-optimization/85286
6246 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
6248 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
6250 * final.c (final_1): Set insn_last_address as well as
6251 insn_current_address.
6253 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6256 * explow.c (emit_stack_probe): Call validize_mem on memory location
6257 before passing it to gen_probe_stack. Create address operand and
6258 legitimize it for the probe_stack_address case.
6260 2018-04-09 Jan Hubicka <jh@suse.cz>
6263 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
6264 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
6265 * tree.c (free_lang_data_in_type): Fix handling of binfos;
6267 (free_lang_data): Rebuild type inheritance graph.
6269 2018-04-09 Martin Sebor <msebor@redhat.com>
6271 * invoke.texi (-finline-small-functions): Mention other optimization
6273 (-findirect-inlining, -fpartial-inlining): Same.
6274 (-finline-functions-called-once): Same.
6275 (-freorder-blocks-and-partition): Same.
6277 2018-04-09 Jan Hubicka <jh@suse.cz>
6280 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
6281 jumps; choose last target that matches the criteria (i.e.
6282 no partition changes for non-crossing jumps).
6283 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
6284 support for redirecting crossing jumps to non-crossing.
6286 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
6288 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
6289 also for naked functions.
6291 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
6293 * config/arc/arc.md (add_shift): New pattern.
6294 (add_shift2): Likewise.
6295 (sub_shift): Likewise.
6296 (sub_shift_cmp0_noout): Likewise.
6297 (compare_si_ashiftsi): Likewise.
6298 (xbfu_cmp0_noout): New combine pattern.
6299 (xbfu_cmp0"): Likewise.
6300 (movsi_set_cc_insn): Place the predicable variant first.
6301 (commutative_binary_cmp0_noout): Remove clobber.
6302 (commutative_binary_cmp0): New pattern.
6303 (noncommutative_binary_cmp0): Likewise.
6304 (noncommutative_binary_cmp0_noout): Likewise.
6305 (noncommutative_binary_comparison_result_used): Removed.
6306 (rsub_cmp0): New pattern.
6307 (rsub_cmp0_noout): Likewise.
6308 (extzvsi): Changed, keep only meaningful variants.
6309 (SQH, SEZ): New iterators.
6310 (SQH_postfix): New mode attribute.
6311 (SEZ_prefix): New code attribute.
6312 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
6313 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
6314 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
6316 (noncommutative_operator): Check the availability of barrel
6319 2018-04-09 Richard Biener <rguenther@suse.de>
6321 PR tree-optimization/85284
6322 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
6323 Only use the niter constraining form of simple_iv when the exit
6326 2018-04-09 Tom de Vries <tom@codesourcery.com>
6329 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
6330 (define_expand "*memory_barrier"): New define_expand.
6331 (define_insn "memory_barrier"): New insn.
6333 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
6335 PR rtl-optimization/80463
6336 PR rtl-optimization/83972
6337 PR rtl-optimization/83480
6339 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
6340 correct producer for the insn.
6341 (tidy_control_flow): Fixup seqnos in case of debug insns.
6343 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
6345 PR rtl-optimization/83913
6347 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
6348 different sched-times when merging exprs.
6350 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
6352 PR rtl-optimization/83962
6354 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
6355 tidy_fallthru_edge and tidy_control_flow.
6357 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
6359 PR rtl-optimization/83530
6361 * sel-sched.c (force_next_insn): New global variable.
6362 (remove_insn_for_debug): When force_next_insn is true, also leave only
6363 next insn in the ready list.
6364 (sel_sched_region): When the region wasn't scheduled, make another pass
6365 over it with force_next_insn set to 1.
6367 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
6369 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
6371 * config/nds32/constants.md (unspec_volatile_element): Add enum values
6372 for interrupt control.
6373 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
6374 functions for interrupt control.
6375 * config/nds32/nds32-intrinsic.md: Likewise.
6376 * config/nds32/nds32_intrinsic.h: Likewise.
6377 * config/nds32/nds32.h (nds32_builtins): Likewise.
6379 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
6381 * config/nds32/nds32.c (nds32_init_machine_status,
6382 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
6383 strict_aligned_p field.
6384 (nds32_expand_to_rtl_hook): New function.
6385 (TARGET_EXPAND_TO_RTL_HOOK): Define.
6386 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
6388 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
6389 Chung-Ju Wu <jasonwucj@gmail.com>
6391 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
6392 * config/nds32/nds32-n7.md: New file.
6393 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
6394 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
6396 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
6397 * config/nds32/nds32.md (pipeline_model): Add n7.
6398 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
6399 * config/nds32/pipelines.md: Include n7 settings.
6401 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
6402 Chung-Ju Wu <jasonwucj@gmail.com>
6404 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
6405 * config/nds32/nds32-e8.md: New file.
6406 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
6407 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
6409 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
6410 * config/nds32/nds32.md (pipeline_model): Add e8.
6411 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
6412 * config/nds32/pipelines.md: Include e8 settings.
6414 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
6415 Chung-Ju Wu <jasonwucj@gmail.com>
6417 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
6418 * config/nds32/nds32-n8.md: New file.
6419 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
6420 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
6422 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
6423 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
6424 * config/nds32/nds32.md (pipeline_model): Add n8.
6425 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
6426 * config/nds32/pipelines.md: Include n8 settings.
6428 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
6429 Chung-Ju Wu <jasonwucj@gmail.com>
6431 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
6432 * config/nds32/nds32-n9-2r1w.md: New file.
6433 * config/nds32/nds32-n9-3r2w.md: New file.
6434 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
6435 nds32_register_ports): New or modify for cpu n9.
6436 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
6438 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
6439 * config/nds32/nds32-utils.c: New file.
6440 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
6441 TARGET_MUL_SLOW): Define.
6442 * config/nds32/nds32.md (pipeline_model): New attribute.
6443 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
6444 New options that support cpu n9.
6445 * config/nds32/pipelines.md: Include n9 settings.
6446 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
6448 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
6450 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
6451 information if necessary.
6452 (output_cond_branch_compare_zero): Likewise.
6453 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
6454 (nds32_target_alignment): Refine for alignment.
6455 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
6456 (FUNCTION_BOUNDARY): Modify.
6457 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
6459 * config/nds32/nds32.opt (malways-align, malign-functions): New.
6461 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
6463 * config/nds32/constants.md (unspec_volatile_element): Add values for
6464 TLB operation and data prefetch.
6465 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
6466 functions for TLB operation and data prefetch.
6467 * config/nds32/nds32-intrinsic.md: Likewise.
6468 * config/nds32/nds32_intrinsic.h: Likewise.
6469 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
6470 (nds32_print_operand): Likewise.
6471 * config/nds32/nds32.h (nds32_builtins): Likewise.
6473 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6474 Andrew Pinski <pinsika@gcc.gnu.org>
6477 * match.pd: Use constant_boolean_node of correct type instead of
6478 boolean_true_node or boolean_false_node for simplifying
6479 pointer comparisons to zero.
6481 2018-04-07 Jakub Jelinek <jakub@redhat.com>
6483 PR tree-optimization/80021
6484 * tree.c (verify_type_variant): Make error call in verify_variant_match
6485 translatable and remove final full stop.
6487 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
6489 * config/nds32/constants.md (unspec_volatile_element): Add
6490 UNSPEC_VOLATILE_EH_RETURN.
6491 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
6492 nds32_output_stack_pop): Support dwarf exception handling process.
6493 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
6494 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
6495 exception handling process.
6496 (nds32_compute_stack_frame): Likewise.
6497 (nds32_return_addr_rtx): Likewise.
6498 (nds32_initial_elimination_offset): Likewise.
6499 (nds32_expand_prologue): Likewise.
6500 (nds32_expand_epilogue): Likewise.
6501 (nds32_dynamic_chain_address): New function.
6502 * config/nds32/nds32.h (machine_function): Add fields for dwarf
6504 (DYNAMIC_CHAIN_ADDRESS): Define.
6505 (EH_RETURN_DATA_REGNO): Define.
6506 (EH_RETURN_STACKADJ_RTX): Define.
6507 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
6508 patterns for dwarf exception handling.
6510 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
6512 * config/nds32/nds32.h: Clean up obsolete macros.
6514 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
6516 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
6517 Add enum values for particular instructions.
6518 * config/nds32/nds32-intrinsic.c: Implementation of expanding
6519 particular intrinsic functions.
6520 * config/nds32/nds32-intrinsic.md: Likewise.
6521 * config/nds32/nds32_intrinsic.h: Likewise.
6522 * config/nds32/nds32.h (nds32_builtins): Likewise.
6523 * config/nds32/nds32.md (type): Add pbsad and pbsada.
6524 (btst, ave): New patterns for particular instructions.
6526 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
6528 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
6529 Add enum values for atomic load/store and memory sync.
6530 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
6532 * config/nds32/nds32-intrinsic.md: Likewise.
6533 * config/nds32/nds32_intrinsic.h: Likewise.
6534 * config/nds32/nds32.h (nds32_builtins): Likewise.
6536 2018-04-07 Jakub Jelinek <jakub@redhat.com>
6538 PR tree-optimization/85257
6539 * fold-const.c (native_encode_vector): If not all elts could fit
6540 and off is -1, return 0 rather than offset.
6541 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
6542 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
6543 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
6544 adjust buffer in native_interpret_expr call.
6546 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
6548 * config/nds32/constants.md (unspec_volatile_element): Add cache
6549 control enum values.
6550 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
6551 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
6552 * config/nds32/nds32.c (nds32_cctl_names): New.
6553 (nds32_print_operand): Handle cache control register names.
6554 * config/nds32/nds32.h (nds32_builtins): New enum values.
6555 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
6557 * config/nds32/nds32.md (type): Add mmu.
6558 * config/nds32/pipelines.md (simple_insn): Add mmu.
6560 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
6562 * config/nds32/nds32.md (type): Remove call.
6563 * config/nds32/pipelines.md (simple_insn): Likewise.
6565 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
6567 * config/nds32/constants.md (unspec_volatile_element): Add
6568 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
6569 UNSPEC_VOLATILE_FMFCFG.
6570 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
6571 description for fmfcfg and fmfcsr.
6572 (bdesc_1arg): Add fmtcsr.
6573 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
6574 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
6575 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
6576 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
6577 unspec_fmfcfg): New patterns.
6578 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
6579 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
6580 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
6581 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
6582 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
6583 __nds32__fmfcfg): Define.
6585 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
6587 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
6588 intrinsic register names.
6589 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
6590 intrinsic register enum values and macros.
6592 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
6594 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
6595 for load/store addressing form.
6596 (nds32_print_operand_address): Likewise.
6598 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
6601 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
6602 based on LABEL_REF. Remove useless assertion.
6603 (pic_address_needs_scratch): Fix formatting.
6604 (sparc_legitimize_pic_address): Minor tweaks.
6605 (sparc_delegitimize_address): Adjust assertion accordingly.
6606 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
6607 into symbolic_operand.
6608 (movsi_high_pic_label_ref): Likewise.
6609 (movsi_lo_sum_pic_label_ref): Likewise.
6610 (movdi_pic_label_ref): Likewise.
6611 (movdi_high_pic_label_ref): Likewise.
6612 (movdi_lo_sum_pic_label_ref): Likewise.
6614 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
6616 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
6617 custom LIB_SPEC setup.
6619 2018-04-06 Ruslan Bukin <br@bsdpad.com>
6620 Kito Cheng <kito.cheng@gmail.com>
6622 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
6623 * config/riscv/freebsd.h: New.
6625 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
6627 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
6628 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
6631 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
6632 Kito Cheng <kito.cheng@gmail.com>
6634 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
6635 nds32_output_call, nds32_symbol_binds_local_p): New functions.
6636 * config/nds32/nds32-protos.h (nds32_output_call,
6637 nds32_output_return): Declare.
6638 * config/nds32/nds32.md: Refine all the call and return patterns.
6640 2018-04-06 Jakub Jelinek <jakub@redhat.com>
6643 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
6644 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
6646 PR rtl-optimization/84872
6647 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
6648 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
6651 2018-04-06 Tamar Christina <tamar.christina@arm.com>
6653 * expr.c (copy_blkmode_to_reg): Revert 254862.
6654 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
6656 2018-04-06 Richard Biener <rguenther@suse.de>
6659 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
6660 after seeing a component reference with an adjacent field. Treat
6661 refs to arrays at struct end of external decls similar to
6662 refs to unconstrained commons.
6664 2018-04-06 Jakub Jelinek <jakub@redhat.com>
6667 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
6668 look through SAVE_EXPRs with non-side-effects argument. Adjust
6670 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
6673 2018-04-06 Richard Biener <rguenther@suse.de>
6676 * alias.c (find_base_term): New wrapper around find_base_term
6677 unwinding CSELIB_VAL_PTR changes.
6678 (find_base_term): Do not restore CSELIB_VAL_PTR during the
6681 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6683 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
6685 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
6686 constant definitions.
6687 ("nop"): lr 0,0 -> nopr r0
6688 ("nop_lr0", "nop_lr1"): New insn definitions.
6690 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
6692 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
6693 NDS32_V3PUSH_AVAILABLE_P macro.
6695 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
6696 Chung-Ju Wu <jasonwucj@gmail.com>
6698 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
6699 (nds32*-*-*): Add float and fpu_config into supported_defaults.
6700 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6701 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
6702 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
6703 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
6704 * config/nds32/constraints.md: New constraints and checking for hard
6705 float configuration.
6706 * config/nds32/iterators.md: New mode iterator and attribute for hard
6707 float configuration.
6708 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
6710 * config/nds32/nds32-fpu.md: New file.
6711 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
6712 deal with hard float code generation.
6713 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
6715 (abi_type, float_reg_number): New enum type.
6716 * config/nds32/nds32-predicates.c: New predicates for hard float.
6717 * config/nds32/nds32-protos.h: Declare functions for hard float.
6718 * config/nds32/nds32.c: Implementation for hard float configuration.
6719 * config/nds32/nds32.h: Definitions for hard float configuration.
6720 * config/nds32/nds32.md: Include hard float machine description and
6721 modify patterns for hard float configuration.
6722 * config/nds32/nds32.opt: New options for hard float configuration.
6723 * config/nds32/predicates.md: New predicates for hard float
6726 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6728 * common/config/nds32/nds32-common.c
6729 (nds32_option_optimization_table): Enable -mreleax-hint by default.
6731 2018-04-05 Jakub Jelinek <jakub@redhat.com>
6734 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
6735 CONSTRUCTOR_ELT (ctor, ...)->value.
6737 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
6740 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
6742 2018-04-05 Tom de Vries <tom@codesourcery.com>
6745 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
6748 2018-04-05 Shiva Chen <shiva0217@gmail.com>
6749 Kito Cheng <kito.cheng@gmail.com>
6751 * config/nds32/constraints.md (U33): Fine-tune checking condition.
6752 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
6753 * config/nds32/nds32.h (nds32_16bit_address_type): Add
6754 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
6756 2018-04-05 Shiva Chen <shiva0217@gmail.com>
6757 Kito Cheng <kito.cheng@gmail.com>
6759 * config/nds32/constraints.md (Ufe): New memory constraint.
6760 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
6761 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
6762 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
6764 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
6765 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
6767 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6769 * config/nds32/nds32.md: Use optimize_size in the condition for
6770 alu-shift instructions.
6772 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6774 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
6776 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6778 * config/nds32/nds32.md (negsi2): Refine pattern.
6780 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
6781 Chung-Ju Wu <jasonwucj@gmail.com>
6783 * config/nds32/iterators.md (shift_rotate): New code iterator.
6784 (shift): New code attribute.
6785 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
6786 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
6787 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
6788 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
6789 bit-wise operations.
6790 (andsi3, *andsi3): Ditto.
6791 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
6792 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
6793 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
6794 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
6795 nds32_ior_operand, nds32_xor_operand): New predicates.
6797 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6799 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
6800 (addsi3, subsi3): ... this.
6802 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6804 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
6806 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6808 * config/nds32/nds32.md: Adjust indention.
6810 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
6812 * config/nds32/nds32.md (feature): New attribute.
6814 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
6816 * config/nds32/nds32.md (subtype): New attribute.
6818 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6821 * config/arm/arm-builtins.c (arm_expand_builtin): Change
6822 expansion to perform a bitwise AND of the argument followed by a
6823 boolean negation of the result.
6825 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
6827 PR rtl-optimization/84878
6828 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
6829 the basic block. Assert the use reference is not artificial and that
6830 it has an associated insn.
6832 2018-04-04 Michael Matz <matz@suse.de>
6834 * builtins.c (compute_objsize): Pass correct operand
6835 to array_at_struct_end_p.
6837 2018-04-04 Richard Biener <rguenther@suse.de>
6840 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
6841 from contexts for DINFO_LEVEL_TERSE and below.
6843 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
6845 * config/nds32/nds32-doubleword.md (move_<mode>): Require
6846 resiter_operand condition.
6847 * config/nds32/nds32.md (*move<mode>): Ditto.
6849 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
6850 Monk Chiang <sh.chiang04@gmail.com>
6852 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
6854 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
6856 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
6858 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
6859 Kito Cheng <kito.cheng@gmail.com>
6861 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
6862 nds32_cond_code_str, output_cond_branch,
6863 output_cond_branch_compare_zero, nds32_expand_cbranch,
6864 nds32_expand_cstore, nds32_expand_movcc,
6865 nds32_output_cbranchsi4_equality_zero,
6866 nds32_output_cbranchsi4_equality_reg,
6867 nds32_output_cbranchsi4_equality_reg_or_const_int,
6868 nds32_output_cbranchsi4_greater_less_zero: New functions.
6869 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
6870 nds32_expand_cstore, nds32_expand_movcc,
6871 nds32_output_cbranchsi4_equality_zero,
6872 nds32_output_cbranchsi4_equality_reg,
6873 nds32_output_cbranchsi4_equality_reg_or_const_int,
6874 nds32_output_cbranchsi4_greater_less_zero): Declare.
6875 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
6876 nds32_rimm11s_operand): New predicates.
6877 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
6878 * config/nds32/nds32.md: Rewrite all the branch and conditional move
6881 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
6883 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
6884 * config/nds32/nds32.md: Ditto.
6885 * config/nds32/pipelines.md: Ditto.
6887 2018-04-04 Richard Biener <rguenther@suse.de>
6889 PR tree-optimization/85168
6890 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
6891 propagating abnormals.
6893 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
6895 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
6897 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
6898 Kito Cheng <kito.cheng@gmail.com>
6900 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
6901 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
6902 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
6903 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
6904 * config/nds32/nds32.md (sibcall_internal): New.
6905 (sibcall_register): Remove.
6906 (sibcall_immediate): Remove.
6907 (sibcall_value_internal): New.
6908 (sibcall_value_register): Remove.
6909 (sibcall_value_immediate): Remove.
6910 * config/nds32/predicates.md (nds32_general_register_operand): New.
6911 (nds32_call_address_operand): New.
6913 2018-04-03 Jakub Jelinek <jakub@redhat.com>
6915 PR rtl-optimization/85167
6916 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
6917 bb_defs if *split_p, instead preinitialize it to NULL.
6919 PR tree-optimization/85156
6920 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
6921 evaluating the argument multiple times.
6923 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
6925 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
6927 (_mm_cvtpd_ps): Likewise.
6928 (_mm_cvttpd_epi32): Likewise.
6929 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
6930 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
6931 vector, pixel, and bool following altivec.h include.
6933 2018-04-03 Martin Sebor <msebor@redhat.com>
6935 * doc/extend.texi (Common Function Attributes): Clarify.
6936 (const attribute): Likewise.
6937 (pure attribute): Likewise.
6939 2018-04-03 Jakub Jelinek <jakub@redhat.com>
6942 * config/i386/i386.c (ix86_expand_vector_set): Use
6943 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
6945 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
6947 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
6948 instructions when changing rounding bits to preserve precision bits
6949 in the x87 control word.
6951 2018-04-03 Martin Liska <mliska@suse.cz>
6953 PR tree-optimization/82491
6954 * rtl.h (strip_offset_and_add): Replace += suboffset with
6955 poly_uint64 () + suboffset.
6957 2018-03-29 Martin Liska <mliska@suse.cz>
6958 Martin Jambor <mjambor@suse.cz>
6961 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
6962 param_type is not an integral or pointer type.
6964 2018-04-03 Richard Biener <rguenther@suse.de>
6966 * sese.h (recompute_all_dominators): Remove.
6968 2018-04-02 Martin Sebor <msebor@redhat.com>
6970 * doc/invoke.texi (-Wrestrict): Fix typos.
6972 2018-04-02 Jim Wilson <jimw@sifive.com>
6974 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
6975 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
6976 (<optab>di3, <optab>si3_extend): Likewise.
6977 (<optab>si3_mask, <optab>si3_mask_1): New.
6978 (<optab>di3_mask, <optab>di3_mask_1): New.
6979 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
6980 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
6981 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
6983 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
6985 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
6988 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
6990 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
6991 (nds32_canonicalize_comparison): New function.
6993 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
6994 Kito Cheng <kito.cheng@gmail.com>
6995 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6997 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
6998 * config/nds32/constants.md (unspec_volatile_element): Add
6999 UNSPEC_VOLATILE_RELAX_GROUP.
7000 * config/nds32/nds32-relax-opt.c: New file.
7001 * config/nds32/nds32-predicates.c
7002 (nds32_symbol_load_store_p): New function.
7003 * config/nds32/nds32-protos.h
7004 (nds32_symbol_load_store_p): Declare function.
7005 (make_pass_nds32_relax_opt): Declare new rtl pass function.
7006 * config/nds32/nds32.c
7007 (nds32_register_pass): New function to register pass.
7008 (nds32_register_passes): New function to register passes.
7009 * config/nds32/nds32.md (relax_group): New pattern.
7010 * config/nds32/nds32.opt (mrelax-hint): New option.
7011 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
7013 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
7015 * config/nds32/t-nds32: Modify files dependency.
7017 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
7019 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
7020 (PROFILE_HOOK): Define its implementation.
7022 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
7024 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
7025 type and 32-bit size.
7027 2018-04-01 Jakub Jelinek <jakub@redhat.com>
7030 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
7031 (V_128_256): New mode iterator.
7032 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
7033 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
7034 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
7036 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
7037 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
7039 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
7042 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
7043 NaN inputs correctly.
7045 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
7048 * config/rs6000/vsx.md (??r): New mode attribute.
7049 (*vsx_mov<mode>_64bit): Use it.
7050 (*vsx_mov<mode>_32bit): Likewise.
7052 2018-03-30 Martin Sebor <msebor@redhat.com>
7054 PR tree-optimization/84818
7055 * builtins.c (check_access): Use warning_n.
7057 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7060 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
7062 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
7065 2018-03-30 Julia Koval <julia.koval@intel.com>
7068 * x86-tune.def (movx, partial_reg_dependency): Enable for
7071 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
7074 * lra-constraints.c (process_alt_operands): Move setting
7075 this_alternative_matches below.
7077 2018-03-29 Martin Liska <mliska@suse.cz>
7080 * doc/invoke.texi: Document how LTO works with debug info.
7081 Describe auto-load support of binutils. Mention 'x86-64'
7082 as valid option value of -march option.
7084 2018-03-29 Jakub Jelinek <jakub@redhat.com>
7086 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
7089 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
7090 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
7091 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
7094 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
7097 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
7098 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
7099 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
7100 for RS6000_BTM_POWERPC64.
7101 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
7102 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
7103 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
7108 2018-03-28 Carl Love <cel@us.ibm.com>
7111 2017-09-27 Carl Love <cel@us.ibm.com>
7113 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7114 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7115 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7118 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7120 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
7121 instead of __vector bool.
7122 (_mm_max_pu8): Likewise.
7123 (_mm_min_pi16): Likewise.
7125 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
7128 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
7129 (DIVWEUO): Likewise.
7131 (DIVDEUO): Likewise.
7132 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
7133 DIVWEUO and DIVDEUO.
7134 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
7135 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
7136 (div_extend): Likewise.
7137 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
7139 (__builtin_divweuo): Likewise.
7140 (__builtin_divdeo): Likewise.
7141 (__builtin_divdeuo): Likewise.
7143 2018-03-28 Jakub Jelinek <jakub@redhat.com>
7146 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
7147 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
7149 PR tree-optimization/82004
7150 * gimple-match-head.c (optimize_pow_to_exp): New function.
7151 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
7152 Don't fold to exp if optimize_pow_to_exp is false.
7154 2018-03-28 Martin Liska <mliska@suse.cz>
7157 * calls.c (initialize_argument_information): Fix trailing space.
7158 * common.opt: Fix typo and provide better explanation for
7159 -fsanitize-coverage option.
7160 * config/i386/i386.opt: Fix typo.
7162 2018-03-28 Jakub Jelinek <jakub@redhat.com>
7163 Martin Liska <mliska@suse.cz>
7166 * gimplify.c (asan_poison_variable): Don't do the check for
7167 gimplify_omp_ctxp here.
7168 (gimplify_decl_expr): Do it here.
7169 (gimplify_target_expr): Likewise.
7171 2018-03-28 Martin Liska <mliska@suse.cz>
7174 * config/i386/i386.c (ix86_function_arg_advance): Do not call
7175 chkp_type_bounds_count if MPX is not enabled.
7177 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
7179 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
7181 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7184 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
7185 function to create the function decl for complex long double
7186 multiply and divide for -mabi=ieeelongdouble.
7187 (init_float128_ieee): Call it.
7189 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
7192 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
7193 -fcf-protection=branch -mibt.
7194 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
7196 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7199 * config/arm/arm.c (arm_valid_symbolic_address): Handle
7200 arm_word_relocations.
7202 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
7205 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
7206 extern array declarations.
7208 2018-03-27 Richard Biener <rguenther@suse.de>
7211 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
7212 explicit single_use checks.
7214 2018-03-27 Richard Biener <rguenther@suse.de>
7216 PR tree-optimization/85082
7217 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
7220 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7222 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
7223 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
7224 Turn on fasynchronous-unwind-tables and funwind-tables.
7226 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
7229 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
7230 (*bmi_blsr_<mode>_ccz): Ditto.
7232 2018-03-26 Tom de Vries <tom@codesourcery.com>
7234 PR tree-optimization/85063
7235 * omp-general.c (offloading_function_p): New function. Factor out
7237 * omp-offload.c (pass_omp_target_link::gate): ... here.
7238 * omp-general.h (offloading_function_p): Declare.
7239 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
7240 with attribute omp declare target for offloading functions.
7242 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
7244 PR tree-optimization/84005
7245 * tree-data-ref.h (get_base_for_alignment): Declare.
7246 * tree-data-ref.c (get_base_for_alignment_1): New function.
7247 (get_base_for_alignment): Likewise.
7248 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7249 get_base_for_alignment to find a suitable base object, instead
7250 of always using drb->base_address.
7252 2018-03-23 Jakub Jelinek <jakub@redhat.com>
7255 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
7256 known size by default.
7258 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
7261 * lra-constraints.c (process_alt_operands): Don't match BLKmode
7262 and non BLKmode operands.
7264 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7267 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
7268 Clean up attributes.
7270 2018-03-23 Richard Biener <rguenther@suse.de>
7273 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
7274 we are going to emit early debug for LTO.
7276 2018-03-23 Jakub Jelinek <jakub@redhat.com>
7279 * function.c (match_asm_constraints_1): Don't optimize if input
7280 doesn't satisfy general_operand predicate for output's mode.
7283 * alias.c (write_dependence_p): Don't require for x_canonicalized
7284 non-VOIDmode if x has VOIDmode.
7287 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
7288 just don't try to optimize it rather than assert it never happens.
7290 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7292 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
7293 macro expansions for definition of ST_INTERNAL_<mode> and
7294 LD_INTERNAL_<mode> builtins.
7295 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
7297 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
7299 (altivec_expand_st_builtin): Likewise.
7300 (altivec_expand_builtin): Remove calls to deleted functions.
7301 (rs6000_address_for_altivec): Delete this function.
7302 * config/rs6000/vector.md: Remove expands for
7303 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
7305 2018-03-22 Sudakshina Das <sudi.das@arm.com>
7308 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
7309 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
7310 re-computing once computed.
7311 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
7312 (arm_init_machine_status): Initialize
7313 machine->static_chain_stack_bytes.
7315 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7318 * doc/extend.texi: Add four new prototypes for vec_ld.
7319 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
7320 definitions for more logical presentation.
7321 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
7322 entries for V1TI variants of __builtin_altivec_ld builtin.
7323 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
7324 handling of V1TI variant of LVX icode pattern.
7325 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
7326 (rs6000_gimple_fold_builtin): Likewise.
7327 (altivec_init_builtins): Add code to define
7328 __builtin_altivec_lvx_v1ti function.
7330 2018-03-22 Jakub Jelinek <jakub@redhat.com>
7333 * function.c (match_asm_constraints_1): Don't do the optimization
7334 if input isn't a REG, SUBREG, MEM or constant.
7336 2018-03-22 Tom de Vries <tom@codesourcery.com>
7338 PR tree-optimization/84956
7339 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
7340 bb_has_abnormal_pred.
7342 2018-03-22 Jakub Jelinek <jakub@redhat.com>
7345 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
7346 DECL_INITIAL (decl) to decl at the end.
7347 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
7350 2018-03-21 Joseph Myers <joseph@codesourcery.com>
7352 * doc/extend.texi (__builtin_tgmath): Document when complex
7353 integer types are treated as _Complex _Float64.
7355 2018-03-21 Tom de Vries <tom@codesourcery.com>
7357 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
7359 2018-03-21 Jakub Jelinek <jakub@redhat.com>
7361 PR tree-optimization/84960
7362 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
7363 if it is ENTRY block, move them into single succ of ENTRY in that case.
7365 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
7367 PR tree-optimization/84811
7368 * poly-int.h (poly_span_traits): Remove the T3 parameter and
7369 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
7370 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
7371 (known_subrange_p): Update accordingly. Cast each value involved
7372 in the size comparison, rather than casting the result of the
7375 2018-03-21 Jakub Jelinek <jakub@redhat.com>
7377 PR tree-optimization/84982
7378 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
7379 by flipping the least significant bit rather than all bits from
7380 bitpos to bitpos + bitsize - 1.
7382 2018-03-21 Nathan Sidwell <nathan@acm.org>
7384 * doc/extend.texi (Deprecated Features): Remove mention of
7385 long-deleted deprecations.
7387 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7390 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
7391 * configure: Regenerate.
7393 2018-03-21 Tom de Vries <tom@codesourcery.com>
7395 PR tree-optimization/83126
7396 * tree-parloops.c (num_phis): New function.
7397 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
7399 2018-03-21 Nathan Sidwell <nathan@acm.org>
7401 * doc/extend.texi (Deprecated Features): Update deprecated flags,
7402 mention anon-struct/union members and trailing attributes.
7404 2018-03-21 Bin Cheng <bin.cheng@arm.com>
7406 PR tree-optimization/84969
7407 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
7408 builtin memset partitions if they set different rhs values.
7410 2018-03-21 Jakub Jelinek <jakub@redhat.com>
7412 PR rtl-optimization/84989
7413 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
7414 VEC_DUPLICATE with scalar result mode.
7416 2018-03-21 Martin Liska <mliska@suse.cz>
7419 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
7420 not intended return statement.
7422 2018-03-21 Martin Liska <mliska@suse.cz>
7425 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
7426 (chkp_find_bound_slots_1): Limit number of iterations.
7428 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
7431 * Minor grammar fixes for x86 options.
7433 2018-03-20 Jakub Jelinek <jakub@redhat.com>
7436 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
7437 holding REG_CFA_RESTORE notes, instead turn them into a USE.
7439 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
7442 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
7443 (altivec_lvx_<mode>_1op): Likewise.
7444 (altivec_stvx_<mode>_2op): Likewise.
7445 (altivec_stvx_<mode>_1op): Likewise.
7446 (altivec_lvx_<VM2:mode>): New define_expand.
7447 (altivec_stvx_<VM2:mode>): Likewise.
7448 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
7449 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
7450 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
7451 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
7452 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
7453 (rs6000_gen_lvx): Likewise.
7454 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
7455 (altivec_expand_stv_builtin): Likewise.
7456 (altivec_expand_builtin): Likewise.
7457 * config/rs6000/vector.md: Likewise.
7459 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7462 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
7465 2018-03-20 Richard Biener <rguenther@suse.de>
7468 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
7469 sign-conversions as zero, fall back to standard scalar_stmt
7472 2018-03-20 Martin Liska <mliska@suse.cz>
7475 * predict.c (rebuild_frequencies): Handle case when we have
7476 PROFILE_ABSENT, but flag_guess_branch_prob is false.
7478 2018-03-20 Jakub Jelinek <jakub@redhat.com>
7481 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
7482 flag_section_anchors.
7483 * varasm.c (use_blocks_for_decl_p): Remove hack for
7484 dw2_force_const_mem.
7487 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
7489 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
7490 be created, use lowpart_subreg of operands[0] rather than operands[0]
7492 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
7493 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
7494 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
7495 and n constraint instead of aarch64_shift_imm_di and Usd.
7496 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
7497 (*aarch64_<optab>_reg_minus<mode>3): ... this.
7499 2018-03-20 Sudakshina Das <sudi.das@arm.com>
7502 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
7503 to favor GPR over NEON registers.
7504 (<shift>di3_neon): Likewise.
7506 2018-03-20 Tom de Vries <tom@codesourcery.com>
7509 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
7510 (nvptx_process_pars): Emit bar.sync asap and alap.
7512 2018-03-20 Tom de Vries <tom@codesourcery.com>
7515 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
7516 seen_label if seen_label is already set.
7518 2018-03-20 Jakub Jelinek <jakub@redhat.com>
7521 * config/i386/i386.c (fold_builtin_cpu): For features above 31
7522 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
7523 Use 1U instead of 1. Formatting fixes.
7526 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
7527 instead of TREE_TYPE (s1) for the return value.
7529 2018-03-19 Jakub Jelinek <jakub@redhat.com>
7531 PR tree-optimization/84946
7532 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
7533 bitsize + bitsize in poly_uint64 rather than poly_int64.
7536 * dwarf2asm.c: Include fold-const.c.
7537 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
7538 of decl rather than decl itself.
7540 PR rtl-optimization/84643
7541 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
7543 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
7546 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
7547 calling assemble_variable.
7549 2018-03-19 Sudakshina Das <sudi.das@arm.com>
7552 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
7553 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
7555 2018-03-19 Jim Wilson <jimw@sifive.com>
7558 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
7559 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
7560 (riscv_first_stack_step): Likewise.
7561 (riscv_option_override): Use STACK_BOUNDARY instead of
7563 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
7565 (BIGGEST_ALIGNMENT): Set to 128.
7566 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
7567 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
7570 2018-03-19 Richard Biener <rguenther@suse.de>
7572 PR tree-optimization/84933
7573 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
7574 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
7576 2018-03-19 Richard Biener <rguenther@suse.de>
7578 PR tree-optimization/84859
7579 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
7580 (cond_if_else_store_replacement): Perform sinking operation on
7581 single-store BBs regardless of MAX_STORES_TO_SINK setting.
7582 Generalize what a BB with a single eligible store is.
7584 2018-03-19 Richard Biener <rguenther@suse.de>
7586 PR tree-optimization/84929
7587 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
7588 chrec_is_positive against non-chrec arg.
7590 2018-03-19 Tamar Christina <tamar.christina@arm.com>
7593 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
7595 2018-03-18 Martin Liska <mliska@suse.cz>
7597 PR rtl-optimization/84635
7598 * regrename.c (build_def_use): Use matches_mode only when
7601 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
7603 PR tree-optimization/84913
7604 * tree-vect-loop.c (vectorizable_reduction): Don't try to
7605 vectorize chains of COND_EXPRs.
7607 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
7609 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
7611 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
7613 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
7615 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
7617 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
7619 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
7620 Kito Cheng <kito.cheng@gmail.com>
7622 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
7623 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
7624 (nds32_adjust_reg_alloc_order): New function.
7625 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
7627 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
7629 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
7630 nds32_print_operand, nds32_print_operand_address): Use
7631 HOST_WIDE_INT_PRINT_DEC instead.
7633 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
7635 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
7637 2018-03-17 Jakub Jelinek <jakub@redhat.com>
7640 * config/i386/i386.c (initial_ix86_tune_features,
7641 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
7643 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
7644 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
7645 rather than 1u << ix86_tune. Formatting fix.
7646 (ix86_option_override_internal): Change ix86_arch_mask from
7647 unsigned int to unsigned HOST_WIDE_INT, initialize to
7648 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
7649 (ix86_function_specific_restore): Likewise.
7651 2018-03-16 Jakub Jelinek <jakub@redhat.com>
7654 * postreload.c (reload_combine_recognize_pattern): Perform
7655 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
7656 truncate_int_for_mode the result for the destination's mode.
7659 * hsa-gen.c (mem_type_for_type): Fix comment typo.
7660 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
7662 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
7665 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
7668 * lra-assigns.c (lra_split_hard_reg_for): Don't use
7669 regno_allocno_class_array and sorted_pseudos.
7670 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
7671 insns where regno is used.
7673 2018-03-16 Martin Liska <mliska@suse.cz>
7676 * multiple_target.c (create_dispatcher_calls): Redirect
7677 reference in the symbol table.
7679 2018-03-16 Martin Liska <mliska@suse.cz>
7682 * multiple_target.c (create_dispatcher_calls): Redirect also
7685 2018-03-16 Jakub Jelinek <jakub@redhat.com>
7689 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
7690 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
7691 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
7693 2018-03-16 Julia Koval <julia.koval@intel.com>
7695 * doc/invoke.texi (Skylake Server): Add CLWB.
7696 Cannonlake): Remove CLWB.
7698 2018-03-16 Jakub Jelinek <jakub@redhat.com>
7700 PR tree-optimization/84841
7701 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
7703 (FLOAT_ONE_CONST_TYPE): Define.
7704 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
7705 (sort_by_operand_rank): Put entries with higher constant_type last
7706 rather than first to match comments.
7708 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
7710 * config/nios2/nios2.md (movsi_internal): Fix thinko in
7713 2018-03-15 Jakub Jelinek <jakub@redhat.com>
7716 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
7717 check and use address of target always.
7719 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
7722 * config/i386/i386.c (indirect_thunk_needed): Update comments.
7723 (indirect_thunk_bnd_needed): Likewise.
7724 (indirect_thunks_used): Likewise.
7725 (indirect_thunks_bnd_used): Likewise.
7726 (indirect_return_needed): New.
7727 (indirect_return_bnd_needed): Likewise.
7728 (output_indirect_thunk_function): Add a bool argument for
7730 (output_indirect_thunk_function): Don't generate alias for
7731 function return thunk.
7732 (ix86_code_end): Call output_indirect_thunk_function to generate
7733 function return thunks.
7734 (ix86_output_function_return): Set indirect_return_bnd_needed
7735 and indirect_return_needed instead of indirect_thunk_bnd_needed
7736 and indirect_thunk_needed.
7738 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
7740 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
7741 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
7742 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
7744 2018-03-15 David Malcolm <dmalcolm@redhat.com>
7745 Paul Hua <paul.hua.gm@gmail.com>
7748 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
7750 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
7752 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
7753 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
7756 2018-03-15 Tamar Christina <tamar.christina@arm.com>
7759 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
7760 instead of GET_MODE_SIZE when comparing Units.
7762 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
7765 * varasm.c (hash_section): Return an unchangeble hash value
7766 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
7767 Return !aarch64_can_use_per_function_literal_pools_p ().
7769 2018-03-15 Jakub Jelinek <jakub@redhat.com>
7772 * optabs.c (emit_conditional_move): Pass address of cmode's copy
7773 rather than address of cmode as last argument to prepare_cmp_insn.
7775 2018-03-15 Julia Koval <julia.koval@intel.com>
7777 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
7778 F_AVX512VNNI, F_AVX512BITALG): New.
7780 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
7783 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
7784 insn for floating-point loads and stores.
7786 2018-03-14 Carl Love <cel@us.ibm.com>
7788 * config/rs6000/rs6000-c.c: Add macro definitions for
7789 ALTIVEC_BUILTIN_VEC_PERMXOR.
7790 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
7791 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
7792 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
7793 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
7795 * config/doc/extend.texi: Add prototypes for vec_permxor.
7797 2018-03-14 David Malcolm <dmalcolm@redhat.com>
7800 * diagnostic-show-locus.c (class layout_point): Convert m_line
7801 from int to linenum_type.
7802 (line_span::comparator): Use linenum "compare" function when
7803 comparing line numbers.
7804 (test_line_span): New function.
7805 (layout_range::contains_point): Convert param "row" from int to
7807 (layout_range::intersects_line_p): Likewise.
7808 (layout::will_show_line_p): Likewise.
7809 (layout::print_source_line): Likewise.
7810 (layout::should_print_annotation_line_p): Likewise.
7811 (layout::print_annotation_line): Likewise.
7812 (layout::print_leading_fixits): Likewise.
7813 (layout::annotation_line_showed_range_p): Likewise.
7814 (struct line_corrections): Likewise for field m_row.
7815 (line_corrections::line_corrections): Likewise for param "row".
7816 (layout::print_trailing_fixits): Likewise.
7817 (layout::get_state_at_point): Likewise.
7818 (layout::get_x_bound_for_row): Likewise.
7819 (layout::print_line): Likewise.
7820 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
7821 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
7822 * input.c (selftest::test_linenum_comparisons): New function.
7823 (selftest::input_c_tests): Call it.
7824 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
7825 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
7826 * selftest.h (ASSERT_GT): New macro.
7827 (ASSERT_GT_AT): New macro.
7828 (ASSERT_LT): New macro.
7829 (ASSERT_LT_AT): New macro.
7831 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
7833 PR rtl-optimization/84780
7834 * combine.c (distribute_links): Don't make a link based on pc_rtx.
7836 2018-03-14 Martin Liska <mliska@suse.cz>
7838 * tree.c (record_node_allocation_statistics): Use
7839 get_stats_node_kind.
7840 (get_stats_node_kind): New function extracted from
7841 record_node_allocation_statistics.
7842 (free_node): Use get_stats_node_kind.
7844 2018-03-14 Richard Biener <rguenther@suse.de>
7846 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
7847 that the value-set of ANTIC_IN doesn't grow.
7850 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
7852 (BB_VISITED_WITH_VISITED_SUCCS): New define.
7853 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
7855 2018-03-14 Julia Koval <julia.koval@intel.com>
7857 * config.gcc (icelake-client, icelake-server): New.
7859 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
7860 (initial_ix86_arch_features): Ditto.
7861 (PTA_SKYLAKE): Add SGX.
7862 (PTA_ICELAKE): Remove.
7863 (PTA_ICELAKE_CLIENT): New.
7864 (PTA_ICELAKE_SERVER): New.
7865 (ix86_option_override_internal): Split up icelake on icelake client and
7867 (get_builtin_code_for_version): Ditto.
7868 (fold_builtin_cpu): Ditto.
7869 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
7870 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
7871 * config/i386/i386.h (processor_type): Ditto.
7872 * doc/invoke.texi: Ditto.
7874 2018-03-14 Jakub Jelinek <jakub@redhat.com>
7877 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
7878 INTEGER_CST offset, add it together with bitpos / 8 and
7879 sign extend based on POINTER_SIZE.
7883 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
7886 * config/i386/constraints.md (Yc): New register constraint.
7887 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
7888 Use Yc constraint for alternative 2 of operand 0. Remove
7889 preferred_for_speed attribute.
7891 2018-03-14 Richard Biener <rguenther@suse.de>
7893 PR tree-optimization/84830
7894 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
7895 with the old one to avoid oscillations.
7897 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
7900 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
7902 (assign_by_spills): Return a flag of reload assignment failure.
7903 Do not process the reload assignment failures. Do not spill other
7904 reload pseudos if they has the same reg class. Update n if
7906 (lra_assign): Add a return arg. Set up from the result of
7907 assign_by_spills call.
7908 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
7909 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
7910 usage_insns if it is not NULL.
7911 (spill_hard_reg_in_range): New function.
7912 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
7913 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
7914 function prototypes.
7915 (lra_assign): Change prototype.
7916 * lra.c (lra): Add code to deal with fails by splitting hard reg
7919 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
7921 * config/riscv/riscv.opt (mrelax): New option.
7922 * config/riscv/riscv.c (riscv_file_start): Emit ".option
7923 "norelax" when riscv_mrelax is disabled.
7924 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
7926 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7929 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
7930 reassociation for int modes.
7932 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
7934 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
7935 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
7937 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
7938 * config/aarch64/aarch64-sve.md
7939 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
7940 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
7941 (*extend<mode><Vwide>2): Rename to...
7942 (aarch64_sve_extend<mode><Vwide>2): ...this.
7943 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
7944 renaming the old pattern to...
7945 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
7947 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
7948 define_expand, renaming the old pattern to...
7949 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
7950 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
7951 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
7952 account when deciding which SVE instruction the optab should use.
7953 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
7955 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
7957 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
7958 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
7959 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
7960 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
7961 (tlsdesc_small_<mode>): Turn a define_expand and use
7962 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
7963 (tlsdesc_small_advsimd_<mode>): ...this.
7964 (tlsdesc_small_sve_<mode>): New pattern.
7966 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
7968 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
7969 (UNSPEC_UMUL_HIGHPART): New constants.
7970 (MUL_HIGHPART): New int iteraor.
7971 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
7972 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
7974 (*<su>mul<mode>3_highpart): New define_insn.
7976 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
7979 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
7982 2018-03-13 Martin Liska <mliska@suse.cz>
7985 * (sem_item_optimizer::sem_item_optimizer): Initialize new
7987 (sem_item_optimizer::~sem_item_optimizer): Release it.
7988 (sem_item_optimizer::merge_classes): Register variable aliases.
7989 (sem_item_optimizer::fixup_pt_set): New function.
7990 (sem_item_optimizer::fixup_points_to_sets): Likewise.
7991 * ipa-icf.h: Declare new variables and functions.
7993 2018-03-13 Jakub Jelinek <jakub@redhat.com>
7996 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
7997 integer_pow2p@2 and test integer_pow2p in condition.
7998 (A < 0 ? C : 0): Similarly for @1.
8001 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
8002 characters starting at p contain '\0' character, don't look beyond
8006 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
8007 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
8010 * reg-stack.c (change_stack): Change update_end var from int to
8011 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
8012 also call set_block_for_insn on the newly added insns and rescan.
8015 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
8016 on the last operand.
8019 * tree.c (stabilize_reference_1): Return save_expr (e) for
8020 STATEMENT_LIST even if it doesn't have side-effects.
8022 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
8024 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
8026 2018-03-12 Renlin Li <renlin.li@arm.com>
8028 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
8029 aarch64_output_scalar_simd_mov_immediate.
8031 2018-03-12 Martin Sebor <msebor@redhat.com>
8033 PR tree-optimization/83456
8034 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
8035 for perfectly overlapping calls to memcpy.
8036 (gimple_fold_builtin_memory_chk): Same.
8037 (gimple_fold_builtin_strcpy): Handle no-warning.
8038 (gimple_fold_builtin_stxcpy_chk): Same.
8039 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
8041 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
8043 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
8044 parameter. Use it for SFmode.
8045 (rs6000_function_arg_advance_1): Adjust.
8046 (rs6000_function_arg): Adjust.
8047 (rs6000_gimplify_va_arg): Pass false for that new parameter.
8049 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
8051 PR rtl-optimization/84169
8052 PR rtl-optimization/84780
8053 * combine.c (can_combine_p): Check for a 2-insn combination whether
8054 the destination register is used between the two insns, too.
8056 2018-03-12 Richard Biener <rguenther@suse.de>
8058 PR tree-optimization/84803
8059 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
8060 for refs DR analysis didn't process.
8062 2018-03-12 Richard Biener <rguenther@suse.de>
8064 PR tree-optimization/84777
8065 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
8066 force-vectorize loops ignore whether we are optimizing for size.
8068 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
8070 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
8071 (TARGET_MD_ASM_ADJUST): Define.
8073 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
8074 Kito Cheng <kito.cheng@gmail.com>
8075 Chung-Ju Wu <jasonwucj@gmail.com>
8077 * config/nds32/nds32.c (nds32_compute_stack_frame,
8078 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
8079 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
8080 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
8081 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
8082 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
8083 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
8084 * config/nds32/nds32.md (prologue, epilogue): Use macro
8085 NDS32_V3PUSH_AVAILABLE_P to do checking.
8087 2018-03-11 Jakub Jelinek <jakub@redhat.com>
8090 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
8091 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
8092 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
8093 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
8094 addition of most attributes on !orig_type_die or the attribute not
8095 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
8097 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
8098 Chung-Ju Wu <jasonwucj@gmail.com>
8100 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
8102 * config/nds32/nds32.opt (mvh): New option.
8104 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
8105 Chung-Ju Wu <jasonwucj@gmail.com>
8107 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
8109 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
8110 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
8113 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
8114 Chung-Ju Wu <jasonwucj@gmail.com>
8116 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
8118 * config/nds32/nds32-multiple.md (strlensi): New pattern.
8119 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
8121 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
8122 Kito Cheng <kito.cheng@gmail.com>
8123 Chung-Ju Wu <jasonwucj@gmail.com>
8125 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
8126 UNSPEC_FFMISM and UNSPEC_FLMISM.
8127 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
8128 for ffb, ffmism and flmism.
8129 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
8130 (unspec_ffmism): Ditto.
8131 (unspec_flmism): Ditto.
8132 (nds32_expand_builtin_impl): Check if string extension is available.
8133 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
8134 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
8136 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
8139 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
8142 * lra-assigns.c (assign_by_spills): Return a flag of reload
8143 assignment failure. Do not process the reload assignment
8144 failures. Do not spill other reload pseudos if they has the same
8146 (lra_assign): Add a return arg. Set up from the result of
8147 assign_by_spills call.
8148 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
8149 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
8150 usage_insns if it is not NULL.
8151 (spill_hard_reg_in_range): New function.
8152 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
8153 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
8154 function prototypes.
8155 (lra_assign): Change prototype.
8156 * lra.c (lra): Add code to deal with fails by splitting hard reg
8159 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
8162 * config/i386/i386.opt: Replace Enforcment with Enforcement.
8164 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
8167 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
8168 (dw_val_node): Add val_symbolic_view.
8169 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
8170 (symview_upper_bound): New.
8171 (new_line_info_table): Initialize symviews_since_reset.
8172 (dwarf2out_source_line): Count symviews_since_reset and set
8173 symview_upper_bound.
8174 (dw_val_equal_p): Handle symview.
8175 (add_AT_symview): New.
8176 (print_dw_val): Handle symview.
8177 (attr_checksum, attr_checksum_ordered): Likewise.
8178 (same_dw_val_p, size_of_die): Likewise.
8179 (value_format, output_die): Likewise.
8180 (add_high_low_attributes): Use add_AT_symview for entry_view.
8181 (dwarf2out_finish): Reset symview_upper_bound, clear
8184 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
8187 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
8188 Add strict argument and use it.
8189 (rs6000_split_multireg_move): Update for new strict argument.
8190 (mem_operand_gpr): Disallow all non-offsettable addresses.
8191 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
8193 2018-03-09 Jakub Jelinek <jakub@redhat.com>
8196 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
8197 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
8198 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
8201 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
8202 decl, use remap_type if we want to use the type.
8204 2018-03-09 Martin Sebor <msebor@redhat.com>
8206 PR tree-optimization/84526
8207 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
8209 (builtin_access::generic_overlap): Be prepared to handle non-array
8212 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
8214 PR rtl-optimization/84682
8215 * lra-constraints.c (process_address_1): Check is_address flag
8216 for address constraints.
8217 (process_alt_operands): Likewise.
8218 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
8219 preprocess_constraints.
8220 * recog.h (preprocess_constraints): Add oploc parameter.
8222 * recog.c (preprocess_constraints): Test address_operand for
8223 CT_ADDRESS constraints.
8225 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
8228 * lra-assigns.c (assign_by_spills): Return a flag of reload
8229 assignment failure. Do not process the reload assignment
8230 failures. Do not spill other reload pseudos if they has the same
8232 (lra_assign): Add a return arg. Set up from the result of
8233 assign_by_spills call.
8234 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
8235 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
8236 usage_insns if it is not NULL.
8237 (spill_hard_reg_in_range): New function.
8238 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
8239 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
8240 function prototypes.
8241 (lra_assign): Change prototype.
8242 * lra.c (lra): Add code to deal with fails by splitting hard reg
8245 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8248 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
8249 Accept complain bool parameter. Only emit errors if it is true.
8250 (arm_parse_cpu_option_name): Likewise.
8251 (arm_target_thumb_only): Adjust callers of the above.
8252 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
8253 prototype to take a default true bool parameter.
8254 (arm_parse_arch_option_name): Likewise.
8256 2018-03-09 David Malcolm <dmalcolm@redhat.com>
8257 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
8261 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
8262 * configure: Regenerate.
8263 * configure.ac ("linker --version-script option"): New.
8264 ("linker soname option"): New.
8266 2018-03-09 Richard Biener <rguenther@suse.de>
8268 PR tree-optimization/84775
8269 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
8270 immediate uses of predicate stmts and mark them modified.
8273 PR tree-optimization/84178
8274 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
8276 (version_loop_for_if_conversion): Delay update_ssa call.
8277 (tree_if_conversion): Delay update_ssa until after predicate
8280 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
8283 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
8284 when the function accesses prior frames.
8286 2018-03-08 Jakub Jelinek <jakub@redhat.com>
8289 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
8290 gen_llsym, otherwise call maybe_gen_llsym.
8293 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
8294 has ',' character inside of it.
8296 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8299 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
8300 as clobbering CC_REGNUM.
8302 2018-03-08 Richard Biener <rguenther@suse.de>
8305 * tree-scalar-evolution.c: Include tree-into-ssa.h.
8306 (follow_copies_to_constant): Do not follow SSA names registered
8309 2018-03-08 Richard Biener <rguenther@suse.de>
8311 PR tree-optimization/84178
8312 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
8314 (version_loop_for_if_conversion): Delay update_ssa call.
8315 (tree_if_conversion): Delay update_ssa until after predicate
8318 2018-03-08 David Malcolm <dmalcolm@redhat.com>
8320 PR tree-optimization/84178
8321 * tree-if-conv.c (release_bb_predicate): Remove the
8322 the assertion that the stmts have NULL use_ops.
8323 Discard the statements, asserting that they haven't
8324 yet been added to a BB.
8326 2018-03-08 Richard Biener <rguenther@suse.de>
8328 PR tree-optimization/84746
8329 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
8330 (phi_translate): Pass in destination ANTIC_OUT set.
8331 (phi_translate_1): Likewise. For a simplified result lookup
8332 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
8333 (phi_translate_set): Adjust.
8334 (do_pre_regular_insertion): Likewise.
8335 (do_pre_partial_partial_insertion): Likewise.
8337 2018-03-08 Martin Liska <mliska@suse.cz>
8339 PR gcov-profile/84735
8340 * doc/gcov.texi: Document usage of profile files.
8341 * gcov-io.h: Document changes in the format.
8343 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
8347 * dwarf2out.c (struct dw_line_info_table): Update comments for
8349 (FORCE_RESET_NEXT_VIEW): New.
8350 (FORCE_RESETTING_VIEW_P): New.
8351 (RESETTING_VIEW_P): Check for -1 too.
8352 (ZERO_VIEW_P): Likewise.
8353 (new_line_info_table): Force-reset next view.
8354 (dwarf2out_begin_function): Likewise.
8355 (dwarf2out_source_line): Simplify zero_view_p initialization.
8356 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
8357 view directly. Omit view when omitting .loc at line 0.
8359 2018-03-08 Jakub Jelinek <jakub@redhat.com>
8361 PR tree-optimization/84740
8362 * tree-switch-conversion.c (process_switch): Call build_constructors
8363 only if info.phi_count is non-zero.
8365 PR tree-optimization/84739
8366 * tree-tailcall.c (find_tail_calls): Check call arguments against
8367 DECL_ARGUMENTS (current_function_decl) rather than
8368 DECL_ARGUMENTS (func) when checking for tail recursion.
8370 2018-03-07 Jakub Jelinek <jakub@redhat.com>
8372 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
8373 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
8374 Volker Reichelt's entry and add entries for people that perform
8375 GCC fuzzy testing and report numerous bugs.
8377 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
8380 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
8381 readonly data in sdata, if that is disabled.
8382 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
8383 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
8384 -mreadonly-in-sdata option.
8386 2018-03-07 Martin Sebor <msebor@redhat.com>
8388 PR tree-optimization/84468
8389 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
8390 basic block when looking for nul assignment.
8392 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
8395 * except.h (output_function_exception_table): Adjust prototype.
8396 * except.c (output_function_exception_table): Remove FNNAME parameter
8397 and add SECTION parameter. Ouput one part of the table at a time.
8398 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
8399 the first part of the exception table and emit unwind directives.
8400 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
8401 (i386_pe_seh_cold_init): Likewise.
8402 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
8403 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
8404 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
8405 (ix86_output_call_insn): Emit a nop in one more case for SEH.
8406 * config/i386/winnt.c: Include except.h.
8407 (struct seh_frame_state): Add reg_offset, after_prologue and
8408 in_cold_section fields.
8409 (i386_pe_seh_end_prologue): Set seh->after_prologue.
8410 (i386_pe_seh_cold_init): New function.
8411 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
8412 to seh->in_cold_section.
8413 (seh_emit_push): Record the offset of the push.
8414 (seh_emit_save): Record the offet of the save.
8415 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
8416 Test seh->after_prologue to disregard the epilogue.
8417 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
8418 (i386_pe_end_cold_function): New function.
8420 2018-03-07 Jakub Jelinek <jakub@redhat.com>
8423 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
8424 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
8427 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
8429 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
8430 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
8433 * multiple_target.c: Include tree-inline.h and intl.h.
8434 (expand_target_clones): Diagnose and fail if node->definition and
8435 !tree_versionable_function_p (node->decl).
8437 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
8439 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
8441 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
8442 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8443 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
8445 2018-03-06 Jakub Jelinek <jakub@redhat.com>
8448 * combine.c (try_combine): Use reg_or_subregno instead of handling
8449 just paradoxical SUBREGs and REGs.
8451 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
8453 * config/arc/arc.c (arc_finalize_pic): Remove function.
8454 (arc_must_save_register): We use single base PIC register, remove
8455 checks to save/restore the PIC register.
8456 (arc_expand_prologue): Likewise.
8457 * config/arc/arc-protos.h (arc_set_default_type_attributes):
8459 (arc_verify_short): Likewise.
8460 (arc_attr_type): Likewise.
8461 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
8462 (walk_stores): Likewise.
8463 (arc_address_cost): Make it static.
8464 (arc_verify_short): Likewise.
8465 (branch_dest): Likewise.
8466 (arc_attr_type): Likewise.
8467 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
8468 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
8469 (arc_final_prescan_insn): Remove inserting the nops due to
8470 hardware hazards. It is done in reorg step.
8471 (insn_length_variant_t): Remove.
8472 (insn_length_parameters_t): Likewise.
8473 (arc_insn_length_parameters): Likewise.
8474 (arc_get_insn_variants): Likewise.
8475 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
8477 2018-03-06 Jakub Jelinek <jakub@redhat.com>
8480 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
8483 PR tree-optimization/84687
8484 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
8486 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
8488 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8490 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
8491 Rename to ppc_speculation_barrier.
8492 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
8493 __builtin_ppc_speculation_barrier.
8495 2018-03-05 Jakub Jelinek <jakub@redhat.com>
8498 * combine.c (combine_simplify_rtx): Don't try to simplify if
8499 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
8502 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
8504 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
8505 to 32 bytes when compiling for POWER9.
8507 2018-03-05 Jakub Jelinek <jakub@redhat.com>
8510 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
8511 regparm >= 3 with no arg reg available also for calls with
8512 flag_force_indirect_call. Pass decl to ix86_function_regparm.
8515 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
8517 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
8519 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
8522 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
8524 2018-03-05 Richard Biener <rguenther@suse.de>
8526 PR tree-optimization/84486
8527 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
8528 When inserting a __builtin_assume_aligned call set the LHS
8529 SSA name alignment info accordingly.
8531 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
8533 PR tree-optimization/84114
8534 * config/aarch64/aarch64.c (aarch64_reassociation_width)
8535 Avoid reassociation of FLOAT_MODE addition.
8537 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
8539 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
8540 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
8541 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
8542 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
8543 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
8544 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
8545 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
8547 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
8548 __builtin_ia32_wbinvd): New builtins.
8549 (SPECIAL_ARGS2): New.
8550 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
8551 (SPECIAL_ARGS2): New.
8552 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
8553 (ix86_valid_target_attribute_inner_p): Ditto.
8554 (ix86_init_mmx_sse_builtins): Add special_args2.
8555 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
8556 TARGET_WBNOINVD_P): New.
8557 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
8558 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
8559 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
8560 * config/i386/immintrin.h (_wbinvd): New intrinsic.
8561 * config/i386/pconfigintrin.h: New file.
8562 * config/i386/wbnoinvdintrin.h: Ditto.
8563 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
8565 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
8567 2018-03-05 Richard Biener <rguenther@suse.de>
8569 PR tree-optimization/84670
8570 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
8572 (BB_VISITED_WITH_VISITED_SUCCS): New define.
8573 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
8574 (compute_antic_aux): Only assert the number of values in ANTIC_IN
8575 doesn't grow if all successors (recursively) were visited at least
8578 2018-03-05 Richard Biener <rguenther@suse.de>
8580 PR tree-optimization/84650
8581 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
8582 if executed in the loop pipeline.
8584 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
8586 * doc/configfiles.texi (Configuration Files): Move info about
8587 conditionalizing $target-protos.h to...
8588 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
8589 differs from $target-protos.h.
8591 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
8592 Chung-Ju Wu <jasonwucj@gmail.com>
8594 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
8595 * config/nds32/nds32-multiple.md (setmemsi): Define.
8596 * config/nds32/nds32-memory-manipulation.c
8597 (nds32_gen_dup_4_byte_to_word_value): New.
8598 (emit_setmem_word_loop): New.
8599 (emit_setmem_byte_loop): New.
8600 (nds32_expand_setmem_loop): New.
8601 (nds32_expand_setmem_loop_v3m): New.
8602 (nds32_expand_setmem_unroll): New.
8603 (nds32_expand_setmem): New.
8605 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
8606 Chung-Ju Wu <jasonwucj@gmail.com>
8608 * config/nds32/nds32-memory-manipulation.c
8609 (nds32_emit_load_store): New.
8610 (nds32_emit_post_inc_load_store): New.
8611 (nds32_emit_mem_move): New.
8612 (nds32_emit_mem_move_block): New.
8613 (nds32_expand_movmemsi_loop_unknown_size): New.
8614 (nds32_expand_movmemsi_loop_known_size): New.
8615 (nds32_expand_movmemsi_loop): New.
8616 (nds32_expand_movmemsi_unroll): New.
8617 (nds32_expand_movmemqi): Rename ...
8618 (nds32_expand_movmemsi): ... to this.
8619 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
8620 (movmemsi): ... to this.
8621 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
8622 (nds32_expand_movmemsi): ... to this.
8624 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
8625 Monk Chiang <sh.chiang04@gmail.com>
8626 Chung-Ju Wu <jasonwucj@gmail.com>
8628 * config/nds32/nds32-protos.h
8629 (nds32_expand_load_multiple): New arguments.
8630 (nds32_expand_store_multiple): Ditto.
8631 (nds32_valid_multiple_load_store): Rename ...
8632 (nds32_valid_multiple_load_store_p): ... to this.
8633 * config/nds32/nds32-memory-manipulation.c
8634 (nds32_expand_load_multiple): Refine implementation.
8635 (nds32_expand_store_multiple): Ditto.
8636 * config/nds32/nds32-multiple.md
8637 (load_multiple): Update nds32_expand_load_multiple interface.
8638 (store_multiple): Update nds32_expand_store_multiple interface.
8639 * config/nds32/nds32-predicates.c
8640 (nds32_valid_multiple_load_store): Rename ...
8641 (nds32_valid_multiple_load_store_p): ... to this and refine
8643 * config/nds32/predicates.md
8644 (nds32_load_multiple_and_update_address_operation): New predicate.
8645 (nds32_store_multiple_and_update_address_operation): New predicate.
8647 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
8648 Chung-Ju Wu <jasonwucj@gmail.com>
8650 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
8651 (combo): New attribute.
8652 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
8654 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
8656 * config/nds32/nds32.opt: Change -mcmodel= default value.
8658 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
8659 Monk Chiang <sh.chiang04@gmail.com>
8660 Chung-Ju Wu <jasonwucj@gmail.com>
8662 * config/nds32/constants.md (unspec_element): New enum.
8663 * config/nds32/constraints.md (Umw): New constraint.
8664 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
8665 * config/nds32/nds32-intrinsic.md: Likewise.
8666 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
8667 (nds32_valid_smw_lwm_base_p): New.
8668 (nds32_output_smw_single_word): New.
8669 (nds32_output_lmw_single_word): New.
8670 (nds32_expand_unaligned_load): New.
8671 (nds32_expand_unaligned_store): New.
8672 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
8673 (nds32_output_smw_single_word): Declare.
8674 (nds32_output_lmw_single_word): Declare.
8675 (nds32_expand_unaligned_load): Declare.
8676 (nds32_expand_unaligned_store): Declare.
8677 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
8678 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
8679 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
8680 NDS32_BUILTIN_UASTORE_DW.
8681 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
8684 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
8685 Kito Cheng <kito.cheng@gmail.com>
8686 Chung-Ju Wu <jasonwucj@gmail.com>
8688 * config/nds32/nds32-intrinsic.c
8689 (nds32_expand_builtin_null_ftype_reg): Delete.
8690 (nds32_expand_builtin_reg_ftype_imm): Ditto.
8691 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
8692 (nds32_read_argument): New.
8693 (nds32_legitimize_target): Ditto.
8694 (nds32_legitimize_argument): Ditto.
8695 (nds32_check_constant_argument): Ditto.
8696 (nds32_expand_unop_builtin): Ditto.
8697 (nds32_expand_unopimm_builtin): Ditto.
8698 (nds32_expand_binop_builtin): Ditto.
8699 (nds32_builtin_decl_impl): Ditto.
8700 (builtin_description): Ditto.
8701 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
8702 (nds32_init_builtins_impl): Ditto.
8703 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
8704 (nds32_builtin_decl): New.
8705 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
8706 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
8708 2018-03-02 Jeff Law <law@redhat.com>
8710 * reorg.c (stop_search_p): Handle DEBUG_INSN.
8711 (redundant_insn, fill_simple_delay_slots): Likewise.
8712 (fill_slots_from_thread): Likewise.
8713 * resource.c (mark_referenced_resources): Likewise.
8714 (mark_set_resources, find_dead_or_set_registers): Likewise.
8716 2018-03-02 Jakub Jelinek <jakub@redhat.com>
8718 * substring-locations.h (format_warning_va): Formatting fix for
8720 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
8722 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
8723 * substring-locations.c: Include intl.h.
8724 (format_warning_va): Turned into small wrapper around
8725 format_warning_n_va, renamed to ...
8726 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
8727 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
8729 (format_warning_at_substring_n): New function.
8730 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
8731 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
8732 format_warning_at_substring with just a shorter name instead of
8733 const function pointer.
8734 (fmtwarn_n): New function.
8735 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
8736 appropriate, get rid of all the fmtstr temporaries, move conditionals
8737 with G_() wrapped string literals directly into fmtwarn arguments,
8738 cast dir.len to (int), formatting fixes.
8740 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
8742 * doc/invoke.texi: Remove "Cilk Plus" references.
8744 2018-03-02 Jakub Jelinek <jakub@redhat.com>
8745 Richard Biener <rguenther@suse.de>
8748 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
8749 for error or warning attributes if CALL_FROM_THUNK_P is set.
8752 2018-03-02 Jakub Jelinek <jakub@redhat.com>
8755 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
8756 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
8759 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
8760 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
8762 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
8763 instead of -1U in last predictors element's probability member.
8765 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
8768 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
8769 arguments if they are comparable.
8771 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
8773 PR tree-optimization/84634
8774 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
8775 masks and masked_loop_p with a single loop_masks, making sure it's
8776 null for bb vectorization.
8778 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
8780 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8781 (vect_analyze_data_ref_access): Use loop->safe_len rather than
8782 loop->force_vectorize to check whether there is no alias.
8784 2018-03-02 Jakub Jelinek <jakub@redhat.com>
8787 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
8789 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
8791 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
8792 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
8793 instead of a loop around prev_real_insn.
8794 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
8798 * config/i386/i386.c (ix86_print_operand): Use conditional
8799 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
8802 2018-03-02 Richard Biener <rguenther@suse.de>
8804 PR tree-optimization/84427
8805 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
8806 (bitmap_set_subtract_values): Rewrite to handle multiple
8809 (prune_clobbered_mems): Likewise.
8810 (phi_translate): Take edge instead of pred/phiblock.
8811 (phi_translate_1): Likewise.
8812 (phi_translate_set): Likewise. Insert all translated
8813 exprs for a value into the set, keeping possibly multiple
8814 expressions per value.
8815 (compute_antic_aux): Adjust for phi_translate changes.
8816 When intersecting union the expressions and prune those
8817 not in the final value set, keeping possibly multiple
8818 expressions per value. Do not use value-insertion
8819 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
8820 all expressions. Add verification that the value-sets
8821 only shrink during iteration.
8822 (compute_partial_antic_aux): Adjust for the phi_translate changes.
8823 (do_pre_regular_insertion): Likewise.
8824 (do_pre_partial_partial_insertion): Likewise.
8826 2018-03-02 Richard Biener <rguenther@suse.de>
8829 * config/darwin.c (saved_debug_info_level): New static global.
8830 (darwin_asm_lto_start): Disable debug info generation for LTO out.
8831 (darwin_asm_lto_end): Restore debug info generation settings.
8833 2018-03-01 Martin Liska <mliska@suse.cz>
8836 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8839 2018-03-01 Richard Biener <rguenther@suse.de>
8842 * dwarf2out.c (gen_variable_die): Properly handle late VLA
8843 type annotation with LTO when debug was disabled at compile-time.
8845 2018-03-01 Matthew Fortune <mfortune@gmail.com>
8847 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
8849 (mips_final_postscan_insn): Likewise.
8851 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
8853 PR rtl-optimization/84528
8854 * alias.c (init_alias_target): Add commentary.
8855 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
8856 a unique base value if the frame pointer is not eliminated
8857 to the stack pointer.
8859 2018-03-01 Tom de Vries <tom@codesourcery.com>
8861 PR rtl-optimization/83327
8862 * lra-int.h (hard_regs_spilled_into): Declare.
8863 * lra.c (hard_regs_spilled_into): Define.
8864 (init_reg_info): Init hard_regs_spilled_into.
8865 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
8866 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
8867 (process_bb_lives): Handle hard_regs_spilled_into.
8868 (lra_create_live_ranges_1): Before doing liveness propagation, clear
8869 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
8871 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
8873 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
8874 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
8875 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
8876 * config/rs6000/aix72.h: New file.
8878 2018-02-28 Jakub Jelinek <jakub@redhat.com>
8880 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
8881 instead of warning_at with conditional singular and plural messages
8885 * stor-layout.c (update_alignment_for_field): For
8886 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
8887 && !DECL_PACKED (field), do the alignment update, just use
8888 only desired_align instead of MAX (type_align, desired_align)
8890 (place_field): Don't do known_align < desired_align handling
8891 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
8892 is non-NULL, instead do it after rli->prev_field handling and
8893 only if not within a bitfield word. For DECL_PACKED (field)
8894 use type_align of BITS_PER_UNIT.
8896 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
8898 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
8899 superfluous parentheses and trailing spaces.
8901 2018-02-28 Richard Biener <rguenther@suse.de>
8903 PR tree-optimization/84584
8904 * graphite-scop-detection.c (scop_detection::add_scop): Discard
8905 SCoPs with fake exit edge.
8907 2018-02-28 Martin Liska <mliska@suse.cz>
8910 * timevar.c (timer::print): Fix format to properly print 100%
8913 2018-02-28 Richard Biener <rguenther@suse.de>
8916 * genmatch.c (capture_info::walk_match): Do not mark
8917 captured expressions without operands as expr_p given
8918 they act more like predicates and should be subject to
8919 "lost tail" side-effect preserving.
8921 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
8923 PR rtl-optimization/81611
8924 * auto-inc-dec.c (attempt_change): Move dead note from
8925 mem_insn if it's the next use of regno
8926 (find_address): Take address use of reg holding
8927 non-incremented value. Add parm to limit search to the named
8929 (merge_in_block): Attempt to use a mem insn that is the next
8930 use of the original regno.
8932 2018-02-27 Martin Sebor <msebor@redhat.com>
8935 * gcc/doc/invoke.texi (-Wmissing-attributes): New option.
8936 * gcc/print-tree.c (print_node): Handle DECL_UNINLINABLE.
8938 2018-02-27 Martin Sebor <msebor@redhat.com>
8940 PR translation/84207
8941 * diagnostic-core.h (warning_n, error_n, inform_n): Change
8942 n argument to unsigned HOST_WIDE_INT.
8943 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
8944 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
8945 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
8946 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
8948 2018-02-27 Richard Biener <rguenther@suse.de>
8950 PR tree-optimization/84512
8951 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
8952 Do not use the estimate returned from record_stmt_cost for
8953 the scalar iteration cost but sum properly using add_stmt_cost.
8955 2018-02-27 Richard Biener <rguenther@suse.de>
8957 PR tree-optimization/84466
8958 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8959 Adjust last change to less strictly validate use operands.
8961 2018-02-27 Martin Liska <mliska@suse.cz>
8963 PR gcov-profile/84548
8964 * gcov.c (process_file): Allow partial overlap and consider it
8965 also as group functions.
8966 (output_lines): Properly calculate range of lines for a group.
8968 2018-02-27 Martin Liska <mliska@suse.cz>
8970 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
8971 'ggc' suffixes. Change first column width.
8972 (timer::print): Fix formatting of the column.
8974 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
8976 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
8977 preserve inline entry blocks for the sake of debug inline
8978 entry point markers alone.
8979 (remove_unused_locals): Suggest in comments a better place to
8980 force the preservation of inline entry blocks that are
8981 otherwise unused, but do not preserve them.
8983 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
8985 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
8987 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
8990 * config/i386/constraints.md (Bs): Replace
8991 ix86_indirect_branch_register with
8992 TARGET_INDIRECT_BRANCH_REGISTER.
8994 * config/i386/i386.md (indirect_jump): Likewise.
8995 (tablejump): Likewise.
8996 (*sibcall_memory): Likewise.
8997 (*sibcall_value_memory): Likewise.
8998 Peepholes of indirect call and jump via memory: Likewise.
8999 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
9000 (*sibcall_value_GOT_32): Likewise.
9001 * config/i386/predicates.md (indirect_branch_operand): Likewise.
9002 (GOT_memory_operand): Likewise.
9003 (call_insn_operand): Likewise.
9004 (sibcall_insn_operand): Likewise.
9005 (GOT32_symbol_operand): Likewise.
9006 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
9008 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
9010 PR rtl-optimization/83496
9011 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
9012 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
9013 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
9014 redundant insn, if any.
9015 (relax_delay_slots): Likewise.
9016 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
9018 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
9020 PR tree-optimization/83965
9021 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
9022 that grouped statements are part of a reduction chain. Return
9023 true if the statement is not marked as a reduction itself but
9025 (vect_recog_dot_prod_pattern): Don't check whether the statement
9026 is part of a group here.
9027 (vect_recog_sad_pattern): Likewise.
9028 (vect_recog_widen_sum_pattern): Likewise.
9030 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
9033 * final.c (rest_of_clean_state): Also look for calls inside sequences.
9035 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
9038 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
9040 (ix86_output_indirect_function_return): New prototype.
9041 (ix86_split_simple_return_pop_internal): Likewise.
9042 * config/i386/i386.c (indirect_return_via_cx): New.
9043 (indirect_return_via_cx_bnd): Likewise.
9044 (indirect_thunk_name): Handle return va CX_REG.
9045 (output_indirect_thunk_function): Create alias for
9046 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
9047 (ix86_output_indirect_jmp): Remove the bool argument.
9048 (ix86_output_indirect_function_return): New function.
9049 (ix86_split_simple_return_pop_internal): Likewise.
9050 * config/i386/i386.md (*indirect_jump): Don't pass false
9051 to ix86_output_indirect_jmp.
9052 (*tablejump_1): Likewise.
9053 (simple_return_pop_internal): Change it to define_insn_and_split.
9054 Call ix86_split_simple_return_pop_internal to split it for
9056 (simple_return_indirect_internal): Call
9057 ix86_output_indirect_function_return instead of
9058 ix86_output_indirect_jmp.
9060 2018-02-26 Jakub Jelinek <jakub@redhat.com>
9063 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
9064 memset and value initialization afterwards.
9066 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
9068 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
9070 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9073 * common/config/aarch64/aarch64-common.c
9074 (aarch_option_optimization_table[]): Switch
9075 off fomit-frame-pointer
9077 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
9078 Chung-Ju Wu <jasonwucj@gmail.com>
9080 * config/nds32/nds32-multiple.md (load_multiple): Disallow
9082 (store_multiple): Ditto.
9084 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
9086 * config.gcc: Add --with-cpu support for nds32 target.
9087 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
9088 * config/nds32/nds32.opt: Add -mcpu= option.
9090 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
9092 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
9093 isel=yes): Warn for these deprecated options.
9095 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
9097 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
9098 ISA_2_5_MASKS_EMBEDDED.
9100 2018-02-23 Jakub Jelinek <jakub@redhat.com>
9102 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
9103 p->max as pointers rather than using iterative_hash_expr.
9105 2018-02-23 Carl Love <cel@us.ibm.com>
9107 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
9108 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
9110 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
9111 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
9112 P8V_BUILTIN_VEC_VUNSIGNED2.
9114 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
9117 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
9118 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
9120 * lra-constraints.c (curr_insn_transform): Set up
9121 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
9123 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
9124 * lra-eliminations.c (spill_pseudos): Ditto.
9125 (process_insn_for_elimination): Ditto.
9126 * lra-lives.c (reg_early_clobber_p): Use the new macros.
9127 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
9128 LRA_NON_CLOBBERED_ALT.
9130 2018-02-22 Martin Sebor <msebor@redhat.com>
9132 PR tree-optimization/84480
9133 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
9134 to maybe_diag_stxncpy_trunc. Call it.
9135 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
9136 from gimple_fold_builtin_strcpy. Print inlining stack.
9137 (handle_builtin_stxncpy): Print inlining stack.
9138 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
9140 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
9143 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
9144 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
9145 and -fcheck-pointer-bounds are used together.
9146 (indirect_thunk_prefix): New enum.
9147 (indirect_thunk_need_prefix): New function.
9148 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
9149 "_nt" instead of "_bnd" for NOTRACK prefix.
9150 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
9151 (output_indirect_thunk_function): Likewise.
9153 (ix86_code_end): Update output_indirect_thunk_function calls.
9154 (ix86_output_indirect_branch_via_reg): Replace
9155 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
9156 (ix86_output_indirect_branch_via_push): Likewise.
9157 (ix86_output_function_return): Likewise.
9158 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
9159 incompatible with -fcf-protection=branch and
9160 -fcheck-pointer-bounds.
9162 2018-02-22 Steve Ellcey <sellcey@cavium.com>
9165 * config/aarch64/aarch64.c (aarch64_print_address_internal):
9166 Change gcc_assert call to output_operand_lossage.
9168 2018-02-22 Steve Ellcey <sellcey@cavium.com>
9170 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
9172 2018-02-22 DJ Delorie <dj@redhat.com>
9173 Sebastian Perta <sebastian.perta@renesas.com>
9174 Oleg Endo <olegendo@gcc.gnu.org>
9176 * config/rx/rx.c (rx_rtx_costs): New function.
9177 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
9179 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
9181 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
9183 2018-02-22 Martin Liska <mliska@suse.cz>
9186 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
9187 Add "native" as a possible value.
9189 2018-02-22 Martin Liska <mliska@suse.cz>
9192 * config/i386/i386.c (ix86_option_override_internal):
9193 Add "native" as a possible value for -march and -mtune.
9195 2018-02-22 Jakub Jelinek <jakub@redhat.com>
9198 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
9199 to all type variants.
9201 PR tree-optimization/84503
9202 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
9203 width as info->bitpos + info->bitsize - start.
9204 (merged_store_group::merge_overlapping): Simplify width computation.
9205 (check_no_overlap): New function.
9206 (imm_store_chain_info::try_coalesce_bswap): Compute expected
9207 start + width and last_order of the group, fail if check_no_overlap
9209 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
9210 to group if check_no_overlap fails.
9212 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
9214 * config/rs6000/altivec.md: Delete contraint arguments to
9215 define_expand, define_split, and define_peephole2, and in
9216 define_insn_and_split if always unused.
9217 * config/rs6000/darwin.md: Ditto.
9218 * config/rs6000/dfp.md: Ditto.
9219 * config/rs6000/rs6000.md: Ditto.
9220 * config/rs6000/sync.md: Ditto.
9221 * config/rs6000/vector.md: Ditto.
9222 * config/rs6000/vsx.md: Ditto.
9224 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
9226 * config/rs6000/altivec.md: Write output control strings as braced
9227 blocks instead of double-quoted strings.
9228 * config/rs6000/darwin.md: Ditto.
9229 * config/rs6000/rs6000.md: Ditto.
9230 * config/rs6000/vector.md: Ditto.
9231 * config/rs6000/vsx.md: Ditto.
9233 2018-02-21 Jason Merrill <jason@redhat.com>
9235 PR c++/84314 - ICE with templates and fastcall attribute.
9236 * attribs.c (build_type_attribute_qual_variant): Remove assert.
9238 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
9240 * ipa-cp.c (determine_versionability): Fix comment typos.
9242 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
9245 * ipa-cp.c (determine_versionability): Do not version functions caling
9248 2018-02-21 Martin Liska <mliska@suse.cz>
9251 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
9252 Add "native" as a possible value.
9253 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
9254 the macro when native cpu detection is available.
9256 2018-02-21 Martin Liska <mliska@suse.cz>
9259 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
9260 Add "native" as a possible value.
9261 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
9262 when native cpu detection is available.
9264 2018-02-21 Jakub Jelinek <jakub@redhat.com>
9265 Martin Sebor <msebor@redhat.com>
9267 PR tree-optimization/84478
9268 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
9270 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
9271 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
9272 support which is conservatively correct, for 2 only stay conservative
9273 for maxlen. Formatting and comment capitalization fixes. Add STRICT
9274 argument to the 2 argument get_range_strlen, adjust 6 arg
9275 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
9277 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
9278 (gimple_fold_builtin_strlen): Pass true as last argument to
9281 2018-02-20 Martin Sebor <msebor@redhat.com>
9284 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
9285 (builtin_memref::set_base_and_offset): Same. Handle inner references.
9286 (builtin_memref::builtin_memref): Factor out parts into
9287 set_base_and_offset and call it.
9289 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
9292 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
9293 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
9294 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
9295 search at the associated MODE_INT.
9297 2018-02-20 Jeff Law <law@redhat.com>
9300 PR tree-optimization/81592
9302 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
9303 for range data rather than using global data.
9304 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
9305 range data rather than using global data.
9306 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
9307 pass it to children as needed.
9308 (struct directive::fmtresult): Similarly.
9309 (struct directive::set_width): Similarly.
9310 (struct directive::set_precision): Similarly.
9311 (format_integer, format_directive, parse_directive): Similarly.
9312 (format_none): Accept unnamed vr_values parameter.
9313 (format_percent, format_floating, format_character): Similarly.
9314 (format_string, format_plain): Similarly.
9315 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
9316 the EVRP range analyzer for range data rather than using global data.
9317 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
9318 gimple-ssa-evrp-analyze.h
9319 (class sprintf_dom_walker): Add after_dom_children member function.
9320 Add evrp_range_analyzer member.
9321 (sprintf_dom_walker::before_dom_children): Call into the EVRP
9322 range analyzer as needed.
9323 (sprintf_dom_walker::after_dom_children): New member function.
9324 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
9326 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
9327 (evrp_range_analyzer::pop_to_marker): Likewise.
9329 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
9331 PR tree-optimization/84419
9332 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
9333 with the required type if its current type is compatible but
9336 2018-02-20 Jakub Jelinek <jakub@redhat.com>
9339 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
9340 after vectorization.
9342 2018-02-20 Martin Liska <mliska@suse.cz>
9345 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
9346 possible values if we don't have a hint.
9348 2018-02-20 Martin Liska <mliska@suse.cz>
9352 * final.c (shorten_branches): Build align_tab array with one
9354 * opts.c (finish_options): Add alignment option limit check.
9355 (MAX_CODE_ALIGN): Likewise.
9356 (MAX_CODE_ALIGN_VALUE): Likewise.
9357 * doc/invoke.texi: Document maximum allowed option value for
9358 all -falign-* options.
9360 2018-02-19 Jakub Jelinek <jakub@redhat.com>
9363 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
9364 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
9365 * var-tracking.c (emit_note_insn_var_location): Remove all references
9366 to NOTE_INSN_CALL_ARG_LOCATION.
9367 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
9368 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
9369 Use copy_rtx_if_shared.
9370 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
9371 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
9372 (dwarf2out_var_location): Remove handling of
9373 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
9375 * final.c (final_scan_insn): Remove all references to
9376 NOTE_INSN_CALL_ARG_LOCATION.
9377 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
9378 before dumping final insns.
9379 * except.c (emit_note_eh_region_end): Remove all references to
9380 NOTE_INSN_CALL_ARG_LOCATION.
9381 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
9382 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
9383 * config/arc/arc.c (hwloop_optimize): Likewise.
9384 * config/arm/arm.c (create_fix_barrier): Likewise.
9385 * config/s390/s390.c (s390_chunkify_start): Likewise.
9386 * config/sh/sh.c (find_barrier): Likewise.
9387 * config/i386/i386.c (rest_of_insert_endbranch,
9388 ix86_seh_fixup_eh_fallthru): Likewise.
9389 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
9390 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
9391 * config/frv/frv.c (frv_function_prologue): Likewise.
9392 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
9394 (note_outside_basic_block_p): Remove all references to
9395 NOTE_INSN_CALL_ARG_LOCATION.
9396 * gengtype.c (adjust_field_rtx_def): Likewise.
9397 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
9399 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
9400 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
9403 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
9406 PR tree-optimization/84452
9407 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
9408 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
9411 2018-02-19 Martin Liska <mliska@suse.cz>
9414 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
9416 2018-02-19 Martin Liska <mliska@suse.cz>
9417 Richard Sandiford <richard.sandiford@linaro.org>
9419 PR tree-optimization/82491
9420 * gimple-fold.c (get_base_constructor): Make earlier bail out
9423 2018-02-19 Carl Love <cel@us.ibm.com>
9425 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
9426 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
9428 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
9429 P8V_BUILTIN_VEC_NEG.
9431 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
9433 * config/rl78/rl78.md (movdf): New define expand.
9435 2018-02-19 Martin Liska <mliska@suse.cz>
9438 * doc/invoke.texi: Fix typo.
9439 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
9441 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
9443 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9444 handle rs6000_single_float and rs6000_double_float specially for
9447 2018-02-16 Jeff Law <law@redhat.com>
9449 * config/rx/rx.c (add_pop_cfi_notes): New function.;
9452 2018-02-16 Jakub Jelinek <jakub@redhat.com>
9455 * ipa-inline.c (inline_small_functions): Fix a typo.
9457 2018-02-16 Nathan Sidwell <nathan@acm.org>
9459 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
9461 2018-02-16 Carl Love <cel@us.ibm.com>
9463 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
9464 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
9465 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
9466 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
9467 expansion to P8V_BUILTIN_VEC_FLOAT2.
9469 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
9471 PR rtl-optimization/70023
9472 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
9473 src_regno into account.
9475 2018-02-16 Carl Love <cel@us.ibm.com>
9477 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
9478 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
9479 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
9480 * config/rs6000/rs6000.c: Remove case statements for
9481 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
9482 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
9483 and P9V_BUILTIN_VEC_VINSERT4B.
9484 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
9485 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
9486 * config/rs6000/vsx.md:
9487 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
9490 2018-02-16 Carl Love <cel@us.ibm.com>
9492 * config/rs6000/altivec.h: Add builtin names vec_extract4b
9494 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
9496 * config/rs6000/rs6000-c.c: Add the definitions for
9497 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
9498 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
9499 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
9500 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
9501 definition for insert4b and define insn *insert3b_internal.
9502 * doc/extend.texi: Add documentation for vec_extract4b.
9504 2018-02-16 Nathan Sidwell <nathan@acm.org>
9506 * doc/extend.texi (Backwards Compatibility): Mention friend
9507 injection. Note for-scope is deprecated.
9508 * doc/invoke.texi (-ffriend-injection): Deprecate.
9510 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
9512 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
9513 that moved to I2, also allow destinations that are a paradoxical
9514 subreg (instead of a normal reg).
9516 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
9519 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
9522 2018-02-16 Richard Biener <rguenther@suse.de>
9524 PR tree-optimization/84037
9525 PR tree-optimization/84016
9527 * config/i386/i386.c (ix86_builtin_vectorization_cost):
9528 Adjust vec_construct for the fact we need additional higher latency
9529 128bit inserts for AVX256 and AVX512 vector builds.
9530 (ix86_add_stmt_cost): Scale vector construction cost for
9533 2018-02-16 Richard Biener <rguenther@suse.de>
9535 PR tree-optimization/84417
9536 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
9537 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
9538 (non_rewritable_lvalue_p): Likewise, use poly-ints.
9540 2018-02-16 Martin Liska <mliska@suse.cz>
9543 * internal-fn.def (ASAN_CHECK): Set proper flags.
9544 (ASAN_MARK): Likewise.
9546 2018-02-16 Julia Koval <julia.koval@intel.com>
9548 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
9549 from PTA_CANNONLAKE.
9551 2018-02-16 Jakub Jelinek <jakub@redhat.com>
9554 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
9555 Use ++iter rather than iter++ for std::list iterators.
9556 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
9557 defer deleting them until all nodes in the forest are processed. Do
9558 free even leaf nodes. Change to_process into auto_vec.
9561 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
9562 * vec.h (vec_default_construct): Use memset instead of placement new
9563 if BROKEN_VALUE_INITIALIZATION is defined.
9564 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
9565 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
9568 PR rtl-optimization/83723
9569 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
9570 * lra.c (lra_substitute_pseudo): Likewise. If true, use
9571 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
9573 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
9575 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
9577 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
9579 PR rtl-optimization/81443
9580 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
9581 from inner REGs to paradoxical SUBREGs.
9583 2018-02-16 Richard Biener <rguenther@suse.de>
9585 PR tree-optimization/84399
9586 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9587 For operands we can analyze at their definition make sure we can
9588 analyze them at each use as well.
9590 2018-02-16 Richard Biener <rguenther@suse.de>
9592 PR tree-optimization/84190
9593 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
9594 volatile accesses if the decl isn't volatile.
9596 2018-02-15 Jason Merrill <jason@redhat.com>
9598 PR c++/84314 - ICE with templates and fastcall attribute.
9599 * attribs.c (build_type_attribute_qual_variant): Don't clobber
9600 TYPE_CANONICAL on an existing type.
9602 2018-02-15 Jakub Jelinek <jakub@redhat.com>
9604 PR tree-optimization/84383
9605 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
9606 dstoff nor call operand_equal_p if dstbase is NULL.
9608 PR tree-optimization/84334
9609 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
9610 also a CONSTANT_CLASS_P, punt.
9612 2018-02-14 Jim Wilson <jimw@sifive.com>
9614 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
9615 first SMALL_OPERAND check. New local min_second_step. Move assert
9616 to where locals are set. Add TARGET_RVC support.
9617 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
9619 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
9621 * doc/invoke.texi: Correct -Wformat-overflow code sample.
9623 2018-02-14 Martin Sebor <msebor@redhat.com>
9625 PR tree-optimization/83698
9626 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
9627 arrays constrain the offset range to their bounds.
9628 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
9629 (builtin_access::overlap): Avoid setting the size of overlap if it's
9631 (maybe_diag_overlap): Also consider arrays when deciding what values
9632 of offsets to include in diagnostics.
9634 2018-02-14 Martin Sebor <msebor@redhat.com>
9637 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
9638 that correspond to the kind of a declaration.
9640 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
9643 * config/pa/pa.md: Load address of PIC label using the linkage table
9644 if the label is nonlocal.
9646 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9648 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
9649 warning message if user requests -maltivec=be.
9650 * doc/invoke.texi: Document deprecation of -maltivec=be.
9652 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
9655 * config/rs6000/rs6000-c.c: Update definitions for
9656 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
9657 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
9659 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9662 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
9663 add _get_ssp intrinsics. Remove argument from
9664 __builtin_ia32_rdssp[d|q].
9665 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
9666 * config/i386/i386-builtin.def: Remove argument from
9667 __builtin_ia32_rdssp[d|q].
9668 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
9669 ix86_expand_special_args_builtin for _rdssp[d|q].
9670 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
9671 Clear register before usage.
9672 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
9673 Add documentation for new _get_ssp and _inc_ssp intrinsics.
9675 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
9677 PR tree-optimization/84357
9678 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
9679 operand 1 of an ARRAY_REF too.
9681 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
9684 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
9685 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
9687 (set_of_reg): New struct.
9688 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
9689 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
9690 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
9692 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
9693 Split into bitclr, bitset, bitinvert patterns if appropriate.
9694 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
9695 use rx_fuse_in_memory_bitop.
9696 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
9697 to named insn, correct maximum insn length.
9699 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
9702 * machmode.def: Define a complex mode for PARTIAL_INT.
9703 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
9705 * doc/rtl.texi: Document CSPImode.
9706 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
9708 (msp430_hard_regno_nregs_with_padding): Likewise.
9710 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
9713 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
9715 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
9717 PR rtl-optimization/84169
9718 * combine.c (try_combine): New variable split_i2i3. Set it to true if
9719 we generated a parallel as new i3 and we split that to new i2 and i3
9720 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
9721 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
9722 those to i2, not i1. Partially rewrite this scan code.
9724 2018-02-13 Jakub Jelinek <jakub@redhat.com>
9727 * stor-layout.c (place_field): For variable length fields, adjust
9728 offset_align afterwards not just based on the field's alignment,
9729 but also on the size.
9732 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
9733 of exps and logs in the use_exp2 case.
9735 2018-02-13 Jeff Law <law@redhat.com>
9737 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
9740 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
9743 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
9747 * common.opt (gas-loc-support, gas-locview-support): New.
9748 (ginline-points, ginternal-reset-location-views): New.
9749 * doc/invoke.texi: Document them. Use @itemx where intended.
9750 (gvariable-location-views): Adjust.
9751 * target.def (reset_location_view): New.
9752 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
9753 (TARGET_RESET_LOCATION_VIEW): New.
9754 * doc/tm.texi: Rebuilt.
9755 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
9756 (dwarf2out_default_as_locview_support): New.
9757 (output_asm_line_debug_info): Use option variables.
9758 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
9759 (output_loc_list): Likewise.
9760 (add_high_low_attributes): Check option variables.
9761 Don't output entry view attribute in strict mode.
9762 (gen_inlined_subroutine_die): Check option variables.
9763 (dwarf2out_inline_entry): Likewise.
9764 (init_sections_and_labels): Likewise.
9765 (dwarf2out_early_finish): Likewise.
9766 (maybe_reset_location_view): New, from...
9767 (dwarf2out_var_location): ... here. Call it.
9768 * debug.h (dwarf2out_default_as_loc_support): Declare.
9769 (dwarf2out_default_as_locview_support): Declare.
9770 * hooks.c (hook_int_rtx_insn_0): New.
9771 * hooks.h (hook_int_rtx_insn_0): Declare.
9772 * toplev.c (process_options): Take -gas-loc-support and
9773 -gas-locview-support from dwarf2out. Enable
9774 -gvariable-location-views by default only with locview
9775 assembler support. Enable -ginternal-reset-location-views by
9776 default only if the target defines the corresponding hook.
9777 Enable -ginline-points by default if location views are
9778 enabled; force it disabled if statement frontiers are
9780 * tree-inline.c (expand_call_inline): Check option variables.
9781 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
9783 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
9785 PR tree-optimization/84321
9786 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
9787 handling. Also check whether the anti-range contains any values
9788 that satisfy the mask; switch to a VR_RANGE if not.
9790 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
9793 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
9795 2018-02-13 Martin Jambor <mjambor@suse.cz>
9798 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
9799 of call statements, also set location of a load to a temporary.
9801 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
9803 * config/rl78/rl78.c (add_vector_labels): New function.
9804 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
9805 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
9806 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
9807 which checks that no arguments are passed.
9808 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
9809 * doc/extend.texi: Documentation for the new attribute.
9811 2018-02-13 Andreas Schwab <schwab@suse.de>
9813 * config/riscv/linux.h (CPP_SPEC): Define.
9815 2018-02-13 Jakub Jelinek <jakub@redhat.com>
9818 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
9819 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
9820 OPTION_MASK_ISA_AES as first argument to def_builtin_const
9821 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
9822 instead of OPTION_MASK_ISA_PCLMUL as first argument to
9823 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
9824 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
9825 temporarily for AES and PCLMUL builtins.
9827 PR tree-optimization/84339
9828 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
9829 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
9833 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
9834 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
9835 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
9837 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
9839 * omp-simd-clone.h: New file.
9840 * omp-simd-clone.c: Include omp-simd-clone.h.
9841 (expand_simd_clones): No longer static.
9842 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
9843 cgraph.h and omp-simd-clone.h.
9844 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
9845 (vect_recog_widen_shift_pattern): Formatting fix.
9846 (vect_pattern_recog_1): Don't check optab for calls.
9849 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
9850 operands[2] into a REG before using gen_lowpart on it.
9852 2018-02-12 Jeff Law <law@redhat.com>
9855 * config/sh/sh.c (find_barrier): Consider a sibling call
9858 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
9859 successfully back substituting a reg.
9861 2018-02-12 Richard Biener <rguenther@suse.de>
9863 PR tree-optimization/84037
9864 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
9865 parameter, move visited init to caller.
9866 (vect_slp_analyze_operations): Separate cost from validity
9867 check, initialize visited once for all instances.
9868 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
9870 * tree-vect-stmts.c (vect_model_simple_cost): Make early
9872 (vect_model_promotion_demotion_cost): Likewise.
9873 (vectorizable_bswap): Guard cost modeling with !slp_node
9874 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
9876 (vectorizable_call): Likewise.
9877 (vectorizable_conversion): Likewise.
9878 (vectorizable_assignment): Likewise.
9879 (vectorizable_shift): Likewise.
9880 (vectorizable_operation): Likewise.
9881 (vectorizable_store): Likewise.
9882 (vectorizable_load): Likewise.
9883 (vectorizable_condition): Likewise.
9884 (vectorizable_comparison): Likewise.
9886 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
9889 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
9890 (ASAN_MARK): Fix fnspec to account for return value, change pointer
9891 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
9893 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
9896 * params.def (inline-min-speedup): Increase from 8 to 15.
9897 (max-inline-insns-auto): Decrease from 40 to 30.
9898 * ipa-split.c (consider_split): Add some buffer for function to
9899 be considered inlining candidate.
9900 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
9903 2018-02-12 Richard Biener <rguenther@suse.de>
9905 PR tree-optimization/84037
9906 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
9907 matched stmts if we cannot swap the non-matched ones.
9909 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
9911 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
9912 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
9913 _mm_maskz_scalef_round_ss): New intrinsics.
9914 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
9915 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
9916 __builtin_ia32_scalefss_round): Remove.
9917 (__builtin_ia32_scalefsd_mask_round,
9918 __builtin_ia32_scalefss_mask_round): New intrinsics.
9919 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
9920 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
9921 ((match_operand:VF_128 2 "<round_nimm_predicate>"
9922 "<round_constraint>")): Changed to ...
9923 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
9924 "<round_scalar_constraint>")): ... this.
9925 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
9926 %0, %1, %2<round_op3>}"): Changed to ...
9927 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
9928 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
9929 %2<round_scalar_mask_op3>}"): ... this.
9930 * config/i386/subst.md (round_scalar_nimm_predicate): New.
9932 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
9934 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
9935 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
9936 (_mm_maskz_sqrt_round_ss): New intrinsics.
9937 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
9938 (__builtin_ia32_sqrtsd_mask_round)
9939 (__builtin_ia32_sqrtss_mask_round): New builtins.
9940 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
9941 (__builtin_ia32_sqrtss_round): Remove.
9942 (__builtin_ia32_sqrtsd_mask_round)
9943 (__builtin_ia32_sqrtss_mask_round): New builtins.
9944 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
9945 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
9946 ((match_operand:VF_128 1 "vector_operand"
9947 "xBm,<round_constraint>")): Changed to ...
9948 ((match_operand:VF_128 1 "vector_operand"
9949 "xBm,<round_scalar_constraint>")): ... this.
9950 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
9951 %0, %2, %<iptr>1<round_op3>}): Changed to ...
9952 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
9953 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
9954 %<iptr>1<round_scalar_mask_op3>}): ... this.
9955 ((set_attr "prefix" "<round_prefix>")): Changed to ...
9956 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
9958 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
9961 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
9962 Cast vec_cmpeq result to correct type.
9963 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
9964 Cast vec_cmpgt result to correct type.
9966 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
9968 * final.c (final_scan_insn_1): Renamed from...
9969 (final_scan_insn): ... this. New wrapper, to recover
9970 seen from the outermost call in recursive ones.
9971 * config/sparc/sparc.c (output_return): Drop seen from call.
9972 (output_sibcall): Likewise.
9973 * config/visium/visium.c (output_branch): Likewise.
9975 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
9977 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
9980 2018-02-10 Alan Modra <amodra@gmail.com>
9983 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
9984 Specify LR as an input.
9986 2018-02-10 Jakub Jelinek <jakub@redhat.com>
9989 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
9990 remove_member_access_dummy_vars): New functions.
9991 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
9992 lower_omp_1, execute_lower_omp): Use them.
9994 PR rtl-optimization/84308
9995 * shrink-wrap.c (spread_components): Release todo vector.
9997 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
9999 PR rtl-optimization/57193
10000 * ira-color.c (struct allocno_color_data): Add member
10001 conflict_allocno_hard_prefs.
10002 (update_conflict_allocno_hard_prefs): New.
10003 (bucket_allocno_compare_func): Add a preference based on
10004 conflict_allocno_hard_prefs.
10005 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
10006 (color_allocnos): Remove a dead code. Initiate
10007 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
10009 2018-02-09 Jakub Jelinek <jakub@redhat.com>
10012 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
10013 constraint from =wa to wa. Avoid a subreg on the output operand,
10014 instead use a pseudo and subreg it in a move.
10015 (p9_xxbrd_<mode>): Changed to ...
10016 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
10017 (p9_xxbrd_v2df): New expander.
10018 (p9_xxbrw_<mode>): Changed to ...
10019 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
10020 (p9_xxbrw_v4sf): New expander.
10022 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
10024 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
10026 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
10029 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
10030 multiply in 32-bit mode.
10031 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
10032 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
10035 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
10037 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
10038 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
10039 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
10040 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
10042 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
10045 * dwarf2out.c (is_trivial_indirect_ref): New function.
10046 (dwarf2out_late_global_decl): Do not generate a location
10047 attribute for variables that have a non-trivial DECL_VALUE_EXPR
10048 and that are not defined in the current unit.
10050 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
10052 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
10053 instead of a libcall for UNORDERED.
10055 2018-02-09 Tamar Christina <tamar.christina@arm.com>
10058 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
10059 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
10061 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10064 * config/s390/s390.c (s390_set_current_function): Invoke
10065 s390_indirect_branch_settings also if fndecl didn't change.
10067 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
10069 * config/rs6000/rs6000.md (blockage): Set length to zero.
10071 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
10073 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
10075 2018-02-09 Jakub Jelinek <jakub@redhat.com>
10078 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
10079 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
10083 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
10084 PARALLEL incoming that failed vt_get_decl_and_offset check.
10086 PR middle-end/84237
10087 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
10088 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
10090 (get_variable_section): For decls in named .bss* sections pass true as
10091 second argument to bss_initializer_p.
10093 2018-02-09 Marek Polacek <polacek@redhat.com>
10094 Jakub Jelinek <jakub@redhat.com>
10097 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
10098 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
10099 Sync some changes from cxx_fold_indirect_ref.
10101 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
10103 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
10105 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
10106 (BLOCK_INLINE_ENTRY_LABEL): New.
10107 (dwarf2out_var_location): Disregard inline entry markers.
10108 (inline_entry_data): New struct.
10109 (inline_entry_data_hasher): New hashtable type.
10110 (inline_entry_data_hasher::hash): New.
10111 (inline_entry_data_hasher::equal): New.
10112 (inline_entry_data_table): New variable.
10113 (add_high_low_attributes): Add DW_AT_entry_pc and
10114 DW_AT_GNU_entry_view attributes if a pending entry is found
10115 in inline_entry_data_table. Add old entry_pc attribute only
10116 if debug nonbinding markers are disabled.
10117 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
10118 markers are enabled.
10119 (block_within_block_p, dwarf2out_inline_entry): New.
10120 (dwarf2out_finish): Check that no entries remained in
10121 inline_entry_data_table.
10122 * final.c (reemit_insn_block_notes): Handle inline entry notes.
10123 (final_scan_insn, notice_source_line): Likewise.
10124 (rest_of_clean_state): Skip inline entry markers.
10125 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
10127 * gimple.c (gimple_build_debug_inline_entry): New.
10128 * gimple.h (enum gimple_debug_subcode): Add
10129 GIMPLE_DEBUG_INLINE_ENTRY.
10130 (gimple_build_debug_inline_entry): Declare.
10131 (gimple_debug_inline_entry_p): New.
10132 (gimple_debug_nonbind_marker_p): Adjust.
10133 * insn-notes.def (INLINE_ENTRY): New.
10134 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
10135 inline entry marker notes.
10136 (print_insn): Likewise.
10137 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
10138 (INSN_DEBUG_MARKER_KIND): Likewise.
10139 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
10140 * tree-inline.c (expand_call_inline): Build and insert
10141 debug_inline_entry stmt.
10142 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
10143 inline entry blocks early, if nonbind markers are enabled.
10144 (dump_scope_block): Dump fragment info.
10145 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
10146 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
10147 (gimple_build_debug_inline_entry): New.
10148 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
10149 Enable/disable inline entry points too.
10150 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
10151 (DEBUG_INSN): Describe inline entry markers.
10153 * common.opt (gvariable-location-views): New.
10154 (gvariable-location-views=incompat5): New.
10155 * config.in: Rebuilt.
10156 * configure: Rebuilt.
10157 * configure.ac: Test assembler for view support.
10158 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
10159 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
10160 * dwarf2out.c (var_loc_view): New typedef.
10161 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
10162 (dwarf2out_locviews_in_attribute): New.
10163 (dwarf2out_locviews_in_loclist): New.
10164 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
10165 (enum dw_line_info_opcode): Add LI_adv_address.
10166 (struct dw_line_info_table): Add view.
10167 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
10168 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
10169 (zero_view_p): New variable.
10170 (ZERO_VIEW_P): New macro.
10171 (output_asm_line_debug_info): New.
10172 (struct var_loc_node): Add view.
10173 (add_AT_view_list, AT_loc_list): New.
10174 (add_var_loc_to_decl): Add view param. Test it against last.
10175 (new_loc_list): Add view params. Record them.
10176 (AT_loc_list_ptr): Handle loc and view lists.
10177 (view_list_to_loc_list_val_node): New.
10178 (print_dw_val): Handle dw_val_class_view_list.
10179 (size_of_die): Likewise.
10180 (value_format): Likewise.
10181 (loc_list_has_views): New.
10182 (gen_llsym): Set vl_symbol too.
10183 (maybe_gen_llsym, skip_loc_list_entry): New.
10184 (dwarf2out_maybe_output_loclist_view_pair): New.
10185 (output_loc_list): Output view list or entries too.
10186 (output_view_list_offset): New.
10187 (output_die): Handle dw_val_class_view_list.
10188 (output_dwarf_version): New.
10189 (output_compilation_unit_header): Use it.
10190 (output_skeleton_debug_sections): Likewise.
10191 (output_rnglists, output_line_info): Likewise.
10192 (output_pubnames, output_aranges): Update version comments.
10193 (output_one_line_info_table): Output view numbers in asm comments.
10194 (dw_loc_list): Determine current endview, pass it to new_loc_list.
10195 Call maybe_gen_llsym.
10196 (loc_list_from_tree_1): Adjust.
10197 (add_AT_location_description): Create view list attribute if
10198 needed, check it's absent otherwise.
10199 (convert_cfa_to_fb_loc_list): Adjust.
10200 (maybe_emit_file): Call output_asm_line_debug_info for test.
10201 (dwarf2out_var_location): Reset views as needed. Precompute
10202 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
10203 attribute. Set view.
10204 (new_line_info_table): Reset next view.
10205 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
10206 (dwarf2out_source_line): Likewise. Output view resets and labels to
10207 the assembler, or select appropriate line info opcodes.
10208 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
10209 (optimize_string_length): Catch it. Adjust.
10210 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
10211 dw_val_class_view_list, and remove it if no longer needed.
10212 (hash_loc_list): Hash view numbers.
10213 (loc_list_hasher::equal): Compare them.
10214 (optimize_location_lists): Check whether a view list symbol is
10215 needed, and whether the locview attribute is present, and
10216 whether they match. Remove the locview attribute if no longer
10218 (index_location_lists): Call skip_loc_list_entry for test.
10219 (dwarf2out_finish): Call output_asm_line_debug_info for test.
10220 Use output_dwarf_version.
10221 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
10222 (struct dw_val_node): Add val_view_list.
10223 * final.c (SEEN_NEXT_VIEW): New.
10224 (set_next_view_needed): New.
10225 (clear_next_view_needed): New.
10226 (maybe_output_next_view): New.
10227 (final_start_function): Rename to...
10228 (final_start_function_1): ... this. Take pointer to FIRST,
10229 add SEEN parameter. Emit param bindings in the initial view.
10230 (final_start_function): Reintroduce SEEN-less interface.
10231 (final): Rename to...
10232 (final_1): ... this. Take SEEN parameter. Output final pending
10233 next view at the end.
10234 (final): Reintroduce seen-less interface.
10235 (final_scan_insn): Output pending next view before switching
10236 sections or ending a block. Mark the next view as needed when
10237 outputting variable locations. Notify debug backend of section
10238 changes, and of location view changes.
10239 (rest_of_handle_final): Adjust.
10240 * toplev.c (process_options): Autodetect value for debug variable
10241 location views option. Warn on incompat5 without -gdwarf-5.
10242 * doc/invoke.texi (gvariable-location-views): New.
10243 (gvariable-location-views=incompat5): New.
10244 (gno-variable-location-views): New.
10246 2018-02-08 David Malcolm <dmalcolm@redhat.com>
10248 PR tree-optimization/84136
10249 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
10250 that the result of find_edge is non-NULL.
10252 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
10255 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
10256 storing integer register in SImode. Fix cost of 256 and 512
10257 byte aligned SSE register store.
10259 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
10261 * config/i386/i386.c (ix86_multiplication_cost): Fix
10262 multiplication cost for TARGET_AVX512DQ.
10264 2018-02-08 Marek Polacek <polacek@redhat.com>
10266 PR tree-optimization/84238
10267 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
10270 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
10272 PR tree-optimization/84265
10273 * tree-vect-stmts.c (vectorizable_store): Don't treat
10274 VMAT_CONTIGUOUS accesses as grouped.
10275 (vectorizable_load): Likewise.
10277 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
10279 PR tree-optimization/81635
10280 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
10281 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
10282 (test_round_for_mask): New functions.
10283 (wide_int_cc_tests): Call test_round_for_mask.
10284 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
10285 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
10286 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
10287 range returned by get_range_info.
10289 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
10292 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
10293 * symtab.c: Include builtins.h
10294 (symtab_node::output_to_lto_symbol_table_p): Move here
10295 from lto-streamer-out.c:output_symbol_p.
10296 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
10297 (output_symbol_p): Move all logic to symtab.c
10298 (produce_symtab): Update.
10300 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10302 * config/s390/s390-opts.h (enum indirect_branch): Define.
10303 * config/s390/s390-protos.h (s390_return_addr_from_memory)
10304 (s390_indirect_branch_via_thunk)
10305 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
10306 (enum s390_indirect_branch_type): Define.
10307 * config/s390/s390.c (struct s390_frame_layout, struct
10308 machine_function): Remove.
10309 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
10310 (indirect_branch_table_label_no, indirect_branch_table_name):
10312 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
10313 (enum s390_indirect_branch_option): Define.
10314 (s390_return_addr_from_memory): New function.
10315 (s390_handle_string_attribute): New function.
10316 (s390_attribute_table): Add new attribute handler.
10317 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
10318 (s390_indirect_branch_via_thunk): New function.
10319 (s390_indirect_branch_via_inline_thunk): New function.
10320 (s390_function_ok_for_sibcall): When jumping via thunk disallow
10321 sibling call optimization for non z10 compiles.
10322 (s390_emit_call): Force indirect branch target to be a single
10323 register. Add r1 clobber for non-z10 compiles.
10324 (s390_emit_epilogue): Emit return jump via return_use expander.
10325 (s390_reorg): Handle JUMP_INSNs as execute targets.
10326 (s390_option_override_internal): Perform validity checks for the
10327 new command line options.
10328 (s390_indirect_branch_attrvalue): New function.
10329 (s390_indirect_branch_settings): New function.
10330 (s390_set_current_function): Invoke s390_indirect_branch_settings.
10331 (s390_output_indirect_thunk_function): New function.
10332 (s390_code_end): Implement target hook.
10333 (s390_case_values_threshold): Implement target hook.
10334 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
10336 * config/s390/s390.h (struct s390_frame_layout)
10337 (struct machine_function): Move here from s390.c.
10338 (TARGET_INDIRECT_BRANCH_NOBP_RET)
10339 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
10340 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
10341 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
10342 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
10343 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
10344 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
10345 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
10346 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
10347 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
10348 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
10349 (mnemonic attribute): Add values which aren't recognized
10351 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
10352 pattern for branch conversion. Fix mnemonic attribute.
10353 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
10354 indirect branch via thunk if requested.
10355 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
10356 ("*indirect_jump"): Disable for branch conversion using out of
10358 ("indirect_jump_via_thunk<mode>_z10")
10359 ("indirect_jump_via_thunk<mode>")
10360 ("indirect_jump_via_inlinethunk<mode>_z10")
10361 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
10362 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
10363 ("casesi_jump_via_inlinethunk<mode>_z10")
10364 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
10365 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
10366 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
10367 ("*indirect2_jump"): Disable for branch conversion.
10368 ("casesi_jump"): Turn into expander and expand patterns for branch
10370 ("return_use"): New expander.
10371 ("*return"): Emit return via thunk and rename it to ...
10372 ("*return<mode>"): ... this one.
10373 * config/s390/s390.opt: Add new options and and enum for the
10376 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
10378 * lra-constraints.c (match_reload): Unconditionally use
10379 gen_lowpart_SUBREG, rather than selecting between that
10380 and equivalent gen_rtx_SUBREG code.
10382 2018-02-08 Richard Biener <rguenther@suse.de>
10384 PR tree-optimization/84233
10385 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
10386 changed flag instead of boguously re-using phi_inserted.
10388 2018-02-08 Martin Jambor <mjambor@suse.cz>
10390 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
10391 static local variables.
10393 2018-02-08 Richard Biener <rguenther@suse.de>
10395 PR tree-optimization/84278
10396 * tree-vect-stmts.c (vectorizable_store): When looking for
10397 smaller vector types to perform grouped strided loads/stores
10398 make sure the mode is supported by the target.
10399 (vectorizable_load): Likewise.
10401 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
10403 * config/aarch64/aarch64.c (aarch64_components_for_bb):
10404 Increase LDP/STP opportunities by adding adjacent callee-saves.
10406 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
10408 PR rtl-optimization/84068
10409 PR rtl-optimization/83459
10410 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
10412 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
10414 PR tree-optimization/84224
10415 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
10416 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
10417 non-zero arguments.
10419 2018-02-07 Iain Sandoe <iain@codesourcery.com>
10422 * config/rs6000/altivec.md (*restore_world): Remove LR use.
10423 * config/rs6000/predicates.md (restore_world_operation): Adjust op
10424 count, remove one USE.
10426 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10428 * doc/install.texi (Configuration): Document the
10429 --with-long-double-format={ibm,ieee} PowerPC configuration
10433 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
10434 Convert from define_expand to be define_insn_and_split. Rework
10435 float/double/_Float128 conversions to QI/HI/SImode to work with
10436 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
10437 conversions to QI/HImode types did a store and then a load to
10438 truncate the value. For conversions to VSX registers, don't split
10439 the insn, instead emit the code directly. Use the code iterator
10440 any_fix to combine signed and unsigned conversions.
10441 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
10442 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
10443 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
10444 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
10445 (fix_<mode>di2_hw): Likewise.
10446 (fixuns_<mode>di2_hw): Likewise.
10447 (fix_<mode>si2_hw): Likewise.
10448 (fixuns_<mode>si2_hw): Likewise.
10449 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
10450 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
10451 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
10452 fix<uns>_trunc<SFDF:mode>si2_p8.
10453 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
10454 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10455 (fix<uns>_<mode>_mem): Likewise.
10456 (fctiw<u>z_<mode>_mem): Likewise.
10457 (fix<uns>_<mode>_mem): Likewise.
10458 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
10459 the register allocator from doing a direct move to the GPRs to do
10460 a store, and instead use the ISA 3.0 store byte/half-word from
10461 vector register instruction. For IEEE 128-bit floating point,
10462 also optimize stores of 32-bit ints.
10463 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
10465 2018-02-07 Alan Hayward <alan.hayward@arm.com>
10467 * genextract.c (push_pathstr_operand): New function to support
10469 (walk_rtx): Call push_pathstr_operand.
10470 (print_path): Support [a-zA-Z].
10472 2018-02-07 Richard Biener <rguenther@suse.de>
10474 PR tree-optimization/84037
10475 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
10476 (cse_and_gimplify_to_preheader): Declare.
10477 (vect_get_place_in_interleaving_chain): Likewise.
10478 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10480 (_loop_vec_info::~_loop_vec_info): Delete it.
10481 (cse_and_gimplify_to_preheader): New function.
10482 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
10483 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
10484 (vectorizable_load): Likewise. For grouped stores always base
10485 the IV on the first element.
10486 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
10487 condition before gimplifying.
10489 2018-02-07 Jakub Jelinek <jakub@redhat.com>
10491 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
10492 *DIV_EXPR and *MOD_EXPR.
10494 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
10497 * config/i386/i386.c (ix86_option_override_internal): Mask out
10498 the CF_SET bit when checking -fcf-protection.
10500 2018-02-07 Tom de Vries <tom@codesourcery.com>
10503 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
10506 2018-02-07 Richard Biener <rguenther@suse.de>
10508 PR tree-optimization/84204
10509 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
10512 PR tree-optimization/84205
10513 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
10514 special-case isl_ast_op_zdiv_r.
10516 PR tree-optimization/84223
10517 * graphite-scop-detection.c (gather_bbs::before_dom_children):
10518 Only add conditions from within the region.
10519 (gather_bbs::after_dom_children): Adjust.
10521 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
10524 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
10525 * config/avr/avr.md: Only post-reload split REG-REG moves if
10526 either register is GENERAL_REG_P.
10528 2018-02-07 Jakub Jelinek <jakub@redhat.com>
10530 PR tree-optimization/84235
10531 * tree-ssa-scopedtables.c
10532 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
10533 if the subtraction is performed in floating point type where NaNs are
10534 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
10535 build 1. Formatting fix.
10537 2018-02-06 Jakub Jelinek <jakub@redhat.com>
10540 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
10541 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
10542 and skip it regardless of bb boundaries. Use CALL_P macro,
10543 don't test INSN_P (insn) together with CALL_P or JUMP_P check
10544 unnecessarily, formatting fix.
10546 2018-02-06 Michael Collison <michael.collison@arm.com>
10548 * config/arm/thumb2.md:
10549 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
10550 (*thumb_mov_notscc): Ditto.
10552 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
10555 * config/rs6000/rs6000.md (su code attribute): Use "u" for
10556 unsigned_fix, not "s".
10558 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10560 * configure.ac (gcc_fn_eh_frame_ro): New function.
10561 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
10562 correct .eh_frame permissions.
10563 * configure: Regenerate.
10565 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
10567 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
10568 irrelevant options.
10570 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10572 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10573 Display warning message for -mno-speculate-indirect-jumps.
10575 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
10577 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
10579 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
10581 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
10583 PR tree-optimization/84225
10584 * tree-eh.c (find_trapping_overflow): Only call
10585 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
10587 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10590 * config/i386/i386.c: Reimplement the check of possible options
10591 -mibt/-mshstk conbination. Change error messages.
10592 * doc/invoke.texi: Fix a typo: remove extra '='.
10594 2018-02-06 Marek Polacek <polacek@redhat.com>
10596 PR tree-optimization/84228
10597 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
10599 2018-02-06 Tamar Christina <tamar.christina@arm.com>
10602 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
10603 emitted arch directives.
10604 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
10605 __ARM_FEATURE_COPROC before changing architectures.
10607 2018-02-06 Richard Biener <rguenther@suse.de>
10609 * config/i386/i386.c (print_reg): Fix typo.
10610 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
10612 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
10614 * configure: Regenerate.
10616 2018-02-05 Martin Sebor <msebor@redhat.com>
10618 PR tree-optimization/83369
10619 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
10622 2018-02-05 Martin Liska <mliska@suse.cz>
10624 * doc/invoke.texi: Cherry-pick upstream r323995.
10626 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
10628 * ira.c (ira_init_register_move_cost): Adjust comment.
10630 2018-02-05 Martin Liska <mliska@suse.cz>
10632 PR gcov-profile/84137
10633 * doc/gcov.texi: Fix typo in documentation.
10635 2018-02-05 Martin Liska <mliska@suse.cz>
10637 PR gcov-profile/83879
10638 * doc/gcov.texi: Document necessity of --dynamic-list-data when
10639 using dlopen functionality.
10641 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
10643 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
10644 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
10645 _mm_maskz_range_ss, _mm_mask_range_round_ss,
10646 _mm_maskz_range_round_ss): New intrinsics.
10647 (__builtin_ia32_rangesd128_round)
10648 (__builtin_ia32_rangess128_round): Remove.
10649 (__builtin_ia32_rangesd128_mask_round,
10650 __builtin_ia32_rangess128_mask_round): New builtins.
10651 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
10652 __builtin_ia32_rangess128_round): Remove.
10653 (__builtin_ia32_rangesd128_mask_round,
10654 __builtin_ia32_rangess128_mask_round): New builtins.
10655 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
10656 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
10657 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
10658 "<round_saeonly_constraint>")): Changed to ...
10659 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
10660 "<round_saeonly_scalar_constraint>")): ... this.
10661 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10662 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
10663 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
10664 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
10665 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
10667 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
10669 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
10671 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
10672 Remove all values except native, 8540 and 8548.
10674 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
10676 * config/i386/i386.c (ix86_output_function_return): Pass
10677 INVALID_REGNUM, instead of -1, as invalid register number to
10678 indirect_thunk_name and output_indirect_thunk.
10680 2018-02-02 Julia Koval <julia.koval@intel.com>
10682 * config.gcc: Add -march=icelake.
10683 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
10684 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
10685 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
10686 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
10687 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
10688 (processor_target_table): Add icelake.
10689 (ix86_option_override_internal): Handle new PTAs.
10690 (get_builtin_code_for_version): Handle icelake.
10691 (M_INTEL_COREI7_ICELAKE): New.
10692 (fold_builtin_cpu): Handle icelake.
10693 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
10694 * doc/invoke.texi: Add -march=icelake.
10696 2018-02-02 Julia Koval <julia.koval@intel.com>
10698 * config/i386/i386.c (ix86_option_override_internal): Change flags type
10699 to wide_int_bitmask.
10700 * wide-int-bitmask.h: New.
10702 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10705 * config/i386/i386.md: Replace Pmode with word_mode in
10706 builtin_setjmp_setup and builtin_longjmp to support x32.
10708 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
10712 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
10714 (rs6000_supported_cpu_names): New static variable.
10715 (linux_cpu_translation_table): Likewise.
10716 (elf_platform) <cpu>: Define new static variable and use it.
10717 Translate kernel AT_PLATFORM name to canonical name if needed.
10718 Error if platform name is unknown.
10720 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
10723 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
10725 2018-02-01 Jeff Law <law@redhat.com>
10728 * config/i386/i386.c (release_scratch_register_on_entry): Add new
10729 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
10730 the scratch if RELEASE_VIA_POP is false.
10731 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
10732 If we have to save a temporary register, decrement SIZE appropriately.
10733 Pass new arguments to release_scratch_register_on_entry.
10734 (ix86_adjust_stack_and_probe): Likewise.
10735 (ix86_emit_probe_stack_range): Pass new arguments to
10736 release_scratch_register_on_entry.
10738 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
10740 PR rtl-optimization/84157
10741 * combine.c (change_zero_ext): Use REG_P predicate in
10742 front of HARD_REGISTER_P predicate.
10744 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
10746 * config/avr/avr.c (avr_option_override): Move disabling of
10747 -fdelete-null-pointer-checks to...
10748 * common/config/avr/avr-common.c (avr_option_optimization_table):
10751 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10753 PR tree-optimization/81635
10754 * tree-data-ref.c (split_constant_offset_1): For types that
10755 wrap on overflow, try to use range info to prove that wrapping
10758 2018-02-01 Renlin Li <renlin.li@arm.com>
10761 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10762 TAILCALL_ADDR_REGS.
10763 (aarch64_register_move_cost): Likewise.
10764 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
10765 TAILCALL_ADDR_REGS.
10766 (REG_CLASS_NAMES): Likewise.
10767 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
10768 TAILCALL_ADDR_REGS. Remove IP registers.
10769 * config/aarch64/aarch64.md (Ucs): Update register constraint.
10771 2018-02-01 Richard Biener <rguenther@suse.de>
10773 * domwalk.h (dom_walker::dom_walker): Add additional constructor
10774 for specifying RPO order and allow NULL for that.
10775 * domwalk.c (dom_walker::dom_walker): Likewise.
10776 (dom_walker::walk): Handle NULL RPO order.
10777 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
10779 (rewrite_update_dom_walker): Likewise.
10780 (mark_def_dom_walker): Likewise.
10782 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10784 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
10785 (aarch64_maybe_expand_sve_subreg_move): Declare.
10786 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
10787 * config/aarch64/predicates.md (aarch64_any_register_operand): New
10789 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
10790 that are semantically a reverse operation.
10791 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
10792 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
10793 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
10795 (aarch64_can_change_mode_class): For big-endian, forbid changes
10796 between two SVE modes if they have different element sizes.
10798 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10800 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
10801 the TImode handling for big-endian targets.
10803 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10805 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
10806 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
10808 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
10809 Remove BSWAP handing for big-endian targets and use the form of
10810 LD1RQ appropariate for the mode.
10812 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10814 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
10815 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
10816 duplicated element.
10818 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
10821 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
10822 check for operands that need to go through aarch64_sve_reload_be.
10824 2018-02-01 Jakub Jelinek <jakub@redhat.com>
10826 PR tree-optimization/81661
10827 PR tree-optimization/84117
10828 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
10829 * tree-eh.c: Include gimplify.h.
10830 (find_trapping_overflow, replace_trapping_overflow,
10831 rewrite_to_non_trapping_overflow): New functions.
10832 * tree-vect-loop.c: Include tree-eh.h.
10833 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
10834 * tree-data-ref.c: Include tree-eh.h.
10835 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
10837 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
10839 PR rtl-optimization/84123
10840 * combine.c (change_zero_ext): Check if hard register satisfies
10841 can_change_dest_mode before calling gen_lowpart_SUBREG.
10843 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
10846 * ira.c (ira_init_register_move_cost): Remove assert.
10848 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
10850 PR rtl-optimization/84071
10851 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
10852 * doc/tm.texi: Regenerate.
10854 2018-01-31 Richard Biener <rguenther@suse.de>
10856 PR tree-optimization/84132
10857 * tree-data-ref.c (analyze_miv_subscript): Properly
10858 check whether evolution_function_is_affine_multivariate_p
10859 before calling gcd_of_steps_may_divide_p.
10861 2018-01-31 Julia Koval <julia.koval@intel.com>
10864 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
10865 * config/i386/i386.md (rdpid_rex64) New.
10866 (rdpid): Make 32bit only.
10868 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
10871 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
10872 an IDENTIFIER_NODE for FUNCTION_TYPE's.
10874 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
10877 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
10879 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
10881 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
10883 PR rtl-optimization/84071
10884 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
10885 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
10887 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
10889 * config/arc/arc.c (arc_handle_aux_attribute): New function.
10890 (arc_attribute_table): Add 'aux' attribute.
10891 (arc_in_small_data_p): Consider aux like variables.
10892 (arc_is_aux_reg_p): New function.
10893 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
10894 (arc_get_aux_arg): New function.
10895 (prepare_move_operands): Handle aux-register access.
10896 (arc_handle_aux_attribute): New function.
10897 * doc/extend.texi (ARC Variable attributes): Add subsection.
10899 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
10901 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
10902 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
10903 (arc_attribute_table): Add 'uncached' attribute.
10904 (arc_print_operand): Print '.di' flag for uncached memory
10906 (arc_in_small_data_p): Do not consider for small data the uncached
10908 (arc_is_uncached_mem_p): New function.
10909 * config/arc/predicates.md (compact_store_memory_operand): Check
10910 for uncached memory accesses.
10911 (nonvol_nonimm_operand): Likewise.
10912 * gcc/doc/extend.texi (ARC Type Attribute): New subsection.
10914 2018-01-31 Jakub Jelinek <jakub@redhat.com>
10917 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
10918 falign-loops=): Add Optimization flag.
10920 2018-01-30 Jeff Law <law@redhat.com>
10923 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
10924 INT_REGISTERS_SAVED. Check it prior to calling
10925 get_scratch_register_on_entry.
10926 (ix86_adjust_stack_and_probe): Similarly.
10927 (ix86_emit_probe_stack_range): Similarly.
10928 (ix86_expand_prologue): Corresponding changes.
10930 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10933 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
10934 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
10936 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
10939 * lra-constraints.c (curr_insn_transform): Process AND in the
10942 2018-01-30 Jakub Jelinek <jakub@redhat.com>
10944 PR rtl-optimization/83986
10945 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
10946 dependence against last_pending_memory_flush in addition to
10947 pending_jump_insns.
10949 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
10951 PR tree-optimization/81611
10952 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
10955 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10958 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
10961 2018-01-30 Richard Biener <rguenther@suse.de>
10962 Jakub Jelinek <jakub@redhat.com>
10964 PR tree-optimization/84111
10965 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
10966 inner loops added during recursion, as they don't have up-to-date
10969 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
10972 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
10973 (can_inline_edge_by_limits_p): ... here.
10974 (can_early_inline_edge_p, check_callers,
10975 update_caller_keys, update_callee_keys, recursive_inlining,
10976 add_new_edges_to_heap, speculation_useful_p,
10977 inline_small_functions,
10978 inline_small_functions, flatten_function,
10979 inline_to_all_callers_1): Update.
10981 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
10983 * profile-count.c (profile_count::combine_with_ipa_count): Handle
10986 2018-01-30 Richard Biener <rguenther@suse.de>
10988 PR tree-optimization/83008
10989 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
10990 invariant and constant vector uses in stmts when they need
10991 more than one stmt.
10993 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10996 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
10997 * configure: Regenerate.
10999 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
11001 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
11003 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
11004 Use gen_rtx_REG rather than gen_lowpart.
11006 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
11008 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
11009 rather than 0 when creating partial subregs.
11011 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
11013 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
11016 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11019 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
11020 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
11021 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
11022 flags. This restores the settings used before the 2017-07-24.
11023 Turning off pre increment/decrement/modify allows IVOPTS to
11024 optimize DF/SF loops where the index is an int.
11026 2018-01-29 Richard Biener <rguenther@suse.de>
11027 Kelvin Nilsen <kelvin@gcc.gnu.org>
11030 * tree-vect-stmts.c (vectorizable_call): Don't call
11031 targetm.vectorize_builtin_md_vectorized_function if callee is
11034 2018-01-22 Carl Love <cel@us.ibm.com>
11036 * doc/extend.tex: Fix typo in second arg in
11037 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
11039 2018-01-29 Richard Biener <rguenther@suse.de>
11041 PR tree-optimization/84086
11042 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
11043 (flush_ssaname_freelist): When SSA names were released reset
11044 the SCEV hash table.
11046 2018-01-29 Richard Biener <rguenther@suse.de>
11048 PR tree-optimization/84057
11049 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
11050 removed paths when removing edges.
11052 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
11054 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
11055 -mfunction-return=@var{choice}.
11057 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
11059 PR diagnostic/84034
11060 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
11061 Handle CR like TAB.
11062 (layout::print_source_line): Likewise.
11063 (test_get_line_width_without_trailing_whitespace): Add test cases.
11065 2018-01-27 Jakub Jelinek <jakub@redhat.com>
11067 PR middle-end/84040
11068 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
11071 2018-01-26 Jim Wilson <jimw@sifive.com>
11073 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
11075 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
11078 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11080 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
11081 and CMP + SUB-immediate -> SUBS.
11083 2018-01-26 Martin Sebor <msebor@redhat.com>
11085 PR tree-optimization/83896
11086 * tree-ssa-strlen.c (get_string_len): Rename...
11087 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
11088 Avoid assuming length is constant.
11089 (handle_char_store): Use HOST_WIDE_INT for string length.
11091 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
11094 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
11095 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
11097 2018-01-26 Richard Biener <rguenther@suse.de>
11099 PR rtl-optimization/84003
11100 * dse.c (record_store): Only record redundant stores when
11101 the earlier store aliases at least all accesses the later one does.
11103 2018-01-26 Jakub Jelinek <jakub@redhat.com>
11105 PR rtl-optimization/83985
11106 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
11107 REG_CFA_RESTORE insns.
11108 (delete_unmarked_insns): Don't ignore separate shrink wrapping
11109 REG_CFA_RESTORE insns here.
11112 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
11113 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
11115 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
11117 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
11118 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
11119 (arc_init): Likewise.
11120 (arc_override_options): Likewise.
11121 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
11123 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
11125 * config/arc/arc.h (TARGET_DBNZ): Define.
11126 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
11127 properly set the tune attribute.
11128 (dbnz): Use TARGET_DBNZ guard.
11129 * config/arc/arc.opt (mtune): Add core3 option.
11131 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
11133 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
11134 recognize new pic like addresses.
11135 (arc_delegitimize_address): Clean up.
11137 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
11139 * config/arc/arc-arches.def: Option mrf16 valid for all
11141 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
11142 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
11143 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
11144 * config/arc/arc-tables.opt: Regenerate.
11145 * config/arc/arc.c (arc_conditional_register_usage): Handle
11146 reduced register file case.
11147 (arc_file_start): Set must have build attributes.
11148 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
11149 mrf16 option value.
11150 * config/arc/arc.opt (mrf16): Add new option.
11151 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
11152 * config/arc/genmultilib.awk: Handle new mrf16 option.
11153 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
11154 * config/arc/t-multilib: Regenerate.
11155 * doc/invoke.texi (ARC Options): Document mrf16 option.
11157 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
11159 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
11160 * config/arc/arc.c (arc_handle_secure_attribute): New function.
11161 (arc_attribute_table): Add 'secure_call' attribute.
11162 (arc_print_operand): Print secure call operand.
11163 (arc_function_ok_for_sibcall): Don't optimize tail calls when
11165 (arc_is_secure_call_p): New function. * config/arc/arc.md
11166 (call_i): Add support for sjli instruction.
11167 (call_value_i): Likewise.
11168 * config/arc/constraints.md (Csc): New constraint.
11170 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
11171 John Eric Martin <John.Martin@emmicro-us.com>
11173 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
11174 * config/arc/arc.c (_arc_jli_section): New struct.
11175 (arc_jli_section): New type.
11176 (rc_jli_sections): New static variable.
11177 (arc_handle_jli_attribute): New function.
11178 (arc_attribute_table): Add jli_always and jli_fixed attribute.
11179 (arc_file_end): New function.
11180 (TARGET_ASM_FILE_END): Define.
11181 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
11182 (arc_add_jli_section): New function.
11183 (jli_call_scan): Likewise.
11184 (arc_reorg): Call jli_call_scan.
11185 (arc_output_addsi): Remove 'S' from printing asm operand.
11186 (arc_is_jli_call_p): New function.
11187 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
11189 (movhi_insn): Likewise.
11190 (movsi_insn): Likewise.
11191 (movsi_set_cc_insn): Likewise.
11192 (loadqi_update): Likewise.
11193 (load_zeroextendqisi_update): Likewise.
11194 (load_signextendqisi_update): Likewise.
11195 (loadhi_update): Likewise.
11196 (load_zeroextendhisi_update): Likewise.
11197 (load_signextendhisi_update): Likewise.
11198 (loadsi_update): Likewise.
11199 (loadsf_update): Likewise.
11200 (movsicc_insn): Likewise.
11201 (bset_insn): Likewise.
11202 (bxor_insn): Likewise.
11203 (bclr_insn): Likewise.
11204 (bmsk_insn): Likewise.
11205 (bicsi3_insn): Likewise.
11206 (cmpsi_cc_c_insn): Likewise.
11207 (movsi_ne): Likewise.
11208 (movsi_cond_exec): Likewise.
11209 (clrsbsi2): Likewise.
11210 (norm_f): Likewise.
11220 (call_i): Remove 'S' asm letter, add jli instruction.
11221 (call_value_i): Likewise.
11222 * config/arc/arc.op (mjli-always): New option.
11223 * config/arc/constraints.md (Cji): New constraint.
11224 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
11226 (subsf3_fpx): Likewise.
11227 (mulsf3_fpx): Likewise.
11228 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
11230 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
11231 function attrbutes.
11232 * doc/invoke.texi (ARC): Document mjli-always option.
11234 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
11236 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
11237 avoid addition with 0 and use incw and decw where possible.
11239 2018-01-26 Richard Biener <rguenther@suse.de>
11241 PR tree-optimization/81082
11242 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
11243 association if it requires casting to unsigned.
11244 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
11245 from fold_plusminus_mult_expr to catch important cases late when
11246 range info is available.
11248 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11250 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
11251 * configure.ac (hidden_linkonce): New test.
11252 * configure: Regenerate.
11253 * config.in: Regenerate.
11255 2018-01-26 Julia Koval <julia.koval@intel.com>
11257 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
11258 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
11259 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
11260 _mm_mask_bitshuffle_epi64_mask): Fix type.
11261 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
11262 USI_FTYPE_V4DI_V4DI_USI): Remove.
11263 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
11264 __builtin_ia32_vpshufbitqmb256_mask,
11265 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
11266 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
11267 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
11269 2018-01-26 Alan Modra <amodra@gmail.com>
11272 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
11273 UNSPEC_VBPERMQ. Sort other unspecs.
11275 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
11277 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
11279 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
11281 PR middle-end/83055
11282 * predict.c (drop_profile): Do not push/pop cfun; update also
11284 (handle_missing_profiles): Fix logic looking for zero profiles.
11286 2018-01-25 Jakub Jelinek <jakub@redhat.com>
11288 PR middle-end/83977
11289 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
11290 on functions with #pragma omp declare simd or functions with simd
11292 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
11293 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
11294 Remove trailing \n from warning_at calls.
11296 2018-01-25 Tom de Vries <tom@codesourcery.com>
11299 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
11300 for neutered workers.
11302 2018-01-24 Joseph Myers <joseph@codesourcery.com>
11305 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
11306 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
11308 2018-01-24 Jeff Law <law@redhat.com>
11311 * i386.c (get_probe_interval): Move to earlier point.
11312 (ix86_compute_frame_layout): If -fstack-clash-protection and
11313 the frame is larger than the probe interval, then use pushes
11314 to save registers rather than reg->mem moves.
11315 (ix86_expand_prologue): Remove conditional for int_registers_saved
11318 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
11321 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
11322 min/max for never referenced object.
11324 2018-01-24 Jakub Jelinek <jakub@redhat.com>
11326 PR middle-end/83977
11327 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
11329 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
11330 attributes from DECL_ATTRIBUTES (decl) without affecting
11331 DECL_ATTRIBUTES (current_function_decl).
11332 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
11333 functions with non-NULL DECL_ABSTRACT_ORIGIN.
11335 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
11337 PR tree-optimization/83979
11338 * fold-const.c (fold_comparison): Use constant_boolean_node
11339 instead of boolean_{true,false}_node.
11341 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
11343 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
11346 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11348 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
11349 Simplify the clause that sets the length attribute.
11350 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
11351 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
11352 clause that sets the length attribute.
11353 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
11355 2018-01-24 Tom de Vries <tom@codesourcery.com>
11358 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
11359 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
11360 Add strict parameter.
11361 (prevent_branch_around_nothing): Insert dummy insn between branch to
11362 label and label with no ptx insn inbetween.
11363 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
11365 2018-01-24 Tom de Vries <tom@codesourcery.com>
11368 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
11369 for neutered threads in warp.
11370 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
11372 2018-01-24 Richard Biener <rguenther@suse.de>
11374 PR tree-optimization/83176
11375 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
11378 2018-01-24 Richard Biener <rguenther@suse.de>
11380 PR tree-optimization/82819
11381 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
11382 code generating pluses that are no-ops in the target precision.
11384 2018-01-24 Richard Biener <rguenther@suse.de>
11386 PR middle-end/84000
11387 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
11389 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
11391 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
11392 to merge probabilities.
11393 * predict.c (probably_never_executed): Also mark as cold functions
11394 with global 0 profile and guessed local profile.
11395 * profile-count.c (profile_probability::combine_with_count): New
11397 * profile-count.h (profile_probability::operator*,
11398 profile_probability::operator*=, profile_probability::operator/,
11399 profile_probability::operator/=): Reduce precision to adjusted
11400 and set value to guessed on contradictory divisions.
11401 (profile_probability::combine_with_freq): Remove.
11402 (profile_probability::combine_wiht_count): Declare.
11403 (profile_count::force_nonzero):: Set to adjusted.
11404 (profile_count::probability_in):: Set quality to adjusted.
11405 * tree-ssa-tail-merge.c (replace_block_by): Use
11406 combine_with_count.
11408 2018-01-23 Andrew Waterman <andrew@sifive.com>
11409 Jim Wilson <jimw@sifive.com>
11411 * config/riscv/riscv.c (riscv_stack_boundary): New.
11412 (riscv_option_override): Set riscv_stack_boundary. Handle
11413 riscv_preferred_stack_boundary_arg.
11414 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
11415 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
11416 (STACK_BOUNDARY): Set to riscv_stack_boundary.
11417 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
11418 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
11419 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
11421 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
11424 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
11425 of struct ix86_frame.
11426 (ix86_expand_epilogue): Likewise. Add a local variable for
11427 the reg_save_offset field in struct ix86_frame.
11429 2018-01-23 Bin Cheng <bin.cheng@arm.com>
11431 PR tree-optimization/82604
11432 * tree-loop-distribution.c (enum partition_kind): New enum item
11433 PKIND_PARTIAL_MEMSET.
11434 (partition_builtin_p): Support above new enum item.
11435 (generate_code_for_partition): Ditto.
11436 (compute_access_range): Differentiate cases that equality can be
11437 proven at all loops, the innermost loops or no loops.
11438 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
11439 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
11440 (finalize_partitions, distribute_loop): Don't fuse partition of
11441 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
11442 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
11443 parloop is enabled.
11445 2018-01-23 Martin Liska <mliska@suse.cz>
11447 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
11448 order to ignore the predictor.
11449 (PRED_POLYMORPHIC_CALL): Likewise.
11450 (PRED_RECURSIVE_CALL): Likewise.
11452 2018-01-23 Martin Liska <mliska@suse.cz>
11454 * tree-profile.c (tree_profiling): Print function header to
11455 aware reader which function we are working on.
11456 * value-prof.c (gimple_find_values_to_profile): Do not print
11457 not interesting value histograms.
11459 2018-01-23 Martin Liska <mliska@suse.cz>
11461 * profile-count.h (enum profile_quality): Add
11462 profile_uninitialized as the first value. Do not number values
11463 as they are zero based.
11464 (profile_count::verify): Update sanity check.
11465 (profile_probability::verify): Likewise.
11467 2018-01-23 Nathan Sidwell <nathan@acm.org>
11469 * doc/invoke.texi (ffor-scope): Deprecate.
11471 2018-01-23 David Malcolm <dmalcolm@redhat.com>
11473 PR tree-optimization/83510
11474 * domwalk.c (set_all_edges_as_executable): New function.
11475 (dom_walker::dom_walker): Convert bool param
11476 "skip_unreachable_blocks" to enum reachability. Move setup of
11477 edge flags to set_all_edges_as_executable and only do it when
11478 reachability is REACHABLE_BLOCKS.
11479 * domwalk.h (enum dom_walker::reachability): New enum.
11480 (dom_walker::dom_walker): Convert bool param
11481 "skip_unreachable_blocks" to enum reachability.
11482 (set_all_edges_as_executable): New decl.
11483 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
11484 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
11486 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
11487 but converting true to REACHABLE_BLOCKS.
11488 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
11490 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
11491 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
11492 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
11494 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
11495 if check_all_array_refs will be called.
11497 2018-01-23 David Malcolm <dmalcolm@redhat.com>
11499 * tree.c (selftest::test_location_wrappers): Add more test
11502 2018-01-23 David Malcolm <dmalcolm@redhat.com>
11504 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
11505 (selftest::test_bit_in_range): Likewise.
11507 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
11510 * doc/sourcebuild.texi (vect_float): Say that the selector
11511 only describes the situation when -funsafe-math-optimizations is on.
11512 (vect_float_strict): Document.
11514 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
11516 PR tree-optimization/83965
11517 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11518 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
11519 instead of checking only for a reduction.
11520 (vect_recog_widen_sum_pattern): Likewise.
11522 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
11524 * predict.c (probably_never_executed): Only use precise profile info.
11525 (compute_function_frequency): Skip after inlining hack since we now
11526 have quality checking.
11528 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
11530 * profile-count.h (profile_probability::very_unlikely,
11531 profile_probability::unlikely, profile_probability::even): Set
11532 precision to guessed.
11534 2018-01-23 Richard Biener <rguenther@suse.de>
11536 PR tree-optimization/83963
11537 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
11538 Properly terminate dominator walk when crossing the exit edge not
11539 when visiting its source block.
11541 2018-01-23 Jakub Jelinek <jakub@redhat.com>
11544 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
11545 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
11547 2018-01-22 Jakub Jelinek <jakub@redhat.com>
11549 PR tree-optimization/83957
11550 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
11551 semicolon after for body surrounded by braces.
11553 PR tree-optimization/83081
11554 * profile-count.h (profile_probability::split): New method.
11555 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
11556 Use profile_probability::split.
11557 (do_compare_rtx_and_jump): Fix adjustment of probabilities
11558 when splitting a single conditional jump into 2.
11560 2018-01-22 David Malcolm <dmalcolm@redhat.com>
11562 PR tree-optimization/69452
11563 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
11566 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11568 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
11569 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
11570 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
11572 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11574 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
11576 * config/rl78/rl78.md (movdi): New define_expand.
11577 * config/rl78/rl78.c (rl78_split_movdi): New function.
11579 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11582 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
11584 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
11585 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
11586 128-bit to produce an UNSPEC move to get the double word with the
11587 signbit and then a shift directly to do signbit.
11588 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
11589 implementation with a new version that just does either a direct
11590 move or a regular move. Move memory interface to separate insns.
11591 Move insns so they are next to the expander.
11592 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
11593 with signbit move. Split big and little endian case.
11594 (signbit<mode>2_dm_mem_le): Likewise.
11595 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
11596 (signbit<mode>2_dm2): Likewise.
11598 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11600 * config/rl78/rl78.md (anddi3): New define_expand.
11602 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11604 * config/rl78/rl78.md (umindi3): New define_expand.
11606 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11608 * config/rl78/rl78.md (smindi3): New define_expand.
11610 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11612 * config/rl78/rl78.md (smaxdi3): New define_expand.
11614 2018-01-22 Carl Love <cel@us.ibm.com>
11616 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
11617 LVX_V1TI): Add macro expansion.
11618 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
11619 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
11620 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
11621 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
11622 Change check to determine if the instruction is a byte reversing
11623 entry. Fix typo in comment.
11624 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
11625 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
11626 Add def_builtin calls for new builtins.
11627 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
11628 Add define_insn expansion.
11630 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11632 * config/rl78/rl78.md (umaxdi3): New define_expand.
11634 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
11636 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
11637 for non-QImode registers.
11639 2018-01-22 Richard Biener <rguenther@suse.de>
11641 PR tree-optimization/83963
11642 * graphite-scop-detection.c (scop_detection::get_sese): Delay
11643 including the loop exit block.
11644 (scop_detection::merge_sese): Likewise.
11645 (scop_detection::add_scop): Do it here instead.
11647 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11649 * doc/sourcebuild.texi (arm_softfloat): Document.
11651 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
11654 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
11655 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
11656 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
11658 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11659 David Edelsohn <dje.gcc@gmail.com>
11662 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
11663 Change "crset eq" to "crset 2".
11664 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
11665 (*call_indirect_aix<mode>_nospec): Likewise.
11666 (*call_value_indirect_aix<mode>_nospec): Likewise.
11667 (*call_indirect_elfv2<mode>_nospec): Likewise.
11668 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
11669 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
11670 change assembly output from . to $.
11671 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
11672 (indirect_jump<mode>_nospec): Change assembly output from . to $.
11673 (*tablejump<mode>_internal1_nospec): Likewise.
11675 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
11678 * config/sh/sh_optimize_sett_clrt.cc:
11679 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
11681 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
11683 PR tree-optimization/83940
11684 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
11685 offset_dt to vect_constant_def rather than vect_unknown_def_type.
11686 (vect_check_load_store_mask): Add a mask_dt_out parameter and
11687 use it to pass back the definition type.
11688 (vect_check_store_rhs): Likewise rhs_dt_out.
11689 (vect_build_gather_load_calls): Add a mask_dt argument and use
11690 it instead of a call to vect_is_simple_use.
11691 (vectorizable_store): Update calls to vect_check_load_store_mask
11692 and vect_check_store_rhs. Use the dt returned by the latter instead
11693 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
11694 instead of calls to vect_is_simple_use. Pass the scalar rather
11695 than the vector operand to vect_is_simple_use when handling
11696 second and subsequent copies of an rhs value.
11697 (vectorizable_load): Update calls to vect_check_load_store_mask
11698 and vect_build_gather_load_calls. Use the cached mask_dt and
11699 gs_info.offset_dt instead of calls to vect_is_simple_use.
11701 2018-01-20 Jakub Jelinek <jakub@redhat.com>
11703 PR middle-end/83945
11704 * tree-emutls.c: Include gimplify.h.
11705 (lower_emutls_2): New function.
11706 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
11707 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
11708 it before further processing.
11711 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
11712 UINTVAL (trueop1) instead of INTVAL (op1).
11714 2018-01-19 Jakub Jelinek <jakub@redhat.com>
11718 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
11719 INCOMING_FRAME_SP_OFFSET if not defined.
11720 (scan_trace): Add ENTRY argument. If true and
11721 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
11722 emit a note to adjust the CFA offset.
11723 (create_cfi_notes): Adjust scan_trace callers.
11724 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
11725 INCOMING_FRAME_SP_OFFSET in the CIE.
11726 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
11727 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
11729 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
11730 * doc/tm.texi: Regenerated.
11732 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11734 PR rtl-optimization/83147
11735 * lra-constraints.c (remove_inheritance_pseudos): Use
11736 lra_substitute_pseudo_within_insn.
11738 2018-01-19 Tom de Vries <tom@codesourcery.com>
11739 Cesar Philippidis <cesar@codesourcery.com>
11742 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
11744 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
11747 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
11748 spaces for function labels.
11750 2018-01-19 Martin Liska <mliska@suse.cz>
11752 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
11753 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
11754 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
11755 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
11756 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
11757 (PRED_CONST_RETURN): Change from 69 to 65.
11758 (PRED_NULL_RETURN): Change from 91 to 71.
11759 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
11760 (PRED_LOOP_GUARD): Change from 66 to 73.
11762 2018-01-19 Martin Liska <mliska@suse.cz>
11764 * predict.c (predict_insn_def): Add new assert.
11765 (struct branch_predictor): Change type to signed integer.
11766 (test_prediction_value_range): Amend test to cover
11767 PROB_UNINITIALIZED.
11768 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
11769 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
11770 (PRED_LOOP_ITERATIONS_MAX): Likewise.
11771 (PRED_LOOP_IV_COMPARE): Likewise.
11772 * predict.h (PROB_UNINITIALIZED): Define new constant.
11774 2018-01-19 Martin Liska <mliska@suse.cz>
11776 * predict.c (dump_prediction): Add new format for
11777 analyze_brprob.py script which is enabled with -details
11779 * profile-count.h (precise_p): New function.
11781 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
11783 PR tree-optimization/83922
11784 * tree-vect-loop.c (vect_verify_full_masking): Return false if
11785 there are no statements that need masking.
11786 (vect_active_double_reduction_p): New function.
11787 (vect_analyze_loop_operations): Use it when handling phis that
11788 are not in the loop header.
11790 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
11792 PR tree-optimization/83914
11793 * tree-vect-loop.c (vectorizable_induction): Don't convert
11794 init_expr or apply the peeling adjustment for inductions
11795 that are nested within the vectorized loop.
11797 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11799 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
11802 2018-01-18 Jakub Jelinek <jakub@redhat.com>
11806 * function.h (gimplify_parameters): Add gimple_seq * argument.
11807 * function.c: Include gimple.h and options.h.
11808 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
11809 for the added local temporaries if needed.
11810 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
11811 if there are any parameter cleanups, wrap whole body into a
11812 try/finally with the cleanups.
11814 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
11817 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11818 Use GET_MODE_CLASS for scalar floating point.
11820 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
11824 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
11825 Fix call of call_cgraph_insertion_hooks.
11827 2018-01-18 Martin Sebor <msebor@redhat.com>
11829 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
11831 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
11834 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
11837 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
11840 * common.opt: (-ffile-prefix-map): New option.
11841 * opts.c (common_handle_option): Defer it.
11842 * opts-global.c (handle_common_deferred_options): Handle it.
11843 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
11844 * file-prefix-map.h: New file.
11845 (remap_debug_filename, add_debug_prefix_map): ...here.
11846 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
11847 * final.c (debug_prefix_map, add_debug_prefix_map
11848 remap_debug_filename): Move to...
11849 * file-prefix-map.c: New file.
11850 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
11851 generalize, get rid of alloca(), use strrchr() instead of strchr().
11852 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
11853 Implement in terms of add_prefix_map().
11854 (remap_macro_filename, remap_debug_filename): Implement in term of
11856 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
11857 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
11858 * dbxout.c: Include file-prefix-map.h.
11859 * varasm.c: Likewise.
11860 * vmsdbgout.c: Likewise.
11861 * xcoffout.c: Likewise.
11862 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
11863 * doc/cppopts.texi (-fmacro-prefix-map): Document.
11864 * doc/invoke.texi (-ffile-prefix-map): Document.
11865 (-fdebug-prefix-map): Update description.
11867 2018-01-18 Martin Liska <mliska@suse.cz>
11869 * config/i386/i386.c (indirect_thunk_name): Document that also
11871 (output_indirect_thunk): Document why both instructions
11872 (pause and lfence) are generated.
11874 2018-01-18 Richard Biener <rguenther@suse.de>
11876 PR tree-optimization/83887
11877 * graphite-scop-detection.c
11878 (scop_detection::get_nearest_dom_with_single_entry): Remove.
11879 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
11880 (scop_detection::merge_sese): Re-implement with a flood-fill
11881 algorithm that properly finds a SESE region if it exists.
11883 2018-01-18 Jakub Jelinek <jakub@redhat.com>
11886 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
11887 pointer_diff optimizations use view_convert instead of convert.
11889 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11891 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
11892 Generate different code for -mno-speculate-indirect-jumps.
11893 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
11894 (*call_indirect_aix<mode>): Disable for
11895 -mno-speculate-indirect-jumps.
11896 (*call_indirect_aix<mode>_nospec): New define_insn.
11897 (*call_value_indirect_aix<mode>): Disable for
11898 -mno-speculate-indirect-jumps.
11899 (*call_value_indirect_aix<mode>_nospec): New define_insn.
11900 (*sibcall_nonlocal_sysv<mode>): Generate different code for
11901 -mno-speculate-indirect-jumps.
11902 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
11904 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
11906 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
11907 long double type, set the flags for noting the default long double
11908 type, even if we don't pass or return a long double type.
11910 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
11913 * ipa-inline.c (flatten_function): Do not overwrite final inlining
11916 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
11918 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
11919 support for merge[hl].
11920 (fold_mergehl_helper): New helper function.
11921 (tree-vector-builder.h): New #include for tree_vector_builder usage.
11922 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
11923 (altivec_vmrglw_direct): Add xxmrglw insn.
11925 2018-01-17 Andrew Waterman <andrew@sifive.com>
11927 * config/riscv/riscv.c (riscv_conditional_register_usage): If
11928 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
11930 2018-01-17 David Malcolm <dmalcolm@redhat.com>
11933 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
11934 call the lto_location_cache before reading the
11935 DECL_SOURCE_LOCATION of the types.
11937 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
11938 Richard Sandiford <richard.sandiford@linaro.org>
11940 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
11941 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
11942 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
11943 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
11944 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
11946 * config/aarch64/constraints.md (aarch64_movti_operand):
11948 * config/aarch64/predicates.md (Uti): Add new constraint.
11950 2018-01-17 Carl Love <cel@us.ibm.com>
11952 * config/rs6000/vsx.md (define_expand xl_len_r,
11953 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
11954 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
11956 (define_expand, define_insn): Move the shift left from the
11957 define_insn to the define_expand for lxvl and stxvl instructions.
11958 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
11959 and XL_LEN_R definitions to PURE.
11961 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
11963 * config/i386/i386.c (indirect_thunk_name): Declare regno
11964 as unsigned int. Compare regno with INVALID_REGNUM.
11965 (output_indirect_thunk): Ditto.
11966 (output_indirect_thunk_function): Ditto.
11967 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
11968 in the call to output_indirect_thunk_function.
11970 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
11972 PR middle-end/83884
11973 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
11974 rather than the size of inner_type to determine the stack slot size
11975 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
11977 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
11980 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
11983 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
11985 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
11986 endian Linux systems to optionally enable multilibs for selecting
11987 the long double type if the user configured an explicit type.
11988 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
11989 have no long double multilibs if not defined.
11990 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
11991 warn if the user used -mabi={ieee,ibm}longdouble and we built
11992 multilibs for long double.
11993 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
11994 appropriate multilib option.
11995 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
11997 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
11998 for building long double multilibs.
11999 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
12001 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
12003 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
12006 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
12008 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
12011 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
12014 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
12017 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
12019 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
12020 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
12022 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
12024 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
12025 different rtl trees depending on TARGET_64BIT.
12026 (rs6000_gen_lvx): Likewise.
12028 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
12030 * config/visium/visium.md (nop): Tweak comment.
12031 (hazard_nop): Likewise.
12033 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12035 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
12036 -mspeculate-indirect-jumps.
12037 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
12038 for -mno-speculate-indirect-jumps.
12039 (*call_indirect_elfv2<mode>_nospec): New define_insn.
12040 (*call_value_indirect_elfv2<mode>): Disable for
12041 -mno-speculate-indirect-jumps.
12042 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
12043 (indirect_jump): Emit different RTL for
12044 -mno-speculate-indirect-jumps.
12045 (*indirect_jump<mode>): Disable for
12046 -mno-speculate-indirect-jumps.
12047 (*indirect_jump<mode>_nospec): New define_insn.
12048 (tablejump): Emit different RTL for
12049 -mno-speculate-indirect-jumps.
12050 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
12051 (tablejumpsi_nospec): New define_expand.
12052 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
12053 (tablejumpdi_nospec): New define_expand.
12054 (*tablejump<mode>_internal1): Disable for
12055 -mno-speculate-indirect-jumps.
12056 (*tablejump<mode>_internal1_nospec): New define_insn.
12057 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
12060 2018-01-16 Artyom Skrobov tyomitch@gmail.com
12062 * caller-save.c (insert_save): Drop unnecessary parameter. All
12065 2018-01-16 Jakub Jelinek <jakub@redhat.com>
12066 Richard Biener <rguenth@suse.de>
12069 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
12070 return early, inline manually is_gimple_sizepos. Make sure if we
12071 call gimplify_expr we don't end up with a gimple constant.
12072 * tree.c (variably_modified_type_p): Don't return true for
12073 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
12074 * gimplify.h (is_gimple_sizepos): Remove.
12076 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
12078 PR tree-optimization/83857
12079 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
12080 vectorizable_live_operation for pure SLP statements.
12081 (vectorizable_live_operation): Handle PHIs.
12083 2018-01-16 Richard Biener <rguenther@suse.de>
12085 PR tree-optimization/83867
12086 * tree-vect-stmts.c (vect_transform_stmt): Precompute
12087 nested_in_vect_loop_p since the scalar stmt may get invalidated.
12089 2018-01-16 Jakub Jelinek <jakub@redhat.com>
12092 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
12093 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
12094 If off is not INTEGER_CST, issue a may not be aligned warning
12095 rather than isn't aligned. Use isn%'t rather than isn't.
12096 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
12098 <case MULT_EXPR>: Improve the case when bottom and one of the
12099 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
12100 operand, in that case check if the other operand is multiple of
12101 bottom divided by the INTEGER_CST operand.
12103 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
12106 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
12107 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
12108 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
12109 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
12110 * config/pa/pa.c (pa_function_arg_advance): Likewise.
12111 (pa_function_arg, pa_arg_partial_bytes): Likewise.
12112 (pa_function_arg_size): New function.
12114 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
12116 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
12117 in a separate statement.
12119 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
12121 PR tree-optimization/83847
12122 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
12123 group gathers and scatters.
12125 2018-01-16 Jakub Jelinek <jakub@redhat.com>
12127 PR rtl-optimization/86620
12128 * params.def (max-sched-ready-insns): Bump minimum value to 1.
12130 PR rtl-optimization/83213
12131 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
12132 to last if both are JUMP_INSNs.
12134 PR tree-optimization/83843
12135 * gimple-ssa-store-merging.c
12136 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
12137 store_immediate_info for bswap/nop orig_stores.
12139 2018-01-15 Andrew Waterman <andrew@sifive.com>
12141 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
12143 <UDIV>: Increase cost if !TARGET_DIV.
12145 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
12147 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
12148 (define_attr "cr_logical_3op"): New.
12149 (cceq_ior_compare): Adjust.
12150 (cceq_ior_compare_complement): Adjust.
12151 (*cceq_rev_compare): Adjust.
12152 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
12153 (is_cracked_insn): Adjust.
12154 (insn_must_be_first_in_group): Adjust.
12155 * config/rs6000/40x.md: Adjust.
12156 * config/rs6000/440.md: Adjust.
12157 * config/rs6000/476.md: Adjust.
12158 * config/rs6000/601.md: Adjust.
12159 * config/rs6000/603.md: Adjust.
12160 * config/rs6000/6xx.md: Adjust.
12161 * config/rs6000/7450.md: Adjust.
12162 * config/rs6000/7xx.md: Adjust.
12163 * config/rs6000/8540.md: Adjust.
12164 * config/rs6000/cell.md: Adjust.
12165 * config/rs6000/e300c2c3.md: Adjust.
12166 * config/rs6000/e500mc.md: Adjust.
12167 * config/rs6000/e500mc64.md: Adjust.
12168 * config/rs6000/e5500.md: Adjust.
12169 * config/rs6000/e6500.md: Adjust.
12170 * config/rs6000/mpc.md: Adjust.
12171 * config/rs6000/power4.md: Adjust.
12172 * config/rs6000/power5.md: Adjust.
12173 * config/rs6000/power6.md: Adjust.
12174 * config/rs6000/power7.md: Adjust.
12175 * config/rs6000/power8.md: Adjust.
12176 * config/rs6000/power9.md: Adjust.
12177 * config/rs6000/rs64.md: Adjust.
12178 * config/rs6000/titan.md: Adjust.
12180 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
12182 * config/i386/predicates.md (indirect_branch_operand): Rewrite
12183 ix86_indirect_branch_register logic.
12185 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
12187 * config/i386/constraints.md (Bs): Update
12188 ix86_indirect_branch_register check. Don't check
12189 ix86_indirect_branch_register with GOT_memory_operand.
12191 * config/i386/predicates.md (GOT_memory_operand): Don't check
12192 ix86_indirect_branch_register here.
12193 (GOT32_symbol_operand): Likewise.
12195 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
12197 * config/i386/predicates.md (constant_call_address_operand):
12198 Rewrite ix86_indirect_branch_register logic.
12199 (sibcall_insn_operand): Likewise.
12201 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
12203 * config/i386/constraints.md (Bs): Replace
12204 ix86_indirect_branch_thunk_register with
12205 ix86_indirect_branch_register.
12207 * config/i386/i386.md (indirect_jump): Likewise.
12208 (tablejump): Likewise.
12209 (*sibcall_memory): Likewise.
12210 (*sibcall_value_memory): Likewise.
12211 Peepholes of indirect call and jump via memory: Likewise.
12212 * config/i386/i386.opt: Likewise.
12213 * config/i386/predicates.md (indirect_branch_operand): Likewise.
12214 (GOT_memory_operand): Likewise.
12215 (call_insn_operand): Likewise.
12216 (sibcall_insn_operand): Likewise.
12217 (GOT32_symbol_operand): Likewise.
12219 2018-01-15 Jakub Jelinek <jakub@redhat.com>
12221 PR middle-end/83837
12222 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
12223 type rather than type addr's type points to.
12224 (expand_omp_atomic_mutex): Likewise.
12225 (expand_omp_atomic): Likewise.
12227 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
12230 * config/i386/i386.c (output_indirect_thunk_function): Use
12231 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
12232 for __x86_return_thunk.
12234 2018-01-15 Richard Biener <rguenther@suse.de>
12236 PR middle-end/83850
12237 * expmed.c (extract_bit_field_1): Fix typo.
12239 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12242 * config/arm/iterators.md (VF): New mode iterator.
12243 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
12244 Remove integer-related logic from pattern.
12245 (neon_vabd<mode>_3): Likewise.
12247 2018-01-15 Jakub Jelinek <jakub@redhat.com>
12249 PR middle-end/82694
12250 * common.opt (fstrict-overflow): No longer an alias.
12251 (fwrapv-pointer): New option.
12252 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
12253 also for pointer types based on flag_wrapv_pointer.
12254 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
12255 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
12256 opts->x_flag_wrapv got set.
12257 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
12258 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
12259 POINTER_TYPE_OVERFLOW_UNDEFINED.
12260 * match.pd: Likewise in address comparison pattern.
12261 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
12263 2018-01-15 Richard Biener <rguenther@suse.de>
12266 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
12267 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
12268 Reset type names to their identifier if their TYPE_DECL doesn't
12269 have linkage (and thus is used for ODR and devirt).
12270 (save_debug_info_for_decl): Remove.
12271 (save_debug_info_for_type): Likewise.
12272 (add_tree_to_fld_list): Adjust.
12273 * tree-pretty-print.c (dump_generic_node): Make dumping of
12274 type names more robust.
12276 2018-01-15 Richard Biener <rguenther@suse.de>
12278 * BASE-VER: Bump to 8.0.1.
12280 2018-01-14 Martin Sebor <msebor@redhat.com>
12283 * builtins.c (check_access): Avoid warning when the no-warning bit
12286 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
12288 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
12289 * ira-color (allocno_hard_regs_compare): Likewise.
12291 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
12294 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
12295 Use .pushsection/.popsection.
12297 2018-01-14 Martin Sebor <msebor@redhat.com>
12300 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
12302 2018-01-14 Jakub Jelinek <jakub@redhat.com>
12304 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
12305 entry from extra_headers.
12306 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
12307 extra_headers, make the list bitwise identical to the i?86-*-* one.
12309 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
12311 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
12312 -mcmodel=large with -mindirect-branch=thunk,
12313 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
12314 -mfunction-return=thunk-extern.
12315 * doc/invoke.texi: Document -mcmodel=large is incompatible with
12316 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
12317 -mfunction-return=thunk and -mfunction-return=thunk-extern.
12319 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
12321 * config/i386/i386.c (print_reg): Print the name of the full
12322 integer register without '%'.
12323 (ix86_print_operand): Handle 'V'.
12324 * doc/extend.texi: Document 'V' modifier.
12326 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
12328 * config/i386/constraints.md (Bs): Disallow memory operand for
12329 -mindirect-branch-register.
12331 * config/i386/predicates.md (indirect_branch_operand): Likewise.
12332 (GOT_memory_operand): Likewise.
12333 (call_insn_operand): Likewise.
12334 (sibcall_insn_operand): Likewise.
12335 (GOT32_symbol_operand): Likewise.
12336 * config/i386/i386.md (indirect_jump): Call convert_memory_address
12337 for -mindirect-branch-register.
12338 (tablejump): Likewise.
12339 (*sibcall_memory): Likewise.
12340 (*sibcall_value_memory): Likewise.
12341 Disallow peepholes of indirect call and jump via memory for
12342 -mindirect-branch-register.
12343 (*call_pop): Replace m with Bw.
12344 (*call_value_pop): Likewise.
12345 (*sibcall_pop_memory): Replace m with Bs.
12346 * config/i386/i386.opt (mindirect-branch-register): New option.
12347 * doc/invoke.texi: Document -mindirect-branch-register option.
12349 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
12351 * config/i386/i386-protos.h (ix86_output_function_return): New.
12352 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
12353 set function_return_type.
12354 (indirect_thunk_name): Add ret_p to indicate thunk for function
12356 (output_indirect_thunk_function): Pass false to
12357 indirect_thunk_name.
12358 (ix86_output_indirect_branch_via_reg): Likewise.
12359 (ix86_output_indirect_branch_via_push): Likewise.
12360 (output_indirect_thunk_function): Create alias for function
12361 return thunk if regno < 0.
12362 (ix86_output_function_return): New function.
12363 (ix86_handle_fndecl_attribute): Handle function_return.
12364 (ix86_attribute_table): Add function_return.
12365 * config/i386/i386.h (machine_function): Add
12366 function_return_type.
12367 * config/i386/i386.md (simple_return_internal): Use
12368 ix86_output_function_return.
12369 (simple_return_internal_long): Likewise.
12370 * config/i386/i386.opt (mfunction-return=): New option.
12371 (indirect_branch): Mention -mfunction-return=.
12372 * doc/extend.texi: Document function_return function attribute.
12373 * doc/invoke.texi: Document -mfunction-return= option.
12375 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
12377 * config/i386/i386-opts.h (indirect_branch): New.
12378 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
12379 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
12380 with local indirect jump when converting indirect call and jump.
12381 (ix86_set_indirect_branch_type): New.
12382 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
12383 (indirectlabelno): New.
12384 (indirect_thunk_needed): Likewise.
12385 (indirect_thunk_bnd_needed): Likewise.
12386 (indirect_thunks_used): Likewise.
12387 (indirect_thunks_bnd_used): Likewise.
12388 (INDIRECT_LABEL): Likewise.
12389 (indirect_thunk_name): Likewise.
12390 (output_indirect_thunk): Likewise.
12391 (output_indirect_thunk_function): Likewise.
12392 (ix86_output_indirect_branch_via_reg): Likewise.
12393 (ix86_output_indirect_branch_via_push): Likewise.
12394 (ix86_output_indirect_branch): Likewise.
12395 (ix86_output_indirect_jmp): Likewise.
12396 (ix86_code_end): Call output_indirect_thunk_function if needed.
12397 (ix86_output_call_insn): Call ix86_output_indirect_branch if
12399 (ix86_handle_fndecl_attribute): Handle indirect_branch.
12400 (ix86_attribute_table): Add indirect_branch.
12401 * config/i386/i386.h (machine_function): Add indirect_branch_type
12402 and has_local_indirect_jump.
12403 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
12405 (tablejump): Likewise.
12406 (*indirect_jump): Use ix86_output_indirect_jmp.
12407 (*tablejump_1): Likewise.
12408 (simple_return_indirect_internal): Likewise.
12409 * config/i386/i386.opt (mindirect-branch=): New option.
12410 (indirect_branch): New.
12413 (thunk-inline): Likewise.
12414 (thunk-extern): Likewise.
12415 * doc/extend.texi: Document indirect_branch function attribute.
12416 * doc/invoke.texi: Document -mindirect-branch= option.
12418 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
12421 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
12423 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
12425 * ipa-inline.c (want_inline_small_function_p): Return false if
12426 inlining has already failed with CIF_FINAL_ERROR.
12427 (update_caller_keys): Call want_inline_small_function_p before
12429 (update_callee_keys): Likewise.
12431 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
12433 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
12435 (rs6000_quadword_masked_address_p): Likewise.
12436 (quad_aligned_load_p): Likewise.
12437 (quad_aligned_store_p): Likewise.
12438 (const_load_sequence_p): Add comment to describe the outer-most loop.
12439 (mimic_memory_attributes_and_flags): New function.
12440 (rs6000_gen_stvx): Likewise.
12441 (replace_swapped_aligned_store): Likewise.
12442 (rs6000_gen_lvx): Likewise.
12443 (replace_swapped_aligned_load): Likewise.
12444 (replace_swapped_load_constant): Capitalize argument name in
12445 comment describing this function.
12446 (rs6000_analyze_swaps): Add a third pass to search for vector loads
12447 and stores that access quad-word aligned addresses and replace
12448 with stvx or lvx instructions when appropriate.
12449 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
12450 New function prototype.
12451 (rs6000_quadword_masked_address_p): Likewise.
12452 (rs6000_gen_lvx): Likewise.
12453 (rs6000_gen_stvx): Likewise.
12454 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
12455 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
12456 when memory address is aligned.
12457 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
12458 this split to select lvx instruction when memory address is aligned.
12459 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
12460 instruction when memory address is aligned.
12461 (*vsx_le_perm_load_v16qi): Likewise.
12462 (four unnamed splitters): Modify to select the stvx instruction
12463 when memory is aligned.
12465 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
12467 * predict.c (determine_unlikely_bbs): Handle correctly BBs
12468 which appears in the queue multiple times.
12470 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12471 Alan Hayward <alan.hayward@arm.com>
12472 David Sherwood <david.sherwood@arm.com>
12474 * tree-vectorizer.h (vec_lower_bound): New structure.
12475 (_loop_vec_info): Add check_nonzero and lower_bounds.
12476 (LOOP_VINFO_CHECK_NONZERO): New macro.
12477 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
12478 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
12479 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
12480 fields. Make seg_len the distance travelled, not including the
12482 (dr_direction_indicator): Declare.
12483 (dr_zero_step_indicator): Likewise.
12484 (dr_known_forward_stride_p): Likewise.
12485 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
12487 (runtime_alias_check_p): Allow runtime alias checks with
12489 (operator ==): Compare access_size and align.
12490 (prune_runtime_alias_test_list): Rework for new distinction between
12491 the access_size and seg_len.
12492 (create_intersect_range_checks_index): Likewise. Cope with polynomial
12494 (get_segment_min_max): New function.
12495 (create_intersect_range_checks): Use it.
12496 (dr_step_indicator): New function.
12497 (dr_direction_indicator): Likewise.
12498 (dr_zero_step_indicator): Likewise.
12499 (dr_known_forward_stride_p): Likewise.
12500 * tree-loop-distribution.c (data_ref_segment_size): Return
12501 DR_STEP * (niters - 1).
12502 (compute_alias_check_pairs): Update call to the dr_with_seg_len
12504 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
12505 (vect_preserves_scalar_order_p): New function, split out from...
12506 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
12507 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
12508 (vect_vfa_access_size): New function.
12509 (vect_vfa_align): Likewise.
12510 (vect_compile_time_alias): Take access_size_a and access_b arguments.
12511 (dump_lower_bound): New function.
12512 (vect_check_lower_bound): Likewise.
12513 (vect_small_gap_p): Likewise.
12514 (vectorizable_with_step_bound_p): Likewise.
12515 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
12516 depencies if the vectorization factor is 1. Convert the checks
12517 for nonzero steps into checks on the bounds of DR_STEP. Try using
12518 a bunds check for variable steps if the minimum required step is
12519 relatively small. Update calls to the dr_with_seg_len
12520 constructor and to vect_compile_time_alias.
12521 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
12523 (vect_loop_versioning): Call it.
12524 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
12526 (vect_estimate_min_profitable_iters): Account for any bounds checks.
12528 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12529 Alan Hayward <alan.hayward@arm.com>
12530 David Sherwood <david.sherwood@arm.com>
12532 * doc/sourcebuild.texi (vect_scatter_store): Document.
12533 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
12535 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
12537 * genopinit.c (main): Add supports_vec_scatter_store and
12538 supports_vec_scatter_store_cached to target_optabs.
12539 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
12540 IFN_MASK_SCATTER_STORE.
12541 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
12543 * internal-fn.h (internal_store_fn_p): Declare.
12544 (internal_fn_stored_value_index): Likewise.
12545 * internal-fn.c (scatter_store_direct): New macro.
12546 (expand_scatter_store_optab_fn): New function.
12547 (direct_scatter_store_optab_supported_p): New macro.
12548 (internal_store_fn_p): New function.
12549 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
12550 IFN_MASK_SCATTER_STORE.
12551 (internal_fn_mask_index): Likewise.
12552 (internal_fn_stored_value_index): New function.
12553 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
12554 for scatter stores.
12555 * optabs-query.h (supports_vec_scatter_store_p): Declare.
12556 * optabs-query.c (supports_vec_scatter_store_p): New function.
12557 * tree-vectorizer.h (vect_get_store_rhs): Declare.
12558 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
12559 true for scatter stores.
12560 (vect_gather_scatter_fn_p): Handle scatter stores too.
12561 (vect_check_gather_scatter): Consider using scatter stores if
12562 supports_vec_scatter_store_p.
12563 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
12564 scatter stores too.
12565 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
12566 internal_fn_stored_value_index.
12567 (check_load_store_masking): Handle scatter stores too.
12568 (vect_get_store_rhs): Make public.
12569 (vectorizable_call): Use internal_store_fn_p.
12570 (vectorizable_store): Handle scatter store internal functions.
12571 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
12572 when deciding whether the end of the group has been reached.
12573 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
12574 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
12575 (mask_scatter_store<mode>): New insns.
12577 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12578 Alan Hayward <alan.hayward@arm.com>
12579 David Sherwood <david.sherwood@arm.com>
12581 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
12582 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
12583 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
12585 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
12586 Use vect_truncate_gather_scatter_offset if we can't treat the
12587 operation as a normal gather load or scatter store.
12588 (get_group_load_store_type): Take the gather_scatter_info
12589 as argument. Try using a gather load or scatter store for
12590 single-element groups.
12591 (get_load_store_type): Update calls to get_group_load_store_type
12592 and vect_use_strided_gather_scatters_p.
12594 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12595 Alan Hayward <alan.hayward@arm.com>
12596 David Sherwood <david.sherwood@arm.com>
12598 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
12599 optional tree argument.
12600 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
12602 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
12603 but continue to use the current value as a fallback.
12604 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
12605 to compare the updates.
12606 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
12607 (get_load_store_type): Use it when handling a strided access.
12608 (vect_get_strided_load_store_ops): New function.
12609 (vect_get_data_ptr_increment): Likewise.
12610 (vectorizable_load): Handle strided gather loads. Always pass
12611 a step to vect_create_data_ref_ptr and bump_vector_ptr.
12613 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12614 Alan Hayward <alan.hayward@arm.com>
12615 David Sherwood <david.sherwood@arm.com>
12617 * doc/md.texi (gather_load@var{m}): Document.
12618 (mask_gather_load@var{m}): Likewise.
12619 * genopinit.c (main): Add supports_vec_gather_load and
12620 supports_vec_gather_load_cached to target_optabs.
12621 * optabs-tree.c (init_tree_optimization_optabs): Use
12622 ggc_cleared_alloc to allocate target_optabs.
12623 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
12624 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
12626 * internal-fn.h (internal_load_fn_p): Declare.
12627 (internal_gather_scatter_fn_p): Likewise.
12628 (internal_fn_mask_index): Likewise.
12629 (internal_gather_scatter_fn_supported_p): Likewise.
12630 * internal-fn.c (gather_load_direct): New macro.
12631 (expand_gather_load_optab_fn): New function.
12632 (direct_gather_load_optab_supported_p): New macro.
12633 (direct_internal_fn_optab): New function.
12634 (internal_load_fn_p): Likewise.
12635 (internal_gather_scatter_fn_p): Likewise.
12636 (internal_fn_mask_index): Likewise.
12637 (internal_gather_scatter_fn_supported_p): Likewise.
12638 * optabs-query.c (supports_at_least_one_mode_p): New function.
12639 (supports_vec_gather_load_p): Likewise.
12640 * optabs-query.h (supports_vec_gather_load_p): Declare.
12641 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
12642 and memory_type field.
12643 (NUM_PATTERNS): Bump to 15.
12644 * tree-vect-data-refs.c: Include internal-fn.h.
12645 (vect_gather_scatter_fn_p): New function.
12646 (vect_describe_gather_scatter_call): Likewise.
12647 (vect_check_gather_scatter): Try using internal functions for
12648 gather loads. Recognize existing calls to a gather load function.
12649 (vect_analyze_data_refs): Consider using gather loads if
12650 supports_vec_gather_load_p.
12651 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
12652 (vect_get_gather_scatter_offset_type): Likewise.
12653 (vect_convert_mask_for_vectype): Likewise.
12654 (vect_add_conversion_to_patterm): Likewise.
12655 (vect_try_gather_scatter_pattern): Likewise.
12656 (vect_recog_gather_scatter_pattern): New pattern recognizer.
12657 (vect_vect_recog_func_ptrs): Add it.
12658 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
12659 internal_fn_mask_index and internal_gather_scatter_fn_p.
12660 (check_load_store_masking): Take the gather_scatter_info as an
12661 argument and handle gather loads.
12662 (vect_get_gather_scatter_ops): New function.
12663 (vectorizable_call): Check internal_load_fn_p.
12664 (vectorizable_load): Likewise. Handle gather load internal
12666 (vectorizable_store): Update call to check_load_store_masking.
12667 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
12668 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
12669 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
12670 (aarch64_gather_scale_operand_d): New predicates.
12671 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
12672 (mask_gather_load<mode>): New insns.
12674 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12675 Alan Hayward <alan.hayward@arm.com>
12676 David Sherwood <david.sherwood@arm.com>
12678 * optabs.def (fold_left_plus_optab): New optab.
12679 * doc/md.texi (fold_left_plus_@var{m}): Document.
12680 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
12681 * internal-fn.c (fold_left_direct): Define.
12682 (expand_fold_left_optab_fn): Likewise.
12683 (direct_fold_left_optab_supported_p): Likewise.
12684 * fold-const-call.c (fold_const_fold_left): New function.
12685 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
12686 * tree-parloops.c (valid_reduction_p): New function.
12687 (gather_scalar_reductions): Use it.
12688 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
12689 (vect_finish_replace_stmt): Declare.
12690 * tree-vect-loop.c (fold_left_reduction_fn): New function.
12691 (needs_fold_left_reduction_p): New function, split out from...
12692 (vect_is_simple_reduction): ...here. Accept reductions that
12693 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
12694 (vect_force_simple_reduction): Also store the reduction type in
12695 the assignment's STMT_VINFO_REDUC_TYPE.
12696 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
12697 (merge_with_identity): New function.
12698 (vect_expand_fold_left): Likewise.
12699 (vectorize_fold_left_reduction): Likewise.
12700 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
12701 scalar phi in place for it. Check for target support and reject
12702 cases that would reassociate the operation. Defer the transform
12703 phase to vectorize_fold_left_reduction.
12704 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
12705 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
12706 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
12708 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12710 * tree-if-conv.c (predicate_mem_writes): Remove redundant
12711 call to ifc_temp_var.
12713 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12714 Alan Hayward <alan.hayward@arm.com>
12715 David Sherwood <david.sherwood@arm.com>
12717 * target.def (legitimize_address_displacement): Take the original
12718 offset as a poly_int.
12719 * targhooks.h (default_legitimize_address_displacement): Update
12721 * targhooks.c (default_legitimize_address_displacement): Likewise.
12722 * doc/tm.texi: Regenerate.
12723 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
12724 as an argument, moving assert of ad->disp == ad->disp_term to...
12725 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
12726 Try calling targetm.legitimize_address_displacement before expanding
12727 the address rather than afterwards, and adjust for the new interface.
12728 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
12729 Match the new hook interface. Handle SVE addresses.
12730 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
12731 new hook interface.
12733 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12735 * Makefile.in (OBJS): Add early-remat.o.
12736 * target.def (select_early_remat_modes): New hook.
12737 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
12738 * doc/tm.texi: Regenerate.
12739 * targhooks.h (default_select_early_remat_modes): Declare.
12740 * targhooks.c (default_select_early_remat_modes): New function.
12741 * timevar.def (TV_EARLY_REMAT): New timevar.
12742 * passes.def (pass_early_remat): New pass.
12743 * tree-pass.h (make_pass_early_remat): Declare.
12744 * early-remat.c: New file.
12745 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
12747 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
12749 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12750 Alan Hayward <alan.hayward@arm.com>
12751 David Sherwood <david.sherwood@arm.com>
12753 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
12754 vfm1 with a bound_epilog parameter.
12755 (vect_do_peeling): Update calls accordingly, and move the prologue
12756 call earlier in the function. Treat the base bound_epilog as 0 for
12757 fully-masked loops and retain vf - 1 for other loops. Add 1 to
12758 this base when peeling for gaps.
12759 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
12760 with fully-masked loops.
12761 (vect_estimate_min_profitable_iters): Handle the single peeled
12762 iteration in that case.
12764 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12765 Alan Hayward <alan.hayward@arm.com>
12766 David Sherwood <david.sherwood@arm.com>
12768 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
12769 single-element interleaving even if the size is not a power of 2.
12770 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
12771 accesses for single-element interleaving if the group size is
12774 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12775 Alan Hayward <alan.hayward@arm.com>
12776 David Sherwood <david.sherwood@arm.com>
12778 * doc/md.texi (fold_extract_last_@var{m}): Document.
12779 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
12780 * optabs.def (fold_extract_last_optab): New optab.
12781 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
12782 * internal-fn.c (fold_extract_direct): New macro.
12783 (expand_fold_extract_optab_fn): Likewise.
12784 (direct_fold_extract_optab_supported_p): Likewise.
12785 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
12786 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12787 EXTRACT_LAST_REDUCTION.
12788 (get_initial_def_for_reduction): Do not create an initial vector
12789 for EXTRACT_LAST_REDUCTION reductions.
12790 (vectorizable_reduction): Leave the scalar phi in place for
12791 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
12792 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
12793 epilogue code for EXTRACT_LAST_REDUCTION and defer the
12794 transform phase to vectorizable_condition.
12795 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
12797 (vect_finish_stmt_generation): ...here.
12798 (vect_finish_replace_stmt): New function.
12799 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
12800 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
12802 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
12804 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12805 Alan Hayward <alan.hayward@arm.com>
12806 David Sherwood <david.sherwood@arm.com>
12808 * doc/md.texi (extract_last_@var{m}): Document.
12809 * optabs.def (extract_last_optab): New optab.
12810 * internal-fn.def (EXTRACT_LAST): New internal function.
12811 * internal-fn.c (cond_unary_direct): New macro.
12812 (expand_cond_unary_optab_fn): Likewise.
12813 (direct_cond_unary_optab_supported_p): Likewise.
12814 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
12815 loops using EXTRACT_LAST.
12816 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
12817 (extract_last_<mode>): ...this optab.
12818 (vec_extract<mode><Vel>): Update accordingly.
12820 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12821 Alan Hayward <alan.hayward@arm.com>
12822 David Sherwood <david.sherwood@arm.com>
12824 * target.def (empty_mask_is_expensive): New hook.
12825 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
12826 * doc/tm.texi: Regenerate.
12827 * targhooks.h (default_empty_mask_is_expensive): Declare.
12828 * targhooks.c (default_empty_mask_is_expensive): New function.
12829 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
12830 if the target says that empty masks are expensive.
12831 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
12833 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
12835 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12836 Alan Hayward <alan.hayward@arm.com>
12837 David Sherwood <david.sherwood@arm.com>
12839 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
12840 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
12841 (vect_use_loop_mask_for_alignment_p): New function.
12842 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
12843 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
12844 niters_skip argument. Make sure that the first niters_skip elements
12845 of the first iteration are inactive.
12846 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
12847 Update call to vect_set_loop_masks_directly.
12848 (get_misalign_in_elems): New function, split out from...
12849 (vect_gen_prolog_loop_niters): ...here.
12850 (vect_update_init_of_dr): Take a code argument that specifies whether
12851 the adjustment should be added or subtracted.
12852 (vect_update_init_of_drs): Likewise.
12853 (vect_prepare_for_masked_peels): New function.
12854 (vect_do_peeling): Skip prologue peeling if we're using a mask
12855 instead. Update call to vect_update_inits_of_drs.
12856 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12858 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
12859 alignment. Do not include the number of peeled iterations in
12860 the minimum threshold in that case.
12861 (vectorizable_induction): Adjust the start value down by
12862 LOOP_VINFO_MASK_SKIP_NITERS iterations.
12863 (vect_transform_loop): Call vect_prepare_for_masked_peels.
12864 Take the number of skipped iterations into account when calculating
12866 * tree-vect-stmts.c (vect_gen_while_not): New function.
12868 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12869 Alan Hayward <alan.hayward@arm.com>
12870 David Sherwood <david.sherwood@arm.com>
12872 * doc/sourcebuild.texi (vect_fully_masked): Document.
12873 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
12874 default value to 0.
12875 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
12877 (vect_analyze_loop_2): ...here. Don't check the vectorization
12878 factor against the number of loop iterations if the loop is
12881 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12882 Alan Hayward <alan.hayward@arm.com>
12883 David Sherwood <david.sherwood@arm.com>
12885 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
12886 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
12887 (dump_groups): Update accordingly.
12888 (iv_use::mem_type): New member variable.
12889 (address_p): New function.
12890 (record_use): Add a mem_type argument and initialize the new
12892 (record_group_use): Add a mem_type argument. Use address_p.
12893 Remove obsolete null checks of base_object. Update call to record_use.
12894 (find_interesting_uses_op): Update call to record_group_use.
12895 (find_interesting_uses_cond): Likewise.
12896 (find_interesting_uses_address): Likewise.
12897 (get_mem_type_for_internal_fn): New function.
12898 (find_address_like_use): Likewise.
12899 (find_interesting_uses_stmt): Try find_address_like_use before
12900 calling find_interesting_uses_op.
12901 (addr_offset_valid_p): Use the iv mem_type field as the type
12902 of the addressed memory.
12903 (add_autoinc_candidates): Likewise.
12904 (get_address_cost): Likewise.
12905 (split_small_address_groups_p): Use address_p.
12906 (split_address_groups): Likewise.
12907 (add_iv_candidate_for_use): Likewise.
12908 (autoinc_possible_for_pair): Likewise.
12909 (rewrite_groups): Likewise.
12910 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
12911 (determine_group_iv_cost): Update after split of USE_ADDRESS.
12912 (get_alias_ptr_type_for_ptr_address): New function.
12913 (rewrite_use_address): Rewrite address uses in calls that were
12914 identified by find_address_like_use.
12916 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12917 Alan Hayward <alan.hayward@arm.com>
12918 David Sherwood <david.sherwood@arm.com>
12920 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
12922 * gimple-expr.h (is_gimple_addressable: Likewise.
12923 * gimple-expr.c (is_gimple_address): Likewise.
12924 * internal-fn.c (expand_call_mem_ref): New function.
12925 (expand_mask_load_optab_fn): Use it.
12926 (expand_mask_store_optab_fn): Likewise.
12928 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12929 Alan Hayward <alan.hayward@arm.com>
12930 David Sherwood <david.sherwood@arm.com>
12932 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
12933 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
12934 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
12935 (cond_umax@var{mode}): Document.
12936 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
12937 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
12938 (cond_umin_optab, cond_umax_optab): New optabs.
12939 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
12940 (COND_IOR, COND_XOR): New internal functions.
12941 * internal-fn.h (get_conditional_internal_fn): Declare.
12942 * internal-fn.c (cond_binary_direct): New macro.
12943 (expand_cond_binary_optab_fn): Likewise.
12944 (direct_cond_binary_optab_supported_p): Likewise.
12945 (get_conditional_internal_fn): New function.
12946 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
12947 Cope with reduction statements that are vectorized as calls rather
12949 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
12950 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
12951 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
12952 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
12953 (UNSPEC_COND_EOR): New unspecs.
12954 (optab): Add mappings for them.
12955 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
12956 (sve_int_op, sve_fp_op): New int attributes.
12958 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
12959 Alan Hayward <alan.hayward@arm.com>
12960 David Sherwood <david.sherwood@arm.com>
12962 * optabs.def (while_ult_optab): New optab.
12963 * doc/md.texi (while_ult@var{m}@var{n}): Document.
12964 * internal-fn.def (WHILE_ULT): New internal function.
12965 * internal-fn.h (direct_internal_fn_supported_p): New override
12966 that takes two types as argument.
12967 * internal-fn.c (while_direct): New macro.
12968 (expand_while_optab_fn): New function.
12969 (convert_optab_supported_p): Likewise.
12970 (direct_while_optab_supported_p): New macro.
12971 * wide-int.h (wi::udiv_ceil): New function.
12972 * tree-vectorizer.h (rgroup_masks): New structure.
12973 (vec_loop_masks): New typedef.
12974 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
12975 and fully_masked_p.
12976 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
12977 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
12978 (vect_max_vf): New function.
12979 (slpeel_make_loop_iterate_ntimes): Delete.
12980 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
12981 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
12982 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
12983 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
12984 internal-fn.h, stor-layout.h and optabs-query.h.
12985 (vect_set_loop_mask): New function.
12986 (add_preheader_seq): Likewise.
12987 (add_header_seq): Likewise.
12988 (interleave_supported_p): Likewise.
12989 (vect_maybe_permute_loop_masks): Likewise.
12990 (vect_set_loop_masks_directly): Likewise.
12991 (vect_set_loop_condition_masked): Likewise.
12992 (vect_set_loop_condition_unmasked): New function, split out from
12993 slpeel_make_loop_iterate_ntimes.
12994 (slpeel_make_loop_iterate_ntimes): Rename to..
12995 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
12996 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
12997 (vect_do_peeling): Update call accordingly.
12998 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
13000 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13001 mask_compare_type, can_fully_mask_p and fully_masked_p.
13002 (release_vec_loop_masks): New function.
13003 (_loop_vec_info): Use it to free the loop masks.
13004 (can_produce_all_loop_masks_p): New function.
13005 (vect_get_max_nscalars_per_iter): Likewise.
13006 (vect_verify_full_masking): Likewise.
13007 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
13008 retries, and free the mask rgroups before retrying. Check loop-wide
13009 reasons for disallowing fully-masked loops. Make the final decision
13010 about whether use a fully-masked loop or not.
13011 (vect_estimate_min_profitable_iters): Do not assume that peeling
13012 for the number of iterations will be needed for fully-masked loops.
13013 (vectorizable_reduction): Disable fully-masked loops.
13014 (vectorizable_live_operation): Likewise.
13015 (vect_halve_mask_nunits): New function.
13016 (vect_double_mask_nunits): Likewise.
13017 (vect_record_loop_mask): Likewise.
13018 (vect_get_loop_mask): Likewise.
13019 (vect_transform_loop): Handle the case in which the final loop
13020 iteration might handle a partial vector. Call vect_set_loop_condition
13021 instead of slpeel_make_loop_iterate_ntimes.
13022 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
13023 (check_load_store_masking): New function.
13024 (prepare_load_store_mask): Likewise.
13025 (vectorizable_store): Handle fully-masked loops.
13026 (vectorizable_load): Likewise.
13027 (supportable_widening_operation): Use vect_halve_mask_nunits for
13029 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
13030 (vect_gen_while): New function.
13031 * config/aarch64/aarch64.md (umax<mode>3): New expander.
13032 (aarch64_uqdec<mode>): New insn.
13034 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13035 Alan Hayward <alan.hayward@arm.com>
13036 David Sherwood <david.sherwood@arm.com>
13038 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
13039 (reduc_xor_scal_optab): New optabs.
13040 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
13041 (reduc_xor_scal_@var{m}): Document.
13042 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
13043 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
13044 internal functions.
13045 * fold-const-call.c (fold_const_call): Handle them.
13046 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
13047 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
13048 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
13049 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
13050 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
13051 (UNSPEC_XORV): New unspecs.
13052 (optab): Add entries for them.
13053 (BITWISEV): New int iterator.
13054 (bit_reduc_op): New int attributes.
13056 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13057 Alan Hayward <alan.hayward@arm.com>
13058 David Sherwood <david.sherwood@arm.com>
13060 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
13061 * internal-fn.def (VEC_SHL_INSERT): New internal function.
13062 * optabs.def (vec_shl_insert_optab): New optab.
13063 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
13064 (duplicate_and_interleave): Likewise.
13065 * tree-vect-loop.c: Include internal-fn.h.
13066 (neutral_op_for_slp_reduction): New function, split out from
13067 get_initial_defs_for_reduction.
13068 (get_initial_def_for_reduction): Handle option 2 for variable-length
13069 vectors by loading the neutral value into a vector and then shifting
13070 the initial value into element 0.
13071 (get_initial_defs_for_reduction): Replace the code argument with
13072 the neutral value calculated by neutral_op_for_slp_reduction.
13073 Use gimple_build_vector for constant-length vectors.
13074 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
13075 but the first group_size elements have a neutral value.
13076 Use duplicate_and_interleave otherwise.
13077 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
13078 Update call to get_initial_defs_for_reduction. Handle SLP
13079 reductions for variable-length vectors by creating one vector
13080 result for each scalar result, with the elements associated
13081 with other scalar results stubbed out with the neutral value.
13082 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
13083 Require IFN_VEC_SHL_INSERT for double reductions on
13084 variable-length vectors, or SLP reductions that have
13085 a neutral value. Require can_duplicate_and_interleave_p
13086 support for variable-length unchained SLP reductions if there
13087 is no neutral value, such as for MIN/MAX reductions. Also require
13088 the number of vector elements to be a multiple of the number of
13089 SLP statements when doing variable-length unchained SLP reductions.
13090 Update call to vect_create_epilog_for_reduction.
13091 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
13092 and remove initial values.
13093 (duplicate_and_interleave): Make public.
13094 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
13095 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
13097 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13098 Alan Hayward <alan.hayward@arm.com>
13099 David Sherwood <david.sherwood@arm.com>
13101 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
13102 (can_duplicate_and_interleave_p): New function.
13103 (vect_get_and_check_slp_defs): Take the vector of statements
13104 rather than just the current one. Remove excess parentheses.
13105 Restriction rejectinon of vect_constant_def and vect_external_def
13106 for variable-length vectors to boolean types, or types for which
13107 can_duplicate_and_interleave_p is false.
13108 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
13109 (duplicate_and_interleave): New function.
13110 (vect_get_constant_vectors): Use gimple_build_vector for
13111 constant-length vectors and suitable variable-length constant
13112 vectors. Use duplicate_and_interleave for other variable-length
13113 vectors. Don't defer the update when inserting new statements.
13115 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13116 Alan Hayward <alan.hayward@arm.com>
13117 David Sherwood <david.sherwood@arm.com>
13119 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
13120 min_profitable_iters doesn't go negative.
13122 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13123 Alan Hayward <alan.hayward@arm.com>
13124 David Sherwood <david.sherwood@arm.com>
13126 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
13127 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
13128 * optabs.def (vec_mask_load_lanes_optab): New optab.
13129 (vec_mask_store_lanes_optab): Likewise.
13130 * internal-fn.def (MASK_LOAD_LANES): New internal function.
13131 (MASK_STORE_LANES): Likewise.
13132 * internal-fn.c (mask_load_lanes_direct): New macro.
13133 (mask_store_lanes_direct): Likewise.
13134 (expand_mask_load_optab_fn): Handle masked operations.
13135 (expand_mask_load_lanes_optab_fn): New macro.
13136 (expand_mask_store_optab_fn): Handle masked operations.
13137 (expand_mask_store_lanes_optab_fn): New macro.
13138 (direct_mask_load_lanes_optab_supported_p): Likewise.
13139 (direct_mask_store_lanes_optab_supported_p): Likewise.
13140 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
13142 (vect_load_lanes_supported): Likewise.
13143 * tree-vect-data-refs.c (strip_conversion): New function.
13144 (can_group_stmts_p): Likewise.
13145 (vect_analyze_data_ref_accesses): Use it instead of checking
13146 for a pair of assignments.
13147 (vect_store_lanes_supported): Take a masked_p parameter.
13148 (vect_load_lanes_supported): Likewise.
13149 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
13150 vect_store_lanes_supported and vect_load_lanes_supported.
13151 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13152 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
13153 parameter. Don't allow gaps for masked accesses.
13154 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
13155 and vect_load_lanes_supported.
13156 (get_load_store_type): Take a masked_p parameter and update
13157 call to get_group_load_store_type.
13158 (vectorizable_store): Update call to get_load_store_type.
13159 Handle IFN_MASK_STORE_LANES.
13160 (vectorizable_load): Update call to get_load_store_type.
13161 Handle IFN_MASK_LOAD_LANES.
13163 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13164 Alan Hayward <alan.hayward@arm.com>
13165 David Sherwood <david.sherwood@arm.com>
13167 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
13169 * config/aarch64/aarch64-protos.h
13170 (aarch64_sve_struct_memory_operand_p): Declare.
13171 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
13172 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
13173 (VPRED, vpred): Handle SVE structure modes.
13174 * config/aarch64/constraints.md (Utx): New constraint.
13175 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
13176 (aarch64_sve_struct_nonimmediate_operand): New predicates.
13177 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
13178 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
13179 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
13180 structure modes. Split into pieces after RA.
13181 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
13182 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
13184 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
13185 SVE structure modes.
13186 (aarch64_classify_address): Likewise.
13187 (sizetochar): Move earlier in file.
13188 (aarch64_print_operand): Handle SVE register lists.
13189 (aarch64_array_mode): New function.
13190 (aarch64_sve_struct_memory_operand_p): Likewise.
13191 (TARGET_ARRAY_MODE): Redefine.
13193 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13194 Alan Hayward <alan.hayward@arm.com>
13195 David Sherwood <david.sherwood@arm.com>
13197 * target.def (array_mode): New target hook.
13198 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
13199 * doc/tm.texi: Regenerate.
13200 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
13201 * hooks.c (hook_optmode_mode_uhwi_none): New function.
13202 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
13203 targetm.array_mode.
13204 * stor-layout.c (mode_for_array): Likewise. Support polynomial
13207 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13208 Alan Hayward <alan.hayward@arm.com>
13209 David Sherwood <david.sherwood@arm.com>
13211 * fold-const.c (fold_binary_loc): Check the argument types
13212 rather than the result type when testing for a vector operation.
13214 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13216 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
13217 * doc/tm.texi: Regenerate.
13219 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13220 Alan Hayward <alan.hayward@arm.com>
13221 David Sherwood <david.sherwood@arm.com>
13223 * doc/invoke.texi (-msve-vector-bits=): Document new option.
13224 (sve): Document new AArch64 extension.
13225 * doc/md.texi (w): Extend the description of the AArch64
13226 constraint to include SVE vectors.
13227 (Upl, Upa): Document new AArch64 predicate constraints.
13228 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
13230 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
13231 (msve-vector-bits=): New option.
13232 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
13233 SVE when these are disabled.
13234 (sve): New extension.
13235 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
13236 modes. Adjust their number of units based on aarch64_sve_vg.
13237 (MAX_BITSIZE_MODE_ANY_MODE): Define.
13238 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
13239 aarch64_addr_query_type.
13240 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
13241 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
13242 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
13243 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
13244 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
13245 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
13246 (aarch64_simd_imm_zero_p): Delete.
13247 (aarch64_check_zero_based_sve_index_immediate): Declare.
13248 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
13249 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
13250 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
13251 (aarch64_sve_float_mul_immediate_p): Likewise.
13252 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
13253 rather than an rtx.
13254 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
13255 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
13256 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
13257 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
13258 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
13259 (aarch64_regmode_natural_size): Likewise.
13260 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
13261 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
13263 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
13264 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
13265 for VG and the SVE predicate registers.
13266 (V_ALIASES): Add a "z"-prefixed alias.
13267 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
13268 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
13269 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
13270 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
13271 (REG_CLASS_NAMES): Add entries for them.
13272 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
13273 and the predicate registers.
13274 (aarch64_sve_vg): Declare.
13275 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
13276 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
13277 (REGMODE_NATURAL_SIZE): Define.
13278 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
13280 * config/aarch64/aarch64.c: Include cfgrtl.h.
13281 (simd_immediate_info): Add a constructor for series vectors,
13282 and an associated step field.
13283 (aarch64_sve_vg): New variable.
13284 (aarch64_dbx_register_number): Handle VG and the predicate registers.
13285 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
13286 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
13287 (VEC_ANY_DATA, VEC_STRUCT): New constants.
13288 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
13289 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
13290 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
13291 (aarch64_get_mask_mode): New functions.
13292 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
13293 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
13294 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
13295 predicate modes and predicate registers. Explicitly restrict
13296 GPRs to modes of 16 bytes or smaller. Only allow FP registers
13297 to store a vector mode if it is recognized by
13298 aarch64_classify_vector_mode.
13299 (aarch64_regmode_natural_size): New function.
13300 (aarch64_hard_regno_caller_save_mode): Return the original mode
13302 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
13303 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
13304 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
13305 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
13307 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
13308 does not overlap dest if the function is frame-related. Handle
13310 (aarch64_split_add_offset): New function.
13311 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
13312 them aarch64_add_offset.
13313 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
13314 and update call to aarch64_sub_sp.
13315 (aarch64_add_cfa_expression): New function.
13316 (aarch64_expand_prologue): Pass extra temporary registers to the
13317 functions above. Handle the case in which we need to emit new
13318 DW_CFA_expressions for registers that were originally saved
13319 relative to the stack pointer, but now have to be expressed
13320 relative to the frame pointer.
13321 (aarch64_output_mi_thunk): Pass extra temporary registers to the
13323 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
13324 IP0 and IP1 values for SVE frames.
13325 (aarch64_expand_vec_series): New function.
13326 (aarch64_expand_sve_widened_duplicate): Likewise.
13327 (aarch64_expand_sve_const_vector): Likewise.
13328 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
13329 Handle SVE constants. Use emit_move_insn to move a force_const_mem
13330 into the register, rather than emitting a SET directly.
13331 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
13332 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
13333 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
13334 (offset_9bit_signed_scaled_p): New functions.
13335 (aarch64_replicate_bitmask_imm): New function.
13336 (aarch64_bitmask_imm): Use it.
13337 (aarch64_cannot_force_const_mem): Reject expressions involving
13338 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
13339 (aarch64_classify_index): Handle SVE indices, by requiring
13340 a plain register index with a scale that matches the element size.
13341 (aarch64_classify_address): Handle SVE addresses. Assert that
13342 the mode of the address is VOIDmode or an integer mode.
13343 Update call to aarch64_classify_symbol.
13344 (aarch64_classify_symbolic_expression): Update call to
13345 aarch64_classify_symbol.
13346 (aarch64_const_vec_all_in_range_p): New function.
13347 (aarch64_print_vector_float_operand): Likewise.
13348 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
13349 "vN" for FP registers with SVE modes. Handle (const ...) vectors
13350 and the FP immediates 1.0 and 0.5.
13351 (aarch64_print_address_internal): Handle SVE addresses.
13352 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
13353 (aarch64_regno_regclass): Handle predicate registers.
13354 (aarch64_secondary_reload): Handle big-endian reloads of SVE
13356 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
13357 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
13358 (aarch64_convert_sve_vector_bits): New function.
13359 (aarch64_override_options): Use it to handle -msve-vector-bits=.
13360 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
13361 rather than an rtx.
13362 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
13363 Handle SVE vector and predicate modes. Accept VL-based constants
13364 that need only one temporary register, and VL offsets that require
13365 no temporary registers.
13366 (aarch64_conditional_register_usage): Mark the predicate registers
13367 as fixed if SVE isn't available.
13368 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
13369 Return true for SVE vector and predicate modes.
13370 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
13371 rather than an unsigned int. Handle SVE modes.
13372 (aarch64_preferred_simd_mode): Update call accordingly. Handle
13374 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
13376 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
13377 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
13378 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
13379 (aarch64_sve_float_mul_immediate_p): New functions.
13380 (aarch64_sve_valid_immediate): New function.
13381 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
13382 Explicitly reject structure modes. Check for INDEX constants.
13383 Handle PTRUE and PFALSE constants.
13384 (aarch64_check_zero_based_sve_index_immediate): New function.
13385 (aarch64_simd_imm_zero_p): Delete.
13386 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
13387 vector modes. Accept constants in the range of CNT[BHWD].
13388 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
13389 ask for an Advanced SIMD mode.
13390 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
13391 (aarch64_simd_vector_alignment): Handle SVE predicates.
13392 (aarch64_vectorize_preferred_vector_alignment): New function.
13393 (aarch64_simd_vector_alignment_reachable): Use it instead of
13395 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
13396 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
13398 (MAX_VECT_LEN): Delete.
13399 (expand_vec_perm_d): Add a vec_flags field.
13400 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
13401 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13402 (aarch64_evpc_ext): Don't apply a big-endian lane correction
13404 (aarch64_evpc_rev): Rename to...
13405 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
13406 (aarch64_evpc_rev_global): New function.
13407 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
13408 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
13410 (aarch64_evpc_sve_tbl): New function.
13411 (aarch64_expand_vec_perm_const_1): Update after rename of
13412 aarch64_evpc_rev. Handle SVE permutes too, trying
13413 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
13414 than aarch64_evpc_tbl.
13415 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
13416 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
13417 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
13418 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
13419 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
13420 (aarch64_expand_sve_vcond): New functions.
13421 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
13422 of aarch64_vector_mode_p.
13423 (aarch64_dwarf_poly_indeterminate_value): New function.
13424 (aarch64_compute_pressure_classes): Likewise.
13425 (aarch64_can_change_mode_class): Likewise.
13426 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
13427 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
13428 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
13429 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
13430 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
13431 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
13432 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
13433 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
13435 (Dn, Dl, Dr): Accept const as well as const_vector.
13436 (Dz): Likewise. Compare against CONST0_RTX.
13437 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
13438 of "vector" where appropriate.
13439 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
13440 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
13441 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
13442 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
13443 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
13444 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
13445 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
13446 (v_int_equiv): Extend to SVE modes.
13447 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
13449 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
13450 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
13451 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
13452 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
13453 (SVE_COND_FP_CMP): New int iterators.
13454 (perm_hilo): Handle the new unpack unspecs.
13455 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
13457 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
13458 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
13459 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
13460 (aarch64_equality_operator, aarch64_constant_vector_operand)
13461 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
13462 (aarch64_sve_nonimmediate_operand): Likewise.
13463 (aarch64_sve_general_operand): Likewise.
13464 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
13465 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
13466 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
13467 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
13468 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
13469 (aarch64_sve_float_arith_immediate): Likewise.
13470 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
13471 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
13472 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
13473 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
13474 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
13475 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
13476 (aarch64_sve_float_arith_operand): Likewise.
13477 (aarch64_sve_float_arith_with_sub_operand): Likewise.
13478 (aarch64_sve_float_mul_operand): Likewise.
13479 (aarch64_sve_vec_perm_operand): Likewise.
13480 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
13481 (aarch64_mov_operand): Accept const_poly_int and const_vector.
13482 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
13483 as well as const_vector.
13484 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
13485 in file. Use CONST0_RTX and CONSTM1_RTX.
13486 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
13487 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
13488 Use aarch64_simd_imm_zero.
13489 * config/aarch64/aarch64-sve.md: New file.
13490 * config/aarch64/aarch64.md: Include it.
13491 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
13492 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
13493 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
13494 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
13495 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
13496 (sve): New attribute.
13497 (enabled): Disable instructions with the sve attribute unless
13499 (movqi, movhi): Pass CONST_POLY_INT operaneds through
13500 aarch64_expand_mov_immediate.
13501 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
13502 CNT[BHSD] immediates.
13503 (movti): Split CONST_POLY_INT moves into two halves.
13504 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
13505 Split additions that need a temporary here if the destination
13506 is the stack pointer.
13507 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
13508 (*add<mode>3_poly_1): New instruction.
13509 (set_clobber_cc): New expander.
13511 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13513 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
13514 parameter and use it instead of GET_MODE_SIZE (innermode). Use
13515 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
13516 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
13517 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
13518 Change innermode from fixed_mode_size to machine_mode.
13519 (simplify_subreg): Update call accordingly. Handle a constant-sized
13520 subreg of a variable-length CONST_VECTOR.
13522 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
13523 Alan Hayward <alan.hayward@arm.com>
13524 David Sherwood <david.sherwood@arm.com>
13526 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
13527 (add_offset_to_base): New function, split out from...
13528 (create_mem_ref): ...here. When handling a scale other than 1,
13529 check first whether the address is valid without the offset.
13530 Add it into the base if so, leaving the index and scale as-is.
13532 2018-01-12 Jakub Jelinek <jakub@redhat.com>
13535 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
13536 fold_for_warn before checking if arg2 is INTEGER_CST.
13538 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
13540 * config/rs6000/predicates.md (load_multiple_operation): Delete.
13541 (store_multiple_operation): Delete.
13542 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
13543 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
13544 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
13545 guarded by TARGET_STRING.
13546 (rs6000_output_load_multiple): Delete.
13547 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13548 OPTION_MASK_STRING / TARGET_STRING handling.
13549 (print_operand) <'N', 'O'>: Add comment that these are unused now.
13550 (const rs6000_opt_masks) <"string">: Change mask to 0.
13551 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
13552 (MASK_STRING): Delete.
13553 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
13555 (load_multiple): Delete.
13562 (store_multiple): Delete.
13569 (movmemsi_8reg): Delete.
13570 (corresponding unnamed define_insn): Delete.
13571 (movmemsi_6reg): Delete.
13572 (corresponding unnamed define_insn): Delete.
13573 (movmemsi_4reg): Delete.
13574 (corresponding unnamed define_insn): Delete.
13575 (movmemsi_2reg): Delete.
13576 (corresponding unnamed define_insn): Delete.
13577 (movmemsi_1reg): Delete.
13578 (corresponding unnamed define_insn): Delete.
13579 * config/rs6000/rs6000.opt (mno-string): New.
13580 (mstring): Replace by deprecation warning stub.
13581 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
13583 2018-01-12 Jakub Jelinek <jakub@redhat.com>
13585 * regrename.c (regrename_do_replace): If replacing the same
13586 reg multiple times, try to reuse last created gen_raw_REG.
13589 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
13590 main to workaround a bug in GDB.
13592 2018-01-12 Tom de Vries <tom@codesourcery.com>
13595 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
13597 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
13599 PR rtl-optimization/80481
13600 * ira-color.c (get_cap_member): New function.
13601 (allocnos_conflict_by_live_ranges_p): Use it.
13602 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
13603 (setup_slot_coalesced_allocno_live_ranges): Ditto.
13605 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
13608 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
13609 (*saddl_se_1): Ditto.
13610 (*ssubsi_1): Ditto.
13611 (*ssubl_se_1): Ditto.
13613 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
13615 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
13616 rather than wi::to_widest for DR_INITs.
13617 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
13618 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
13619 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
13621 (vect_analyze_group_access_1): Note that here.
13623 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
13625 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
13626 polynomial type sizes.
13628 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
13630 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
13631 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
13632 (gimple_add_tmp_var): Likewise.
13634 2018-01-12 Martin Liska <mliska@suse.cz>
13636 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
13637 (gimple_alloc_sizes): Likewise.
13638 (dump_gimple_statistics): Use PRIu64 in printf format.
13639 * gimple.h: Change uint64_t to int.
13641 2018-01-12 Martin Liska <mliska@suse.cz>
13643 * tree-core.h: Use uint64_t instead of int.
13644 * tree.c (tree_node_counts): Likewise.
13645 (tree_node_sizes): Likewise.
13646 (dump_tree_statistics): Use PRIu64 in printf format.
13648 2018-01-12 Martin Liska <mliska@suse.cz>
13650 * Makefile.in: As qsort_chk is implemented in vec.c, add
13651 vec.o to linkage of gencfn-macros.
13652 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
13653 passing the info to record_node_allocation_statistics.
13654 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
13656 * ggc-common.c (struct ggc_usage): Add operator== and use
13657 it in operator< and compare function.
13658 * mem-stats.h (struct mem_usage): Likewise.
13659 * vec.c (struct vec_usage): Remove operator< and compare
13660 function. Can be simply inherited.
13662 2018-01-12 Martin Jambor <mjambor@suse.cz>
13665 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
13666 * tree-ssa-math-opts.c: Include domwalk.h.
13667 (convert_mult_to_fma_1): New function.
13668 (fma_transformation_info): New type.
13669 (fma_deferring_state): Likewise.
13670 (cancel_fma_deferring): New function.
13671 (result_of_phi): Likewise.
13672 (last_fma_candidate_feeds_initial_phi): Likewise.
13673 (convert_mult_to_fma): Added deferring logic, split actual
13674 transformation to convert_mult_to_fma_1.
13675 (math_opts_dom_walker): New type.
13676 (math_opts_dom_walker::after_dom_children): New method, body moved
13677 here from pass_optimize_widening_mul::execute, added deferring logic
13679 (pass_optimize_widening_mul::execute): Moved most of code to
13680 math_opts_dom_walker::after_dom_children.
13681 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
13682 * config/i386/i386.c (ix86_option_override_internal): Added
13683 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
13685 2018-01-12 Richard Biener <rguenther@suse.de>
13688 * dwarf2out.c (gen_variable_die): Do not reset old_die for
13689 inline instance vars.
13691 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
13694 * config/rx/rx.c (rx_is_restricted_memory_address):
13695 Handle SUBREG case.
13697 2018-01-12 Richard Biener <rguenther@suse.de>
13699 PR tree-optimization/80846
13700 * target.def (split_reduction): New target hook.
13701 * targhooks.c (default_split_reduction): New function.
13702 * targhooks.h (default_split_reduction): Declare.
13703 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
13704 target requests first reduce vectors by combining low and high
13706 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
13707 (get_vectype_for_scalar_type_and_size): Export.
13708 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
13709 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
13710 * doc/tm.texi: Regenerate.
13711 * config/i386/i386.c (ix86_split_reduction): Implement
13712 TARGET_VECTORIZE_SPLIT_REDUCTION.
13714 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
13717 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
13718 in PIC mode except for TARGET_VXWORKS_RTP.
13719 * config/sparc/sparc.c: Include cfgrtl.h.
13720 (TARGET_INIT_PIC_REG): Define.
13721 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
13722 (sparc_pic_register_p): New predicate.
13723 (sparc_legitimate_address_p): Use it.
13724 (sparc_legitimize_pic_address): Likewise.
13725 (sparc_delegitimize_address): Likewise.
13726 (sparc_mode_dependent_address_p): Likewise.
13727 (gen_load_pcrel_sym): Remove 4th parameter.
13728 (load_got_register): Adjust call to above. Remove obsolete stuff.
13729 (sparc_expand_prologue): Do not call load_got_register here.
13730 (sparc_flat_expand_prologue): Likewise.
13731 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
13732 (sparc_use_pseudo_pic_reg): New function.
13733 (sparc_init_pic_reg): Likewise.
13734 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
13735 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
13737 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
13739 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13740 Add item for branch_cost.
13742 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
13744 PR rtl-optimization/83565
13745 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
13746 not extend the result to a larger mode for rotate operations.
13747 (num_sign_bit_copies1): Likewise.
13749 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13752 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
13754 Use values-Xc.o for -pedantic.
13755 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
13757 2018-01-12 Martin Liska <mliska@suse.cz>
13760 * ipa-devirt.c (final_warning_record::grow_type_warnings):
13762 (possible_polymorphic_call_targets): Use it.
13763 (ipa_devirt): Likewise.
13765 2018-01-12 Martin Liska <mliska@suse.cz>
13767 * profile-count.h (enum profile_quality): Use 0 as invalid
13768 enum value of profile_quality.
13770 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
13772 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
13773 -mext-string options.
13775 2018-01-12 Richard Biener <rguenther@suse.de>
13777 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
13778 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
13779 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13781 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13783 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13785 * configure.ac (--with-long-double-format): Add support for the
13786 configuration option to change the default long double format on
13788 * config.gcc (powerpc*-linux*-*): Likewise.
13789 * configure: Regenerate.
13790 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
13791 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
13792 used without modification.
13794 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13796 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
13797 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
13798 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
13799 MISC_BUILTIN_SPEC_BARRIER.
13800 (rs6000_init_builtins): Likewise.
13801 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
13803 (speculation_barrier): New define_insn.
13804 * doc/extend.texi: Document __builtin_speculation_barrier.
13806 2018-01-11 Jakub Jelinek <jakub@redhat.com>
13809 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
13810 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
13811 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
13813 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
13814 integral modes instead of "ss" and "sd".
13815 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
13816 vectors with 32-bit and 64-bit elements.
13817 (vecdupssescalarmodesuffix): New mode attribute.
13818 (vec_dup<mode>): Use it.
13820 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
13823 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
13824 frame if argument is passed on stack.
13826 2018-01-11 Jakub Jelinek <jakub@redhat.com>
13829 * ree.c (combine_reaching_defs): Optimize also
13830 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
13831 reg2=any_extend(exp); reg1=reg2;, formatting fix.
13833 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
13835 PR middle-end/83189
13836 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
13838 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
13840 PR middle-end/83718
13841 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
13842 after they are computed.
13844 2018-01-11 Bin Cheng <bin.cheng@arm.com>
13846 PR tree-optimization/83695
13847 * gimple-loop-linterchange.cc
13848 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
13849 reset cached scev information after interchange.
13850 (pass_linterchange::execute): Remove call to scev_reset_htab.
13852 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13854 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
13855 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
13856 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
13857 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
13858 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
13859 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
13860 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
13861 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
13862 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
13863 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
13864 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
13865 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
13866 (V_lane_reg): Likewise.
13867 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
13869 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
13870 (vfmal_lane_low<mode>_intrinsic,
13871 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
13872 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
13873 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
13874 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
13875 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
13876 vfmsl_lane_high<mode>_intrinsic): New define_insns.
13878 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13880 * config/arm/arm-cpus.in (fp16fml): New feature.
13881 (ALL_SIMD): Add fp16fml.
13882 (armv8.2-a): Add fp16fml as an option.
13883 (armv8.3-a): Likewise.
13884 (armv8.4-a): Add fp16fml as part of fp16.
13885 * config/arm/arm.h (TARGET_FP16FML): Define.
13886 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
13888 * config/arm/arm-modes.def (V2HF): Define.
13889 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
13890 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
13891 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
13892 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
13893 vfmsl_low, vfmsl_high): New set of builtins.
13894 * config/arm/iterators.md (PLUSMINUS): New code iterator.
13895 (vfml_op): New code attribute.
13896 (VFMLHALVES): New int iterator.
13897 (VFML, VFMLSEL): New mode attributes.
13898 (V_reg): Define mapping for V2HF.
13899 (V_hi, V_lo): New mode attributes.
13900 (VF_constraint): Likewise.
13901 (vfml_half, vfml_half_selector): New int attributes.
13902 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
13904 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
13905 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
13907 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
13908 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
13909 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
13910 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
13912 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
13913 Document new effective target and option set.
13915 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13917 * config/arm/arm-cpus.in (armv8_4): New feature.
13918 (ARMv8_4a): New fgroup.
13919 (armv8.4-a): New arch.
13920 * config/arm/arm-tables.opt: Regenerate.
13921 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
13922 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
13923 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
13924 Add matching rules for -march=armv8.4-a and extensions.
13925 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
13927 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
13930 * config/rx/rx.md (BW): New mode attribute.
13931 (sync_lock_test_and_setsi): Add mode suffix to insn output.
13933 2018-01-11 Richard Biener <rguenther@suse.de>
13935 PR tree-optimization/83435
13936 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
13937 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
13938 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
13940 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
13941 Alan Hayward <alan.hayward@arm.com>
13942 David Sherwood <david.sherwood@arm.com>
13944 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
13946 (aarch64_classify_address): Initialize it. Track polynomial offsets.
13947 (aarch64_print_address_internal): Use it to check for a zero offset.
13949 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
13950 Alan Hayward <alan.hayward@arm.com>
13951 David Sherwood <david.sherwood@arm.com>
13953 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
13954 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
13955 Return a poly_int64 rather than a HOST_WIDE_INT.
13956 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
13957 rather than a HOST_WIDE_INT.
13958 * config/aarch64/aarch64.h (aarch64_frame): Protect with
13959 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
13960 hard_fp_offset, frame_size, initial_adjust, callee_offset and
13961 final_offset from HOST_WIDE_INT to poly_int64.
13962 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
13963 to_constant when getting the number of units in an Advanced SIMD
13965 (aarch64_builtin_vectorized_function): Check for a constant number
13967 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
13969 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
13970 attribute instead of GET_MODE_NUNITS.
13971 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
13972 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
13973 GET_MODE_SIZE for fixed-size registers.
13974 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
13975 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
13976 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
13977 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
13978 (aarch64_print_operand, aarch64_print_address_internal)
13979 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
13980 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
13981 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
13982 Handle polynomial GET_MODE_SIZE.
13983 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
13984 wider than SImode without modification.
13985 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
13986 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
13987 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
13988 passing and returning SVE modes.
13989 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
13990 rather than GEN_INT.
13991 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
13992 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
13993 (aarch64_allocate_and_probe_stack_space): Likewise.
13994 (aarch64_layout_frame): Cope with polynomial offsets.
13995 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
13996 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
13997 polynomial offsets.
13998 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
13999 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
14000 poly_int64 rather than a HOST_WIDE_INT.
14001 (aarch64_get_separate_components, aarch64_process_components)
14002 (aarch64_expand_prologue, aarch64_expand_epilogue)
14003 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
14004 (aarch64_anchor_offset): New function, split out from...
14005 (aarch64_legitimize_address): ...here.
14006 (aarch64_builtin_vectorization_cost): Handle polynomial
14007 TYPE_VECTOR_SUBPARTS.
14008 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
14010 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
14011 number of elements from the PARALLEL rather than the mode.
14012 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
14013 rather than GET_MODE_BITSIZE.
14014 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
14015 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
14016 (aarch64_expand_vec_perm_const_1): Handle polynomial
14017 d->perm.length () and d->perm elements.
14018 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
14019 Apply to_constant to d->perm elements.
14020 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
14021 polynomial CONST_VECTOR_NUNITS.
14022 (aarch64_move_pointer): Take amount as a poly_int64 rather
14024 (aarch64_progress_pointer): Avoid temporary variable.
14025 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
14026 the mode attribute instead of GET_MODE.
14028 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
14029 Alan Hayward <alan.hayward@arm.com>
14030 David Sherwood <david.sherwood@arm.com>
14032 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
14033 x exists before using it.
14034 (aarch64_add_constant_internal): Rename to...
14035 (aarch64_add_offset_1): ...this. Replace regnum with separate
14036 src and dest rtxes. Handle the case in which they're different,
14037 including when the offset is zero. Replace scratchreg with an rtx.
14038 Use 2 additions if there is no spare register into which we can
14039 move a 16-bit constant.
14040 (aarch64_add_constant): Delete.
14041 (aarch64_add_offset): Replace reg with separate src and dest
14042 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
14043 Use aarch64_add_offset_1.
14044 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
14045 an rtx rather than an int. Take the delta as a poly_int64
14046 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
14047 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
14048 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
14049 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
14050 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
14051 and aarch64_add_sp.
14052 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
14053 aarch64_add_constant.
14055 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
14057 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14058 Use scalar_float_mode.
14060 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
14062 * config/aarch64/aarch64-simd.md
14063 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
14064 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
14065 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
14066 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
14067 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
14068 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
14069 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
14070 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
14071 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
14072 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
14074 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14077 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
14078 targ_options->x_arm_arch_string is non NULL.
14080 2018-01-11 Tamar Christina <tamar.christina@arm.com>
14082 * config/aarch64/aarch64.h
14083 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
14085 2018-01-11 Sudakshina Das <sudi.das@arm.com>
14088 * expmed.c (emit_store_flag_force): Swap if const op0
14089 and change VOIDmode to mode of op0.
14091 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
14093 PR rtl-optimization/83761
14094 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
14095 than bytes to mode_for_size.
14097 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
14099 PR middle-end/83189
14100 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
14101 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
14104 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
14106 PR middle-end/83575
14107 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
14108 when in layout mode.
14109 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
14110 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
14113 2018-01-10 Michael Collison <michael.collison@arm.com>
14115 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
14116 * config/aarch64/aarch64-option-extension.def: Add
14117 AARCH64_OPT_EXTENSION of 'fp16fml'.
14118 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
14119 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
14120 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
14121 * config/aarch64/constraints.md (Ui7): New constraint.
14122 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
14123 (VFMLA_SEL_W): Ditto.
14126 (VFMLA16_LOW): New int iterator.
14127 (VFMLA16_HIGH): Ditto.
14128 (UNSPEC_FMLAL): New unspec.
14129 (UNSPEC_FMLSL): Ditto.
14130 (UNSPEC_FMLAL2): Ditto.
14131 (UNSPEC_FMLSL2): Ditto.
14132 (f16mac): New code attribute.
14133 * config/aarch64/aarch64-simd-builtins.def
14134 (aarch64_fmlal_lowv2sf): Ditto.
14135 (aarch64_fmlsl_lowv2sf): Ditto.
14136 (aarch64_fmlalq_lowv4sf): Ditto.
14137 (aarch64_fmlslq_lowv4sf): Ditto.
14138 (aarch64_fmlal_highv2sf): Ditto.
14139 (aarch64_fmlsl_highv2sf): Ditto.
14140 (aarch64_fmlalq_highv4sf): Ditto.
14141 (aarch64_fmlslq_highv4sf): Ditto.
14142 (aarch64_fmlal_lane_lowv2sf): Ditto.
14143 (aarch64_fmlsl_lane_lowv2sf): Ditto.
14144 (aarch64_fmlal_laneq_lowv2sf): Ditto.
14145 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
14146 (aarch64_fmlalq_lane_lowv4sf): Ditto.
14147 (aarch64_fmlsl_lane_lowv4sf): Ditto.
14148 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
14149 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
14150 (aarch64_fmlal_lane_highv2sf): Ditto.
14151 (aarch64_fmlsl_lane_highv2sf): Ditto.
14152 (aarch64_fmlal_laneq_highv2sf): Ditto.
14153 (aarch64_fmlsl_laneq_highv2sf): Ditto.
14154 (aarch64_fmlalq_lane_highv4sf): Ditto.
14155 (aarch64_fmlsl_lane_highv4sf): Ditto.
14156 (aarch64_fmlalq_laneq_highv4sf): Ditto.
14157 (aarch64_fmlsl_laneq_highv4sf): Ditto.
14158 * config/aarch64/aarch64-simd.md:
14159 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
14160 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
14161 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
14162 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
14163 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
14164 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
14165 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
14166 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
14167 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
14168 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
14169 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
14170 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
14171 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
14172 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
14173 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
14174 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
14175 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
14176 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
14177 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
14178 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
14179 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
14180 (vfmlsl_low_u32): Ditto.
14181 (vfmlalq_low_u32): Ditto.
14182 (vfmlslq_low_u32): Ditto.
14183 (vfmlal_high_u32): Ditto.
14184 (vfmlsl_high_u32): Ditto.
14185 (vfmlalq_high_u32): Ditto.
14186 (vfmlslq_high_u32): Ditto.
14187 (vfmlal_lane_low_u32): Ditto.
14188 (vfmlsl_lane_low_u32): Ditto.
14189 (vfmlal_laneq_low_u32): Ditto.
14190 (vfmlsl_laneq_low_u32): Ditto.
14191 (vfmlalq_lane_low_u32): Ditto.
14192 (vfmlslq_lane_low_u32): Ditto.
14193 (vfmlalq_laneq_low_u32): Ditto.
14194 (vfmlslq_laneq_low_u32): Ditto.
14195 (vfmlal_lane_high_u32): Ditto.
14196 (vfmlsl_lane_high_u32): Ditto.
14197 (vfmlal_laneq_high_u32): Ditto.
14198 (vfmlsl_laneq_high_u32): Ditto.
14199 (vfmlalq_lane_high_u32): Ditto.
14200 (vfmlslq_lane_high_u32): Ditto.
14201 (vfmlalq_laneq_high_u32): Ditto.
14202 (vfmlslq_laneq_high_u32): Ditto.
14203 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
14204 (AARCH64_FL_FOR_ARCH8_4): New.
14205 (AARCH64_ISA_F16FML): New ISA flag.
14206 (TARGET_F16FML): New feature flag for fp16fml.
14207 (doc/invoke.texi): Document new fp16fml option.
14209 2018-01-10 Michael Collison <michael.collison@arm.com>
14211 * config/aarch64/aarch64-builtins.c:
14212 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
14213 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
14214 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
14215 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
14216 (AARCH64_ISA_SHA3): New ISA flag.
14217 (TARGET_SHA3): New feature flag for sha3.
14218 * config/aarch64/iterators.md (sha512_op): New int attribute.
14219 (CRYPTO_SHA512): New int iterator.
14220 (UNSPEC_SHA512H): New unspec.
14221 (UNSPEC_SHA512H2): Ditto.
14222 (UNSPEC_SHA512SU0): Ditto.
14223 (UNSPEC_SHA512SU1): Ditto.
14224 * config/aarch64/aarch64-simd-builtins.def
14225 (aarch64_crypto_sha512hqv2di): New builtin.
14226 (aarch64_crypto_sha512h2qv2di): Ditto.
14227 (aarch64_crypto_sha512su0qv2di): Ditto.
14228 (aarch64_crypto_sha512su1qv2di): Ditto.
14229 (aarch64_eor3qv8hi): Ditto.
14230 (aarch64_rax1qv2di): Ditto.
14231 (aarch64_xarqv2di): Ditto.
14232 (aarch64_bcaxqv8hi): Ditto.
14233 * config/aarch64/aarch64-simd.md:
14234 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
14235 (aarch64_crypto_sha512su0qv2di): Ditto.
14236 (aarch64_crypto_sha512su1qv2di): Ditto.
14237 (aarch64_eor3qv8hi): Ditto.
14238 (aarch64_rax1qv2di): Ditto.
14239 (aarch64_xarqv2di): Ditto.
14240 (aarch64_bcaxqv8hi): Ditto.
14241 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
14242 (vsha512h2q_u64): Ditto.
14243 (vsha512su0q_u64): Ditto.
14244 (vsha512su1q_u64): Ditto.
14245 (veor3q_u16): Ditto.
14246 (vrax1q_u64): Ditto.
14247 (vxarq_u64): Ditto.
14248 (vbcaxq_u16): Ditto.
14249 * config/arm/types.md (crypto_sha512): New type attribute.
14250 (crypto_sha3): Ditto.
14251 (doc/invoke.texi): Document new sha3 option.
14253 2018-01-10 Michael Collison <michael.collison@arm.com>
14255 * config/aarch64/aarch64-builtins.c:
14256 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
14257 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
14258 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
14259 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
14260 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
14261 (AARCH64_ISA_SM4): New ISA flag.
14262 (TARGET_SM4): New feature flag for sm4.
14263 * config/aarch64/aarch64-simd-builtins.def
14264 (aarch64_sm3ss1qv4si): Ditto.
14265 (aarch64_sm3tt1aq4si): Ditto.
14266 (aarch64_sm3tt1bq4si): Ditto.
14267 (aarch64_sm3tt2aq4si): Ditto.
14268 (aarch64_sm3tt2bq4si): Ditto.
14269 (aarch64_sm3partw1qv4si): Ditto.
14270 (aarch64_sm3partw2qv4si): Ditto.
14271 (aarch64_sm4eqv4si): Ditto.
14272 (aarch64_sm4ekeyqv4si): Ditto.
14273 * config/aarch64/aarch64-simd.md:
14274 (aarch64_sm3ss1qv4si): Ditto.
14275 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
14276 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
14277 (aarch64_sm4eqv4si): Ditto.
14278 (aarch64_sm4ekeyqv4si): Ditto.
14279 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
14280 (sm3part_op): Ditto.
14281 (CRYPTO_SM3TT): Ditto.
14282 (CRYPTO_SM3PART): Ditto.
14283 (UNSPEC_SM3SS1): New unspec.
14284 (UNSPEC_SM3TT1A): Ditto.
14285 (UNSPEC_SM3TT1B): Ditto.
14286 (UNSPEC_SM3TT2A): Ditto.
14287 (UNSPEC_SM3TT2B): Ditto.
14288 (UNSPEC_SM3PARTW1): Ditto.
14289 (UNSPEC_SM3PARTW2): Ditto.
14290 (UNSPEC_SM4E): Ditto.
14291 (UNSPEC_SM4EKEY): Ditto.
14292 * config/aarch64/constraints.md (Ui2): New constraint.
14293 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
14294 * config/arm/types.md (crypto_sm3): New type attribute.
14295 (crypto_sm4): Ditto.
14296 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
14297 (vsm3tt1aq_u32): Ditto.
14298 (vsm3tt1bq_u32): Ditto.
14299 (vsm3tt2aq_u32): Ditto.
14300 (vsm3tt2bq_u32): Ditto.
14301 (vsm3partw1q_u32): Ditto.
14302 (vsm3partw2q_u32): Ditto.
14303 (vsm4eq_u32): Ditto.
14304 (vsm4ekeyq_u32): Ditto.
14305 (doc/invoke.texi): Document new sm4 option.
14307 2018-01-10 Michael Collison <michael.collison@arm.com>
14309 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
14310 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
14311 (AARCH64_FL_FOR_ARCH8_4): New.
14312 (AARCH64_FL_V8_4): New flag.
14313 (doc/invoke.texi): Document new armv8.4-a option.
14315 2018-01-10 Michael Collison <michael.collison@arm.com>
14317 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
14318 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
14319 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
14320 * config/aarch64/aarch64-option-extension.def: Add
14321 AARCH64_OPT_EXTENSION of 'sha2'.
14322 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
14323 (crypto): Disable sha2 and aes if crypto disabled.
14324 (crypto): Enable aes and sha2 if enabled.
14325 (simd): Disable sha2 and aes if simd disabled.
14326 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
14328 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
14329 (TARGET_SHA2): New feature flag for sha2.
14330 (TARGET_AES): New feature flag for aes.
14331 * config/aarch64/aarch64-simd.md:
14332 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
14333 conditional on TARGET_AES.
14334 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
14335 (aarch64_crypto_sha1hsi): Make pattern conditional
14337 (aarch64_crypto_sha1hv4si): Ditto.
14338 (aarch64_be_crypto_sha1hv4si): Ditto.
14339 (aarch64_crypto_sha1su1v4si): Ditto.
14340 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
14341 (aarch64_crypto_sha1su0v4si): Ditto.
14342 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
14343 (aarch64_crypto_sha256su0v4si): Ditto.
14344 (aarch64_crypto_sha256su1v4si): Ditto.
14345 (doc/invoke.texi): Document new aes and sha2 options.
14347 2018-01-10 Martin Sebor <msebor@redhat.com>
14349 PR tree-optimization/83781
14350 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
14353 2018-01-11 Martin Sebor <msebor@gmail.com>
14354 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14356 PR tree-optimization/83501
14357 PR tree-optimization/81703
14359 * tree-ssa-strlen.c (get_string_cst): Rename...
14360 (get_string_len): ...to this. Handle global constants.
14361 (handle_char_store): Adjust.
14363 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
14364 Jim Wilson <jimw@sifive.com>
14366 * config/riscv/riscv-protos.h (riscv_output_return): New.
14367 * config/riscv/riscv.c (struct machine_function): New naked_p field.
14368 (riscv_attribute_table, riscv_output_return),
14369 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
14370 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
14371 (riscv_compute_frame_info): Only compute frame->mask if not a naked
14373 (riscv_expand_prologue): Add early return for naked function.
14374 (riscv_expand_epilogue): Likewise.
14375 (riscv_function_ok_for_sibcall): Return false for naked function.
14376 (riscv_set_current_function): New.
14377 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
14378 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
14379 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
14380 * doc/extend.texi (RISC-V Function Attributes): New.
14382 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
14384 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
14385 check for 128-bit long double before checking TCmode.
14386 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
14387 128-bit long doubles before checking TFmode or TCmode.
14388 (FLOAT128_IBM_P): Likewise.
14390 2018-01-10 Martin Sebor <msebor@redhat.com>
14392 PR tree-optimization/83671
14393 * builtins.c (c_strlen): Unconditionally return zero for the empty
14395 Use -Warray-bounds for warnings.
14396 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
14397 for non-constant array indices with COMPONENT_REF, arrays of
14398 arrays, and pointers to arrays.
14399 (gimple_fold_builtin_strlen): Determine and set length range for
14400 non-constant character arrays.
14402 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
14404 PR middle-end/81897
14405 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
14408 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
14410 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
14412 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
14415 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
14416 VECTOR_MEM_ALTIVEC_OR_VSX_P.
14417 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
14418 indexed_or_indirect_operand predicate.
14419 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
14420 (*vsx_le_perm_load_v8hi): Likewise.
14421 (*vsx_le_perm_load_v16qi): Likewise.
14422 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
14423 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
14424 (*vsx_le_perm_store_v8hi): Likewise.
14425 (*vsx_le_perm_store_v16qi): Likewise.
14426 (eight unnamed splitters): Likewise.
14428 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
14430 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
14431 * config/rs6000/emmintrin.h: Likewise.
14432 * config/rs6000/mmintrin.h: Likewise.
14433 * config/rs6000/xmmintrin.h: Likewise.
14435 2018-01-10 David Malcolm <dmalcolm@redhat.com>
14438 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
14440 * tree.c (tree_nop_conversion): Return true for location wrapper
14442 (maybe_wrap_with_location): New function.
14443 (selftest::check_strip_nops): New function.
14444 (selftest::test_location_wrappers): New function.
14445 (selftest::tree_c_tests): Call it.
14446 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
14447 (maybe_wrap_with_location): New decl.
14448 (EXPR_LOCATION_WRAPPER_P): New macro.
14449 (location_wrapper_p): New inline function.
14450 (tree_strip_any_location_wrapper): New inline function.
14452 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
14455 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
14456 stack_realign_offset for the largest alignment of stack slot
14458 (ix86_find_max_used_stack_alignment): New function.
14459 (ix86_finalize_stack_frame_flags): Use it. Set
14460 max_used_stack_alignment if we don't realign stack.
14461 * config/i386/i386.h (machine_function): Add
14462 max_used_stack_alignment.
14464 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
14466 * config/arm/arm.opt (-mbranch-cost): New option.
14467 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
14470 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
14473 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
14474 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
14476 2018-01-10 Richard Biener <rguenther@suse.de>
14479 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
14480 early out so it also covers the case where we have a non-NULL
14483 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
14485 PR tree-optimization/83753
14486 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
14487 for non-strided grouped accesses if the number of elements is 1.
14489 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
14492 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
14493 * i386.h (TARGET_USE_GATHER): Define.
14494 * x86-tune.def (X86_TUNE_USE_GATHER): New.
14496 2018-01-10 Martin Liska <mliska@suse.cz>
14499 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
14500 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
14502 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
14503 CLEANUP_NO_PARTITIONING is not set.
14505 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
14507 * doc/rtl.texi: Remove documentation of (const ...) wrappers
14508 for vectors, as a partial revert of r254296.
14509 * rtl.h (const_vec_p): Delete.
14510 (const_vec_duplicate_p): Don't test for vector CONSTs.
14511 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
14512 * expmed.c (make_tree): Likewise.
14515 * common.md (E, F): Use CONSTANT_P instead of checking for
14517 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
14518 checking for CONST_VECTOR.
14520 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
14522 PR middle-end/83575
14523 * predict.c (force_edge_cold): Handle in more sane way edges
14524 with no prediction.
14526 2018-01-09 Carl Love <cel@us.ibm.com>
14528 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
14530 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
14531 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
14532 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
14533 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
14534 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
14535 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
14536 * config/rs6000/rs6000-protos.h: Add extern defition for
14537 rs6000_generate_float2_double_code.
14538 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
14540 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
14541 (float2_v2df): Add define_expand.
14543 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
14546 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
14547 op_mode in the force_to_mode call.
14549 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
14551 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
14552 instead of checking each element individually.
14553 (aarch64_evpc_uzp): Likewise.
14554 (aarch64_evpc_zip): Likewise.
14555 (aarch64_evpc_ext): Likewise.
14556 (aarch64_evpc_rev): Likewise.
14557 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
14558 instead of checking each element individually. Return true without
14560 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
14561 whether all selected elements come from the same input, instead of
14562 checking each element individually. Remove calls to gen_rtx_REG,
14563 start_sequence and end_sequence and instead assert that no rtl is
14566 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
14568 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
14569 order of HIGH and CONST checks.
14571 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
14573 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
14574 if the destination isn't an SSA_NAME.
14576 2018-01-09 Richard Biener <rguenther@suse.de>
14578 PR tree-optimization/83668
14579 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
14581 (canonicalize_loop_form): ... here, renamed from ...
14582 (canonicalize_loop_closed_ssa_form): ... this and amended to
14583 swap successor edges for loop exit blocks to make us use
14584 the RPO order we need for initial schedule generation.
14586 2018-01-09 Joseph Myers <joseph@codesourcery.com>
14588 PR tree-optimization/64811
14589 * match.pd: When optimizing comparisons with Inf, avoid
14590 introducing or losing exceptions from comparisons with NaN.
14592 2018-01-09 Martin Liska <mliska@suse.cz>
14595 * asan.c (shadow_mem_size): Add gcc_assert.
14597 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
14599 Don't save registers in main().
14602 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
14603 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
14604 * config/avr/avr.c (avr_set_current_function): Don't error if
14605 naked, OS_task or OS_main are specified at the same time.
14606 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
14608 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
14610 * common/config/avr/avr-common.c (avr_option_optimization_table):
14611 Switch on -mmain-is-OS_task for optimizing compilations.
14613 2018-01-09 Richard Biener <rguenther@suse.de>
14615 PR tree-optimization/83572
14616 * graphite.c: Include cfganal.h.
14617 (graphite_transform_loops): Connect infinite loops to exit
14618 and remove fake edges at the end.
14620 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
14622 * ipa-inline.c (edge_badness): Revert accidental checkin.
14624 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
14627 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
14628 symbols; not inline clones.
14630 2018-01-09 Jakub Jelinek <jakub@redhat.com>
14633 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
14634 hard registers. Formatting fixes.
14636 PR preprocessor/83722
14637 * gcc.c (try_generate_repro): Pass
14638 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
14639 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
14642 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
14643 Kito Cheng <kito.cheng@gmail.com>
14645 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
14646 (riscv_leaf_function_p): Delete.
14647 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
14649 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14651 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
14653 (do_ifelse): New function.
14654 (do_isel): New function.
14655 (do_sub3): New function.
14656 (do_add3): New function.
14657 (do_load_mask_compare): New function.
14658 (do_overlap_load_compare): New function.
14659 (expand_compare_loop): New function.
14660 (expand_block_compare): Call expand_compare_loop() when appropriate.
14661 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
14662 option description.
14663 (-mblock-compare-inline-loop-limit): New option.
14665 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14668 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
14669 Reverse order of second and third operands in first alternative.
14670 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
14671 of first and second elements in UNSPEC_VPERMR vector.
14672 (altivec_expand_vec_perm_le): Likewise.
14674 2018-01-08 Jeff Law <law@redhat.com>
14676 PR rtl-optimizatin/81308
14677 * tree-switch-conversion.c (cfg_altered): New file scoped static.
14678 (process_switch): If group_case_labels makes a change, then set
14680 (pass_convert_switch::execute): If a switch is converted, then
14681 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
14683 PR rtl-optimization/81308
14684 * recog.c (split_all_insns): Conditionally cleanup the CFG after
14687 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
14689 PR target/83663 - Revert r255946
14690 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
14691 generation for cases where splatting a value is not useful.
14692 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
14693 across a vec_duplicate and a paradoxical subreg forming a vector
14694 mode to a vec_concat.
14696 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14698 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
14699 -march=armv8.3-a variants.
14700 * config/arm/t-multilib: Likewise.
14701 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
14703 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14705 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
14707 (cceq_ior_compare_complement): Give it a name so I can use it, and
14708 change boolean_or_operator predicate to boolean_operator so it can
14709 be used to generate a crand.
14710 (eqne): New code iterator.
14711 (bd/bd_neg): New code_attrs.
14712 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
14713 a single define_insn.
14714 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
14715 decrement (bdnzt/bdnzf/bdzt/bdzf).
14716 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
14717 with the new names of the branch decrement patterns, and added the
14718 names of the branch decrement conditional patterns.
14720 2018-01-08 Richard Biener <rguenther@suse.de>
14722 PR tree-optimization/83563
14723 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
14726 2018-01-08 Richard Biener <rguenther@suse.de>
14728 PR middle-end/83713
14729 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
14731 2018-01-08 Richard Biener <rguenther@suse.de>
14733 PR tree-optimization/83685
14734 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
14735 references to abnormals.
14737 2018-01-08 Richard Biener <rguenther@suse.de>
14740 * dwarf2out.c (output_indirect_strings): Handle empty
14741 skeleton_debug_str_hash.
14742 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
14744 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
14746 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
14747 (emit_store_direct): Likewise.
14748 (arc_trampoline_adjust_address): Likewise.
14749 (arc_asm_trampoline_template): New function.
14750 (arc_initialize_trampoline): Use asm_trampoline_template.
14751 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
14752 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
14753 * config/arc/arc.md (flush_icache): Delete pattern.
14755 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
14757 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
14758 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
14761 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
14764 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
14765 by not USED_FOR_TARGET.
14766 (make_pass_resolve_sw_modes): Likewise.
14768 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
14770 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
14773 2018-01-08 Richard Biener <rguenther@suse.de>
14775 PR middle-end/83580
14776 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
14778 2018-01-08 Richard Biener <rguenther@suse.de>
14780 PR middle-end/83517
14781 * match.pd ((t * 2) / 2) -> t): Add missing :c.
14783 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
14785 PR middle-end/81897
14786 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
14787 basic blocks with a small number of successors.
14788 (convert_control_dep_chain_into_preds): Improve handling of
14790 (dump_predicates): Split apart into...
14791 (dump_pred_chain): ...here...
14792 (dump_pred_info): ...and here.
14793 (can_one_predicate_be_invalidated_p): Add debugging printfs.
14794 (can_chain_union_be_invalidated_p): Improve check for invalidation
14796 (uninit_uses_cannot_happen): Avoid unnecessary if
14797 convert_control_dep_chain_into_preds yielded nothing.
14799 2018-01-06 Martin Sebor <msebor@redhat.com>
14801 PR tree-optimization/83640
14802 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
14803 subtracting negative offset from size.
14804 (builtin_access::overlap): Adjust offset bounds of the access to fall
14805 within the size of the object if possible.
14807 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
14809 PR rtl-optimization/83699
14810 * expmed.c (extract_bit_field_1): Restrict the vector usage of
14811 extract_bit_field_as_subreg to cases in which the extracted
14812 value is also a vector.
14814 * lra-constraints.c (process_alt_operands): Test for the equivalence
14815 substitutions when detecting a possible reload cycle.
14817 2018-01-06 Jakub Jelinek <jakub@redhat.com>
14820 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
14821 by default if flag_selective_schedling{,2}. Formatting fixes.
14823 PR rtl-optimization/83682
14824 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
14825 if it has non-VECTOR_MODE element mode.
14826 (vec_duplicate_p): Likewise.
14828 PR middle-end/83694
14829 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
14830 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
14832 2018-01-05 Jakub Jelinek <jakub@redhat.com>
14835 * config/i386/i386-builtin.def
14836 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
14837 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
14838 Require also OPTION_MASK_ISA_AVX512F in addition to
14839 OPTION_MASK_ISA_GFNI.
14840 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
14841 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
14842 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
14843 to OPTION_MASK_ISA_GFNI.
14844 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
14845 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
14846 OPTION_MASK_ISA_AVX512BW.
14847 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
14848 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
14849 addition to OPTION_MASK_ISA_GFNI.
14850 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
14851 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
14852 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
14853 to OPTION_MASK_ISA_GFNI.
14854 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
14855 a requirement for all ISAs rather than any of them with a few
14857 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
14859 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
14860 bitmasks to be enabled with 3 exceptions, instead of requiring any
14861 enabled ISA with lots of exceptions.
14862 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
14863 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
14864 Change avx512bw in isa attribute to avx512f.
14865 * config/i386/sgxintrin.h: Add license boilerplate.
14866 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
14867 to __AVX512F__ and __AVX512VL to __AVX512VL__.
14868 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
14869 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
14871 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
14872 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
14873 temporarily sse2 rather than sse if not enabled already.
14876 * config/i386/sse.md (VI248_VLBW): Rename to ...
14877 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
14878 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
14879 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
14880 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
14881 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
14882 mode iterator instead of VI248_VLBW.
14884 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
14886 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
14887 (record_modified): Skip clobbers; add debug output.
14888 (param_change_prob): Use sreal frequencies.
14890 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
14892 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14893 punt for user-aligned variables.
14895 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
14897 * tree-chrec.c (chrec_contains_symbols): Return true for
14900 2018-01-05 Sudakshina Das <sudi.das@arm.com>
14903 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
14904 of (x|y) == x for BICS pattern.
14906 2018-01-05 Jakub Jelinek <jakub@redhat.com>
14908 PR tree-optimization/83605
14909 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
14910 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
14913 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
14915 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
14916 * config/epiphany/rtems.h: New file.
14918 2018-01-04 Jakub Jelinek <jakub@redhat.com>
14919 Uros Bizjak <ubizjak@gmail.com>
14922 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
14923 QIreg_operand instead of register_operand predicate.
14924 * config/i386/i386.c (ix86_rop_should_change_byte_p,
14925 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
14926 comments instead of -fmitigate[-_]rop.
14928 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14931 * cgraphunit.c (symbol_table::compile): Switch to text_section
14932 before calling assembly_start debug hook.
14933 * run-rtl-passes.c (run_rtl_passes): Likewise.
14936 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14938 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14939 range_int_cst_p rather than !symbolic_range_p before calling
14940 extract_range_from_multiplicative_op_1.
14942 2018-01-04 Jeff Law <law@redhat.com>
14944 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
14945 redundant test in assertion.
14947 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14949 * doc/rtl.texi: Document machine_mode wrapper classes.
14951 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14953 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
14954 using tree_to_uhwi.
14956 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14958 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
14959 the VEC_PERM_EXPR fold to fail.
14961 2018-01-04 Jakub Jelinek <jakub@redhat.com>
14964 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
14965 to switched_sections.
14967 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14970 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
14971 test for d.testing.
14973 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
14976 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
14977 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
14979 2018-01-04 Jakub Jelinek <jakub@redhat.com>
14982 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
14983 is BLKmode and bitpos not zero or mode change is needed.
14985 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
14988 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
14991 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
14994 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
14995 instead of MULT rtx. Update all corresponding splitters.
14996 (*saddl_se): Ditto.
14997 (*ssub<modesuffix>): Ditto.
14998 (*ssubl_se): Ditto.
14999 (*cmp_sadd_di): Update split patterns.
15000 (*cmp_sadd_si): Ditto.
15001 (*cmp_sadd_sidi): Ditto.
15002 (*cmp_ssub_di): Ditto.
15003 (*cmp_ssub_si): Ditto.
15004 (*cmp_ssub_sidi): Ditto.
15005 * config/alpha/predicates.md (const23_operand): New predicate.
15006 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
15007 Look for ASHIFT, not MULT inner operand.
15008 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
15010 2018-01-04 Martin Liska <mliska@suse.cz>
15012 PR gcov-profile/83669
15013 * gcov.c (output_intermediate_file): Add version to intermediate
15015 * doc/gcov.texi: Document new field 'version' in intermediate
15016 file format. Fix location of '-k' option of gcov command.
15018 2018-01-04 Martin Liska <mliska@suse.cz>
15021 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
15023 2018-01-04 Jakub Jelinek <jakub@redhat.com>
15025 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
15027 2018-01-03 Martin Sebor <msebor@redhat.com>
15029 PR tree-optimization/83655
15030 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
15031 checking calls with invalid arguments.
15033 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15035 * tree-vect-stmts.c (vect_get_store_rhs): New function.
15036 (vectorizable_mask_load_store): Delete.
15037 (vectorizable_call): Return false for masked loads and stores.
15038 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
15039 instead of gimple_assign_rhs1.
15040 (vectorizable_load): Handle IFN_MASK_LOAD.
15041 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
15043 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15045 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
15047 (vectorizable_mask_load_store): ...here.
15048 (vectorizable_load): ...and here.
15050 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15052 * tree-vect-stmts.c (vect_build_all_ones_mask)
15053 (vect_build_zero_merge_argument): New functions, split out from...
15054 (vectorizable_load): ...here.
15056 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15058 * tree-vect-stmts.c (vect_check_store_rhs): New function,
15060 (vectorizable_mask_load_store): ...here.
15061 (vectorizable_store): ...and here.
15063 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15065 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
15067 (vectorizable_mask_load_store): ...here.
15069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15071 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
15072 (vect_model_store_cost): Take a vec_load_store_type instead of a
15074 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
15075 (vect_model_store_cost): Take a vec_load_store_type instead of a
15077 (vectorizable_mask_load_store): Update accordingly.
15078 (vectorizable_store): Likewise.
15079 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
15081 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15083 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
15084 IFN_MASK_LOAD calls here rather than...
15085 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
15087 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15088 Alan Hayward <alan.hayward@arm.com>
15089 David Sherwood <david.sherwood@arm.com>
15091 * expmed.c (extract_bit_field_1): For vector extracts,
15092 fall back to extract_bit_field_as_subreg if vec_extract
15095 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15096 Alan Hayward <alan.hayward@arm.com>
15097 David Sherwood <david.sherwood@arm.com>
15099 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
15100 they are variable or constant sized.
15101 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
15102 slots for constant-sized data.
15104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15105 Alan Hayward <alan.hayward@arm.com>
15106 David Sherwood <david.sherwood@arm.com>
15108 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
15109 handling COND_EXPRs with boolean comparisons, try to find a better
15110 basis for the mask type than the boolean itself.
15112 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15114 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
15115 is calculated and how it can be overridden.
15116 * genmodes.c (max_bitsize_mode_any_mode): New variable.
15117 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
15119 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
15122 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15123 Alan Hayward <alan.hayward@arm.com>
15124 David Sherwood <david.sherwood@arm.com>
15126 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
15127 Remove the mode argument.
15128 (aarch64_simd_valid_immediate): Remove the mode and inverse
15130 * config/aarch64/iterators.md (bitsize): New iterator.
15131 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
15132 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
15133 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
15134 aarch64_simd_valid_immediate.
15135 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
15136 (aarch64_reg_or_bic_imm): Likewise.
15137 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
15138 with an insn_type enum and msl with a modifier_type enum.
15139 Replace element_width with a scalar_mode. Change the shift
15140 to unsigned int. Add constructors for scalar_float_mode and
15141 scalar_int_mode elements.
15142 (aarch64_vect_float_const_representable_p): Delete.
15143 (aarch64_can_const_movi_rtx_p)
15144 (aarch64_simd_scalar_immediate_valid_for_move)
15145 (aarch64_simd_make_constant): Update call to
15146 aarch64_simd_valid_immediate.
15147 (aarch64_advsimd_valid_immediate_hs): New function.
15148 (aarch64_advsimd_valid_immediate): Likewise.
15149 (aarch64_simd_valid_immediate): Remove mode and inverse
15150 arguments. Rewrite to use the above. Use const_vec_duplicate_p
15151 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
15152 and aarch64_float_const_representable_p on the result.
15153 (aarch64_output_simd_mov_immediate): Remove mode argument.
15154 Update call to aarch64_simd_valid_immediate and use of
15155 simd_immediate_info.
15156 (aarch64_output_scalar_simd_mov_immediate): Update call
15159 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15160 Alan Hayward <alan.hayward@arm.com>
15161 David Sherwood <david.sherwood@arm.com>
15163 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
15164 (mode_nunits): Likewise CONST_MODE_NUNITS.
15165 * machmode.def (ADJUST_NUNITS): Document.
15166 * genmodes.c (mode_data::need_nunits_adj): New field.
15167 (blank_mode): Update accordingly.
15168 (adj_nunits): New variable.
15169 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
15171 (emit_mode_size_inline): Set need_bytesize_adj for all modes
15172 listed in adj_nunits.
15173 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
15174 listed in adj_nunits. Don't emit case statements for such modes.
15175 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
15176 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
15177 nothing if adj_nunits is nonnull.
15178 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
15179 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
15180 (emit_mode_fbit): Update use of print_maybe_const_decl.
15181 (emit_move_size): Likewise. Treat the array as non-const
15183 (emit_mode_adjustments): Handle adj_nunits.
15185 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15187 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
15188 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
15189 (VECTOR_MODES): Use it.
15190 (make_vector_modes): Take the prefix as an argument.
15192 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15193 Alan Hayward <alan.hayward@arm.com>
15194 David Sherwood <david.sherwood@arm.com>
15196 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
15197 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
15198 for MODE_VECTOR_BOOL.
15199 * machmode.def (VECTOR_BOOL_MODE): Document.
15200 * genmodes.c (VECTOR_BOOL_MODE): New macro.
15201 (make_vector_bool_mode): New function.
15202 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
15204 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15205 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
15207 * stor-layout.c (int_mode_for_mode): Likewise.
15208 * tree.c (build_vector_type_for_mode): Likewise.
15209 * varasm.c (output_constant_pool_2): Likewise.
15210 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
15211 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
15212 for MODE_VECTOR_BOOL.
15213 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
15214 of mode class checks.
15215 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
15216 instead of a list of mode class checks.
15217 (expand_vector_scalar_condition): Likewise.
15218 (type_for_widest_vector_mode): Handle BImode as an inner mode.
15220 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15221 Alan Hayward <alan.hayward@arm.com>
15222 David Sherwood <david.sherwood@arm.com>
15224 * machmode.h (mode_size): Change from unsigned short to
15226 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
15227 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
15228 or if measurement_type is not polynomial.
15229 (fixed_size_mode::includes_p): Check for constant-sized modes.
15230 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
15231 return a poly_uint16 rather than an unsigned short.
15232 (emit_mode_size): Change the type of mode_size from unsigned short
15233 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
15234 (emit_mode_adjustments): Cope with polynomial vector sizes.
15235 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
15237 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
15239 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
15240 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
15241 * caller-save.c (setup_save_areas): Likewise.
15242 (replace_reg_with_saved_mem): Likewise.
15243 * calls.c (emit_library_call_value_1): Likewise.
15244 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
15245 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
15246 (gen_lowpart_for_combine): Likewise.
15247 * convert.c (convert_to_integer_1): Likewise.
15248 * cse.c (equiv_constant, cse_insn): Likewise.
15249 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
15250 (cselib_subst_to_values): Likewise.
15251 * dce.c (word_dce_process_block): Likewise.
15252 * df-problems.c (df_word_lr_mark_ref): Likewise.
15253 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
15254 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
15255 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
15256 (rtl_for_decl_location): Likewise.
15257 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
15258 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
15259 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
15260 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
15261 (expand_expr_real_1): Likewise.
15262 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
15263 (pad_below): Likewise.
15264 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
15265 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
15266 * ira.c (get_subreg_tracking_sizes): Likewise.
15267 * ira-build.c (ira_create_allocno_objects): Likewise.
15268 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
15269 (ira_sort_regnos_for_alter_reg): Likewise.
15270 * ira-costs.c (record_operand_costs): Likewise.
15271 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
15272 (resolve_simple_move): Likewise.
15273 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
15274 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
15275 (lra_constraints): Likewise.
15276 (CONST_POOL_OK_P): Reject variable-sized modes.
15277 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
15278 (add_pseudo_to_slot, lra_spill): Likewise.
15279 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15280 * optabs-query.c (get_best_extraction_insn): Likewise.
15281 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
15282 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
15283 (expand_mult_highpart, valid_multiword_target_p): Likewise.
15284 * recog.c (offsettable_address_addr_space_p): Likewise.
15285 * regcprop.c (maybe_mode_change): Likewise.
15286 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
15287 * regrename.c (build_def_use): Likewise.
15288 * regstat.c (dump_reg_info): Likewise.
15289 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
15290 (find_reloads, find_reloads_subreg_address): Likewise.
15291 * reload1.c (eliminate_regs_1): Likewise.
15292 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
15293 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
15294 (simplify_binary_operation_1, simplify_subreg): Likewise.
15295 * targhooks.c (default_function_arg_padding): Likewise.
15296 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
15297 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
15298 (verify_gimple_assign_ternary): Likewise.
15299 * tree-inline.c (estimate_move_cost): Likewise.
15300 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15301 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
15302 (get_address_cost_ainc): Likewise.
15303 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
15304 (vect_supportable_dr_alignment): Likewise.
15305 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15306 (vectorizable_reduction): Likewise.
15307 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
15308 (vectorizable_operation, vectorizable_load): Likewise.
15309 * tree.c (build_same_sized_truth_vector_type): Likewise.
15310 * valtrack.c (cleanup_auto_inc_dec): Likewise.
15311 * var-tracking.c (emit_note_insn_var_location): Likewise.
15312 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
15313 (ADDR_VEC_ALIGN): Likewise.
15315 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15316 Alan Hayward <alan.hayward@arm.com>
15317 David Sherwood <david.sherwood@arm.com>
15319 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
15321 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
15322 or if measurement_type is polynomial.
15323 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
15324 * combine.c (make_extraction): Likewise.
15325 * dse.c (find_shift_sequence): Likewise.
15326 * dwarf2out.c (mem_loc_descriptor): Likewise.
15327 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
15328 (extract_bit_field, extract_low_bits): Likewise.
15329 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
15330 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
15331 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
15332 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
15333 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
15334 * reload.c (find_reloads): Likewise.
15335 * reload1.c (alter_reg): Likewise.
15336 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
15337 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
15338 * tree-if-conv.c (predicate_mem_writes): Likewise.
15339 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15340 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
15341 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15342 * valtrack.c (dead_debug_insert_temp): Likewise.
15343 * varasm.c (mergeable_constant_section): Likewise.
15344 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
15346 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15347 Alan Hayward <alan.hayward@arm.com>
15348 David Sherwood <david.sherwood@arm.com>
15350 * expr.c (expand_assignment): Cope with polynomial mode sizes
15351 when assigning to a CONCAT.
15353 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15354 Alan Hayward <alan.hayward@arm.com>
15355 David Sherwood <david.sherwood@arm.com>
15357 * machmode.h (mode_precision): Change from unsigned short to
15359 (mode_to_precision): Return a poly_uint16 rather than an unsigned
15361 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
15362 or if measurement_type is not polynomial.
15363 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
15364 in which the mode is already known to be a scalar_int_mode.
15365 * genmodes.c (emit_mode_precision): Change the type of mode_precision
15366 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
15368 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
15369 for GET_MODE_PRECISION.
15370 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
15371 for GET_MODE_PRECISION.
15372 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
15374 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
15375 (expand_field_assignment, make_extraction): Likewise.
15376 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
15377 (get_last_value): Likewise.
15378 * convert.c (convert_to_integer_1): Likewise.
15379 * cse.c (cse_insn): Likewise.
15380 * expr.c (expand_expr_real_1): Likewise.
15381 * lra-constraints.c (simplify_operand_subreg): Likewise.
15382 * optabs-query.c (can_atomic_load_p): Likewise.
15383 * optabs.c (expand_atomic_load): Likewise.
15384 (expand_atomic_store): Likewise.
15385 * ree.c (combine_reaching_defs): Likewise.
15386 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
15387 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
15388 * tree.h (type_has_mode_precision_p): Likewise.
15389 * ubsan.c (instrument_si_overflow): Likewise.
15391 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15392 Alan Hayward <alan.hayward@arm.com>
15393 David Sherwood <david.sherwood@arm.com>
15395 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
15396 polynomial numbers of units.
15397 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
15398 (valid_vector_subparts_p): New function.
15399 (build_vector_type): Remove temporary shim and take the number
15400 of units as a poly_uint64 rather than an int.
15401 (build_opaque_vector_type): Take the number of units as a
15402 poly_uint64 rather than an int.
15403 * tree.c (build_vector_from_ctor): Handle polynomial
15404 TYPE_VECTOR_SUBPARTS.
15405 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
15406 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
15407 (build_vector_from_val): If the number of units is variable,
15408 use build_vec_duplicate_cst for constant operands and
15409 VEC_DUPLICATE_EXPR otherwise.
15410 (make_vector_type): Remove temporary is_constant ().
15411 (build_vector_type, build_opaque_vector_type): Take the number of
15412 units as a poly_uint64 rather than an int.
15413 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
15415 * cfgexpand.c (expand_debug_expr): Likewise.
15416 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
15417 (store_constructor, expand_expr_real_1): Likewise.
15418 (const_scalar_mask_from_tree): Likewise.
15419 * fold-const-call.c (fold_const_reduction): Likewise.
15420 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
15421 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
15422 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
15423 (fold_relational_const): Likewise.
15424 (native_interpret_vector): Likewise. Change the size from an
15425 int to an unsigned int.
15426 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
15427 TYPE_VECTOR_SUBPARTS.
15428 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
15429 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
15430 duplicating a non-constant operand into a variable-length vector.
15431 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
15432 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
15433 * ipa-icf.c (sem_variable::equals): Likewise.
15434 * match.pd: Likewise.
15435 * omp-simd-clone.c (simd_clone_subparts): Likewise.
15436 * print-tree.c (print_node): Likewise.
15437 * stor-layout.c (layout_type): Likewise.
15438 * targhooks.c (default_builtin_vectorization_cost): Likewise.
15439 * tree-cfg.c (verify_gimple_comparison): Likewise.
15440 (verify_gimple_assign_binary): Likewise.
15441 (verify_gimple_assign_ternary): Likewise.
15442 (verify_gimple_assign_single): Likewise.
15443 * tree-pretty-print.c (dump_generic_node): Likewise.
15444 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15445 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
15446 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
15447 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
15448 (vect_shift_permute_load_chain): Likewise.
15449 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
15450 (expand_vector_condition, optimize_vector_constructor): Likewise.
15451 (lower_vec_perm, get_compute_type): Likewise.
15452 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15453 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
15454 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
15455 (vect_recog_mask_conversion_pattern): Likewise.
15456 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
15457 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
15458 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
15459 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
15460 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
15461 (vectorizable_shift, vectorizable_operation, vectorizable_store)
15462 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
15463 (supportable_widening_operation): Likewise.
15464 (supportable_narrowing_operation): Likewise.
15465 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
15467 * varasm.c (output_constant): Likewise.
15469 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15470 Alan Hayward <alan.hayward@arm.com>
15471 David Sherwood <david.sherwood@arm.com>
15473 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
15474 so that both the length == 3 and length != 3 cases set up their
15475 own permute vectors. Add comments explaining why we know the
15476 number of elements is constant.
15477 (vect_permute_load_chain): Likewise.
15479 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15480 Alan Hayward <alan.hayward@arm.com>
15481 David Sherwood <david.sherwood@arm.com>
15483 * machmode.h (mode_nunits): Change from unsigned char to
15485 (ONLY_FIXED_SIZE_MODES): New macro.
15486 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
15487 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
15488 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
15490 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
15491 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
15492 or if measurement_type is not polynomial.
15493 * genmodes.c (ZERO_COEFFS): New macro.
15494 (emit_mode_nunits_inline): Make mode_nunits_inline return a
15496 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
15497 Use ZERO_COEFFS when emitting initializers.
15498 * data-streamer.h (bp_pack_poly_value): New function.
15499 (bp_unpack_poly_value): Likewise.
15500 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
15501 for GET_MODE_NUNITS.
15502 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
15503 for GET_MODE_NUNITS.
15504 * tree.c (make_vector_type): Remove temporary shim and make
15505 the real function take the number of units as a poly_uint64
15506 rather than an int.
15507 (build_vector_type_for_mode): Handle polynomial nunits.
15508 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
15509 * emit-rtl.c (const_vec_series_p_1): Likewise.
15510 (gen_rtx_CONST_VECTOR): Likewise.
15511 * fold-const.c (test_vec_duplicate_folding): Likewise.
15512 * genrecog.c (validate_pattern): Likewise.
15513 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
15514 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
15515 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
15516 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
15517 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
15518 * rtlanal.c (subreg_get_info): Likewise.
15519 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15520 (vect_grouped_load_supported): Likewise.
15521 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15522 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
15523 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15524 (simplify_const_unary_operation, simplify_binary_operation_1)
15525 (simplify_const_binary_operation, simplify_ternary_operation)
15526 (test_vector_ops_duplicate, test_vector_ops): Likewise.
15527 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
15528 instead of CONST_VECTOR_NUNITS.
15529 * varasm.c (output_constant_pool_2): Likewise.
15530 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
15531 explicit-encoded elements in the XVEC for variable-length vectors.
15533 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15535 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
15537 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15538 Alan Hayward <alan.hayward@arm.com>
15539 David Sherwood <david.sherwood@arm.com>
15541 * coretypes.h (fixed_size_mode): Declare.
15542 (fixed_size_mode_pod): New typedef.
15543 * builtins.h (target_builtins::x_apply_args_mode)
15544 (target_builtins::x_apply_result_mode): Change type to
15545 fixed_size_mode_pod.
15546 * builtins.c (apply_args_size, apply_result_size, result_vector)
15547 (expand_builtin_apply_args_1, expand_builtin_apply)
15548 (expand_builtin_return): Update accordingly.
15550 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15552 * cse.c (hash_rtx_cb): Hash only the encoded elements.
15553 * cselib.c (cselib_hash_rtx): Likewise.
15554 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
15555 CONST_VECTOR encoding.
15557 2018-01-03 Jakub Jelinek <jakub@redhat.com>
15558 Jeff Law <law@redhat.com>
15561 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
15562 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
15563 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
15564 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
15567 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
15568 explicitly probe *sp in a noreturn function if there were any callee
15569 register saves or frame pointer is needed.
15571 2018-01-03 Jakub Jelinek <jakub@redhat.com>
15574 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
15575 BLKmode for ternary, binary or unary expressions.
15578 * var-tracking.c (delete_vta_debug_insn): New inline function.
15579 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
15580 insns from get_insns () to NULL instead of each bb separately.
15581 Use delete_vta_debug_insn. No longer static.
15582 (vt_debug_insns_local, variable_tracking_main_1): Adjust
15583 delete_vta_debug_insns callers.
15584 * rtl.h (delete_vta_debug_insns): Declare.
15585 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
15586 instead of variable_tracking_main.
15588 2018-01-03 Martin Sebor <msebor@redhat.com>
15590 PR tree-optimization/83603
15591 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
15592 arguments past the endof the argument list in functions declared
15593 without a prototype.
15594 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
15595 Avoid checking when arguments are null.
15597 2018-01-03 Martin Sebor <msebor@redhat.com>
15600 * doc/extend.texi (attribute const): Fix a typo.
15601 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
15602 issuing -Wsuggest-attribute for void functions.
15604 2018-01-03 Martin Sebor <msebor@redhat.com>
15606 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
15607 offset_int::from instead of wide_int::to_shwi.
15608 (maybe_diag_overlap): Remove assertion.
15609 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
15610 * gimple-ssa-sprintf.c (format_directive): Same.
15611 (parse_directive): Same.
15612 (sprintf_dom_walker::compute_format_length): Same.
15613 (try_substitute_return_value): Same.
15615 2018-01-03 Jeff Law <law@redhat.com>
15617 PR middle-end/83654
15618 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
15619 non-constant residual for zero at runtime and avoid probing in
15620 that case. Reorganize code for trailing problem to mirror handling
15623 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15625 PR tree-optimization/83501
15626 * tree-ssa-strlen.c (get_string_cst): New.
15627 (handle_char_store): Call get_string_cst.
15629 2018-01-03 Martin Liska <mliska@suse.cz>
15631 PR tree-optimization/83593
15632 * tree-ssa-strlen.c: Include tree-cfg.h.
15633 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
15634 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
15635 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
15637 (strlen_dom_walker::before_dom_children): Call
15638 gimple_purge_dead_eh_edges. Dump tranformation with details
15640 (strlen_dom_walker::before_dom_children): Update call by adding
15641 new argument cleanup_eh.
15642 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
15644 2018-01-03 Martin Liska <mliska@suse.cz>
15647 * cif-code.def (VARIADIC_THUNK): New enum value.
15648 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
15651 2018-01-03 Jan Beulich <jbeulich@suse.com>
15653 * sse.md (mov<mode>_internal): Tighten condition for when to use
15654 vmovdqu<ssescalarsize> for TI and OI modes.
15656 2018-01-03 Jakub Jelinek <jakub@redhat.com>
15658 Update copyright years.
15660 2018-01-03 Martin Liska <mliska@suse.cz>
15663 * ipa-visibility.c (function_and_variable_visibility): Skip
15664 functions with noipa attribure.
15666 2018-01-03 Jakub Jelinek <jakub@redhat.com>
15668 * gcc.c (process_command): Update copyright notice dates.
15669 * gcov-dump.c (print_version): Ditto.
15670 * gcov.c (print_version): Ditto.
15671 * gcov-tool.c (print_version): Ditto.
15672 * gengtype.c (create_file): Ditto.
15673 * doc/cpp.texi: Bump @copying's copyright year.
15674 * doc/cppinternals.texi: Ditto.
15675 * doc/gcc.texi: Ditto.
15676 * doc/gccint.texi: Ditto.
15677 * doc/gcov.texi: Ditto.
15678 * doc/install.texi: Ditto.
15679 * doc/invoke.texi: Ditto.
15681 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15683 * vector-builder.h (vector_builder::m_full_nelts): Change from
15684 unsigned int to poly_uint64.
15685 (vector_builder::full_nelts): Update prototype accordingly.
15686 (vector_builder::new_vector): Likewise.
15687 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
15688 (vector_builder::operator ==): Likewise.
15689 (vector_builder::finalize): Likewise.
15690 * int-vector-builder.h (int_vector_builder::int_vector_builder):
15691 Take the number of elements as a poly_uint64 rather than an
15693 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
15694 from unsigned int to poly_uint64.
15695 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
15696 (vec_perm_indices::new_vector): Likewise.
15697 (vec_perm_indices::length): Likewise.
15698 (vec_perm_indices::nelts_per_input): Likewise.
15699 (vec_perm_indices::input_nelts): Likewise.
15700 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
15701 number of elements per input as a poly_uint64 rather than an
15702 unsigned int. Use the original encoding for variable-length
15703 vectors, rather than clamping each individual element.
15704 For the second and subsequent elements in each pattern,
15705 clamp the step and base before clamping their sum.
15706 (vec_perm_indices::series_p): Handle polynomial element counts.
15707 (vec_perm_indices::all_in_range_p): Likewise.
15708 (vec_perm_indices_to_tree): Likewise.
15709 (vec_perm_indices_to_rtx): Likewise.
15710 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15711 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
15712 (tree_vector_builder::new_binary_operation): Handle polynomial
15713 element counts. Return false if we need to know the number
15714 of elements at compile time.
15715 * fold-const.c (fold_vec_perm): Punt if the number of elements
15716 isn't known at compile time.
15718 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15720 * vec-perm-indices.h (vec_perm_builder): Change element type
15721 from HOST_WIDE_INT to poly_int64.
15722 (vec_perm_indices::element_type): Update accordingly.
15723 (vec_perm_indices::clamp): Handle polynomial element_types.
15724 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
15725 (vec_perm_indices::all_in_range_p): Likewise.
15726 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
15728 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
15729 polynomial vec_perm_indices element types.
15730 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
15731 * fold-const.c (fold_vec_perm): Likewise.
15732 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
15733 * tree-vect-generic.c (lower_vec_perm): Likewise.
15734 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15735 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
15736 element type to HOST_WIDE_INT.
15738 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15739 Alan Hayward <alan.hayward@arm.com>
15740 David Sherwood <david.sherwood@arm.com>
15742 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
15743 rather than an int. Use plus_constant.
15744 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
15745 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
15747 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15748 Alan Hayward <alan.hayward@arm.com>
15749 David Sherwood <david.sherwood@arm.com>
15751 * calls.c (emit_call_1, expand_call): Change struct_value_size from
15752 a HOST_WIDE_INT to a poly_int64.
15754 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15755 Alan Hayward <alan.hayward@arm.com>
15756 David Sherwood <david.sherwood@arm.com>
15758 * calls.c (load_register_parameters): Cope with polynomial
15759 mode sizes. Require a constant size for BLKmode parameters
15760 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
15761 forces a parameter to be padded at the lsb end in order to
15762 fill a complete number of words, require the parameter size
15763 to be ordered wrt UNITS_PER_WORD.
15765 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15766 Alan Hayward <alan.hayward@arm.com>
15767 David Sherwood <david.sherwood@arm.com>
15769 * reload1.c (spill_stack_slot_width): Change element type
15770 from unsigned int to poly_uint64_pod.
15771 (alter_reg): Treat mode sizes as polynomial.
15773 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15774 Alan Hayward <alan.hayward@arm.com>
15775 David Sherwood <david.sherwood@arm.com>
15777 * reload.c (complex_word_subreg_p): New function.
15778 (reload_inner_reg_of_subreg, push_reload): Use it.
15780 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15781 Alan Hayward <alan.hayward@arm.com>
15782 David Sherwood <david.sherwood@arm.com>
15784 * lra-constraints.c (process_alt_operands): Reject matched
15785 operands whose sizes aren't ordered.
15786 (match_reload): Refer to this check here.
15788 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15789 Alan Hayward <alan.hayward@arm.com>
15790 David Sherwood <david.sherwood@arm.com>
15792 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
15793 that the mode size is in the set {1, 2, 4, 8, 16}.
15795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15796 Alan Hayward <alan.hayward@arm.com>
15797 David Sherwood <david.sherwood@arm.com>
15799 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
15800 Use plus_constant instead of gen_rtx_PLUS.
15802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15803 Alan Hayward <alan.hayward@arm.com>
15804 David Sherwood <david.sherwood@arm.com>
15806 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
15807 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
15808 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
15809 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
15810 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
15811 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
15812 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
15813 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
15814 * config/i386/i386.c (ix86_push_rounding): ...this new function.
15815 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
15817 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
15818 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
15819 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
15820 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
15821 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
15822 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
15823 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
15824 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
15825 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
15826 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
15828 * expr.c (emit_move_resolve_push): Treat the input and result
15829 of PUSH_ROUNDING as a poly_int64.
15830 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
15831 (emit_push_insn): Likewise.
15832 * lra-eliminations.c (mark_not_eliminable): Likewise.
15833 * recog.c (push_operand): Likewise.
15834 * reload1.c (elimination_effects): Likewise.
15835 * rtlanal.c (nonzero_bits1): Likewise.
15836 * calls.c (store_one_arg): Likewise. Require the padding to be
15837 known at compile time.
15839 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15840 Alan Hayward <alan.hayward@arm.com>
15841 David Sherwood <david.sherwood@arm.com>
15843 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
15844 Use plus_constant instead of gen_rtx_PLUS.
15846 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15847 Alan Hayward <alan.hayward@arm.com>
15848 David Sherwood <david.sherwood@arm.com>
15850 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
15851 rather than an int.
15853 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15854 Alan Hayward <alan.hayward@arm.com>
15855 David Sherwood <david.sherwood@arm.com>
15857 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
15858 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
15859 via stack temporaries. Treat the mode size as polynomial too.
15861 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15862 Alan Hayward <alan.hayward@arm.com>
15863 David Sherwood <david.sherwood@arm.com>
15865 * expr.c (expand_expr_real_2): When handling conversions involving
15866 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
15867 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
15868 as a poly_uint64 too.
15870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15871 Alan Hayward <alan.hayward@arm.com>
15872 David Sherwood <david.sherwood@arm.com>
15874 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
15876 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15877 Alan Hayward <alan.hayward@arm.com>
15878 David Sherwood <david.sherwood@arm.com>
15880 * combine.c (can_change_dest_mode): Handle polynomial
15881 REGMODE_NATURAL_SIZE.
15882 * expmed.c (store_bit_field_1): Likewise.
15883 * expr.c (store_constructor): Likewise.
15884 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
15885 and polynomial REGMODE_NATURAL_SIZE.
15886 (gen_lowpart_common): Likewise.
15887 * reginfo.c (record_subregs_of_mode): Likewise.
15888 * rtlanal.c (read_modify_subreg_p): Likewise.
15890 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15891 Alan Hayward <alan.hayward@arm.com>
15892 David Sherwood <david.sherwood@arm.com>
15894 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
15895 numbers of elements.
15897 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15898 Alan Hayward <alan.hayward@arm.com>
15899 David Sherwood <david.sherwood@arm.com>
15901 * match.pd: Cope with polynomial numbers of vector elements.
15903 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15904 Alan Hayward <alan.hayward@arm.com>
15905 David Sherwood <david.sherwood@arm.com>
15907 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
15908 in a POINTER_PLUS_EXPR.
15910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15911 Alan Hayward <alan.hayward@arm.com>
15912 David Sherwood <david.sherwood@arm.com>
15914 * omp-simd-clone.c (simd_clone_subparts): New function.
15915 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
15916 (ipa_simd_modify_function_body): Likewise.
15918 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15919 Alan Hayward <alan.hayward@arm.com>
15920 David Sherwood <david.sherwood@arm.com>
15922 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
15923 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
15924 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
15925 (expand_vector_condition, vector_element): Likewise.
15926 (subparts_gt): New function.
15927 (get_compute_type): Use subparts_gt.
15928 (count_type_subparts): Delete.
15929 (expand_vector_operations_1): Use subparts_gt instead of
15930 count_type_subparts.
15932 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15933 Alan Hayward <alan.hayward@arm.com>
15934 David Sherwood <david.sherwood@arm.com>
15936 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
15937 (vect_compile_time_alias): ...this new function. Do the calculation
15938 on poly_ints rather than trees.
15939 (vect_prune_runtime_alias_test_list): Update call accordingly.
15941 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15942 Alan Hayward <alan.hayward@arm.com>
15943 David Sherwood <david.sherwood@arm.com>
15945 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
15947 (vect_schedule_slp_instance): Likewise.
15949 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15950 Alan Hayward <alan.hayward@arm.com>
15951 David Sherwood <david.sherwood@arm.com>
15953 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
15954 constant and extern definitions for variable-length vectors.
15955 (vect_get_constant_vectors): Note that the number of units
15956 is known to be constant.
15958 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15959 Alan Hayward <alan.hayward@arm.com>
15960 David Sherwood <david.sherwood@arm.com>
15962 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
15963 of units as polynomial. Choose between WIDE and NARROW based
15966 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15967 Alan Hayward <alan.hayward@arm.com>
15968 David Sherwood <david.sherwood@arm.com>
15970 * tree-vect-stmts.c (simd_clone_subparts): New function.
15971 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
15973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15974 Alan Hayward <alan.hayward@arm.com>
15975 David Sherwood <david.sherwood@arm.com>
15977 * tree-vect-stmts.c (vectorizable_call): Treat the number of
15978 vectors as polynomial. Use build_index_vector for
15979 IFN_GOMP_SIMD_LANE.
15981 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15982 Alan Hayward <alan.hayward@arm.com>
15983 David Sherwood <david.sherwood@arm.com>
15985 * tree-vect-stmts.c (get_load_store_type): Treat the number of
15986 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
15987 for variable-length vectors.
15988 (vectorizable_mask_load_store): Treat the number of units as
15989 polynomial, asserting that it is constant if the condition has
15990 already been enforced.
15991 (vectorizable_store, vectorizable_load): Likewise.
15993 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15994 Alan Hayward <alan.hayward@arm.com>
15995 David Sherwood <david.sherwood@arm.com>
15997 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
15998 of units as polynomial. Punt if we can't tell at compile time
15999 which vector contains the final result.
16001 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16002 Alan Hayward <alan.hayward@arm.com>
16003 David Sherwood <david.sherwood@arm.com>
16005 * tree-vect-loop.c (vectorizable_induction): Treat the number
16006 of units as polynomial. Punt on SLP inductions. Use an integer
16007 VEC_SERIES_EXPR for variable-length integer reductions. Use a
16008 cast of such a series for variable-length floating-point
16011 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16012 Alan Hayward <alan.hayward@arm.com>
16013 David Sherwood <david.sherwood@arm.com>
16015 * tree.h (build_index_vector): Declare.
16016 * tree.c (build_index_vector): New function.
16017 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
16018 of units as polynomial, forcibly converting it to a constant if
16019 vectorizable_reduction has already enforced the condition.
16020 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
16021 to create a {1,2,3,...} vector.
16022 (vectorizable_reduction): Treat the number of units as polynomial.
16023 Choose vectype_in based on the largest scalar element size rather
16024 than the smallest number of units. Enforce the restrictions
16027 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16028 Alan Hayward <alan.hayward@arm.com>
16029 David Sherwood <david.sherwood@arm.com>
16031 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
16032 number of units as polynomial.
16034 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16035 Alan Hayward <alan.hayward@arm.com>
16036 David Sherwood <david.sherwood@arm.com>
16038 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
16039 * target.def (autovectorize_vector_sizes): Return the vector sizes
16040 by pointer, using vector_sizes rather than a bitmask.
16041 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
16042 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
16043 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
16045 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
16046 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
16047 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
16048 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
16049 * omp-general.c (omp_max_vf): Likewise.
16050 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16051 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
16052 * tree-vect-loop.c (vect_analyze_loop): Likewise.
16053 * tree-vect-slp.c (vect_slp_bb): Likewise.
16054 * doc/tm.texi: Regenerate.
16055 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
16057 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
16058 the vector size as a poly_uint64 rather than an unsigned int.
16059 (current_vector_size): Change from an unsigned int to a poly_uint64.
16060 (get_vectype_for_scalar_type): Update accordingly.
16061 * tree.h (build_truth_vector_type): Take the size and number of
16062 units as a poly_uint64 rather than an unsigned int.
16063 (build_vector_type): Add a temporary overload that takes
16064 the number of units as a poly_uint64 rather than an unsigned int.
16065 * tree.c (make_vector_type): Likewise.
16066 (build_truth_vector_type): Take the number of units as a poly_uint64
16067 rather than an unsigned int.
16069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16070 Alan Hayward <alan.hayward@arm.com>
16071 David Sherwood <david.sherwood@arm.com>
16073 * target.def (get_mask_mode): Take the number of units and length
16074 as poly_uint64s rather than unsigned ints.
16075 * targhooks.h (default_get_mask_mode): Update accordingly.
16076 * targhooks.c (default_get_mask_mode): Likewise.
16077 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16078 * doc/tm.texi: Regenerate.
16080 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16081 Alan Hayward <alan.hayward@arm.com>
16082 David Sherwood <david.sherwood@arm.com>
16084 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
16085 * omp-general.c (omp_max_vf): Likewise.
16086 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
16087 (expand_omp_simd): Handle polynomial safelen.
16088 * omp-low.c (omplow_simd_context): Add a default constructor.
16089 (omplow_simd_context::max_vf): Change from int to poly_uint64.
16090 (lower_rec_simd_input_clauses): Update accordingly.
16091 (lower_rec_input_clauses): Likewise.
16093 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16094 Alan Hayward <alan.hayward@arm.com>
16095 David Sherwood <david.sherwood@arm.com>
16097 * tree-vectorizer.h (vect_nunits_for_cost): New function.
16098 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
16099 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
16100 (vect_analyze_slp_cost): Likewise.
16101 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
16102 (vect_model_load_cost): Likewise.
16104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16105 Alan Hayward <alan.hayward@arm.com>
16106 David Sherwood <david.sherwood@arm.com>
16108 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
16109 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
16110 from an unsigned int * to a poly_uint64_pod *.
16111 (calculate_unrolling_factor): New function.
16112 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
16114 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16115 Alan Hayward <alan.hayward@arm.com>
16116 David Sherwood <david.sherwood@arm.com>
16118 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
16119 from an unsigned int to a poly_uint64.
16120 (_loop_vec_info::slp_unrolling_factor): Likewise.
16121 (_loop_vec_info::vectorization_factor): Change from an int
16123 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
16124 (vect_get_num_vectors): New function.
16125 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
16126 (vect_get_num_copies): Use vect_get_num_vectors.
16127 (vect_analyze_data_ref_dependences): Change max_vf from an int *
16128 to an unsigned int *.
16129 (vect_analyze_data_refs): Change min_vf from an int * to a
16131 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
16132 than an unsigned HOST_WIDE_INT.
16133 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
16134 (vect_analyze_data_ref_dependence): Change max_vf from an int *
16135 to an unsigned int *.
16136 (vect_analyze_data_ref_dependences): Likewise.
16137 (vect_compute_data_ref_alignment): Handle polynomial vf.
16138 (vect_enhance_data_refs_alignment): Likewise.
16139 (vect_prune_runtime_alias_test_list): Likewise.
16140 (vect_shift_permute_load_chain): Likewise.
16141 (vect_supportable_dr_alignment): Likewise.
16142 (dependence_distance_ge_vf): Take the vectorization factor as a
16143 poly_uint64 rather than an unsigned HOST_WIDE_INT.
16144 (vect_analyze_data_refs): Change min_vf from an int * to a
16146 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
16147 vfm1 as a poly_uint64 rather than an int. Make the same change
16148 for the returned bound_scalar.
16149 (vect_gen_vector_loop_niters): Handle polynomial vf.
16150 (vect_do_peeling): Likewise. Update call to
16151 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
16152 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
16154 * tree-vect-loop.c (vect_determine_vectorization_factor)
16155 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
16156 (vect_get_known_peeling_cost): Likewise.
16157 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
16158 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
16159 (vect_transform_loop): Likewise. Use the lowest possible VF when
16160 updating the upper bounds of the loop.
16161 (vect_min_worthwhile_factor): Make static. Return an unsigned int
16162 rather than an int.
16163 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
16164 polynomial unroll factors.
16165 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
16166 (vect_make_slp_decision): Likewise.
16167 (vect_supported_load_permutation_p): Likewise, and polynomial
16169 (vect_analyze_slp_cost): Handle polynomial vf.
16170 (vect_slp_analyze_node_operations): Likewise.
16171 (vect_slp_analyze_bb_1): Likewise.
16172 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
16173 than an unsigned HOST_WIDE_INT.
16174 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
16175 (vectorizable_load): Handle polynomial vf.
16176 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
16178 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
16180 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16181 Alan Hayward <alan.hayward@arm.com>
16182 David Sherwood <david.sherwood@arm.com>
16184 * match.pd: Handle bit operations involving three constants
16185 and try to fold one pair.
16187 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16189 * tree-vect-loop-manip.c: Include gimple-fold.h.
16190 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
16191 niters_maybe_zero parameters. Handle other cases besides a step of 1.
16192 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
16193 Add a path that uses a step of VF instead of 1, but disable it
16195 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
16196 and niters_no_overflow parameters. Update calls to
16197 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
16198 Create a new SSA name if the latter choses to use a ste other
16199 than zero, and return it via niters_vector_mult_vf_var.
16200 * tree-vect-loop.c (vect_transform_loop): Update calls to
16201 vect_do_peeling, vect_gen_vector_loop_niters and
16202 slpeel_make_loop_iterate_ntimes.
16203 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
16204 (vect_gen_vector_loop_niters): Update declarations after above changes.
16206 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
16208 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
16209 128-bit round to integer instructions.
16210 (ceil<mode>2): Likewise.
16211 (btrunc<mode>2): Likewise.
16212 (round<mode>2): Likewise.
16214 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16216 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
16217 unaligned VSX load/store on P8/P9.
16218 (expand_block_clear): Allow the use of unaligned VSX
16219 load/store on P8/P9.
16221 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16223 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
16225 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
16226 swap associated with both a load and a store.
16228 2018-01-02 Andrew Waterman <andrew@sifive.com>
16230 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
16231 * config/riscv/riscv.md (clear_cache): Use it.
16233 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
16235 * web.c: Remove out-of-date comment.
16237 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16239 * expr.c (fixup_args_size_notes): Check that any existing
16240 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
16241 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
16242 (emit_single_push_insn): ...here.
16244 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16246 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
16247 (const_vector_encoded_nelts): New function.
16248 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
16249 (const_vector_int_elt, const_vector_elt): Declare.
16250 * emit-rtl.c (const_vector_int_elt_1): New function.
16251 (const_vector_elt): Likewise.
16252 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
16253 of CONST_VECTOR_ELT.
16255 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16257 * expr.c: Include rtx-vector-builder.h.
16258 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
16259 directly on the tree encoding.
16260 (const_vector_from_tree): Likewise.
16261 * optabs.c: Include rtx-vector-builder.h.
16262 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
16263 sequence of "u" values.
16264 * vec-perm-indices.c: Include rtx-vector-builder.h.
16265 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
16266 directly on the vec_perm_indices encoding.
16268 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16270 * doc/rtl.texi (const_vector): Describe new encoding scheme.
16271 * Makefile.in (OBJS): Add rtx-vector-builder.o.
16272 * rtx-vector-builder.h: New file.
16273 * rtx-vector-builder.c: Likewise.
16274 * rtl.h (rtx_def::u2): Add a const_vector field.
16275 (CONST_VECTOR_NPATTERNS): New macro.
16276 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
16277 (CONST_VECTOR_DUPLICATE_P): Likewise.
16278 (CONST_VECTOR_STEPPED_P): Likewise.
16279 (CONST_VECTOR_ENCODED_ELT): Likewise.
16280 (const_vec_duplicate_p): Check for a duplicated vector encoding.
16281 (unwrap_const_vec_duplicate): Likewise.
16282 (const_vec_series_p): Check for a non-duplicated vector encoding.
16283 Say that the function only returns true for integer vectors.
16284 * emit-rtl.c: Include rtx-vector-builder.h.
16285 (gen_const_vec_duplicate_1): Delete.
16286 (gen_const_vector): Call gen_const_vec_duplicate instead of
16287 gen_const_vec_duplicate_1.
16288 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
16289 (gen_const_vec_duplicate): Use rtx_vector_builder.
16290 (gen_const_vec_series): Likewise.
16291 (gen_rtx_CONST_VECTOR): Likewise.
16292 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
16293 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
16294 Build a new vector rather than modifying a CONST_VECTOR in-place.
16295 (handle_special_swappables): Update call accordingly.
16296 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
16297 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
16298 Build a new vector rather than modifying a CONST_VECTOR in-place.
16299 (handle_special_swappables): Update call accordingly.
16301 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16303 * simplify-rtx.c (simplify_const_binary_operation): Use
16304 CONST_VECTOR_ELT instead of XVECEXP.
16306 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16308 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
16309 the selector elements to be different from the data elements
16310 if the selector is a VECTOR_CST.
16311 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
16312 ssizetype for the selector.
16314 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16316 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
16317 before testing each element individually.
16318 * tree-vect-generic.c (lower_vec_perm): Likewise.
16320 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16322 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
16323 * selftest-run-tests.c (selftest::run_tests): Call it.
16324 * vector-builder.h (vector_builder::operator ==): New function.
16325 (vector_builder::operator !=): Likewise.
16326 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
16327 (vec_perm_indices::all_from_input_p): New function.
16328 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
16329 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
16330 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
16331 instead of reading the VECTOR_CST directly. Detect whether both
16332 vector inputs are the same before constructing the vec_perm_indices,
16333 and update the number of inputs argument accordingly. Use the
16334 utility functions added above. Only construct sel2 if we need to.
16336 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16338 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
16339 the broadcast of the low byte.
16340 (expand_mult_highpart): Use an explicit encoding for the permutes.
16341 * optabs-query.c (can_mult_highpart_p): Likewise.
16342 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16343 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
16344 (vectorizable_bswap): Likewise.
16345 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
16346 explicit encoding for the power-of-2 permutes.
16347 (vect_permute_store_chain): Likewise.
16348 (vect_grouped_load_supported): Likewise.
16349 (vect_permute_load_chain): Likewise.
16351 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16353 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
16354 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
16355 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
16356 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16357 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
16358 (vect_gen_perm_mask_any): Likewise.
16360 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16362 * int-vector-builder.h: New file.
16363 * vec-perm-indices.h: Include int-vector-builder.h.
16364 (vec_perm_indices): Redefine as an int_vector_builder.
16365 (auto_vec_perm_indices): Delete.
16366 (vec_perm_builder): Redefine as a stand-alone class.
16367 (vec_perm_indices::vec_perm_indices): New function.
16368 (vec_perm_indices::clamp): Likewise.
16369 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
16370 (vec_perm_indices::new_vector): New function.
16371 (vec_perm_indices::new_expanded_vector): Update for new
16372 vec_perm_indices class.
16373 (vec_perm_indices::rotate_inputs): New function.
16374 (vec_perm_indices::all_in_range_p): Operate directly on the
16375 encoded form, without computing elided elements.
16376 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
16377 encoding. Update for new vec_perm_indices class.
16378 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
16379 the given vec_perm_builder.
16380 (expand_vec_perm_var): Update vec_perm_builder constructor.
16381 (expand_mult_highpart): Use vec_perm_builder instead of
16382 auto_vec_perm_indices.
16383 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
16384 vec_perm_indices instead of auto_vec_perm_indices. Use a single
16385 or double series encoding as appropriate.
16386 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
16387 vec_perm_indices instead of auto_vec_perm_indices.
16388 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
16389 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
16390 (vect_permute_store_chain): Likewise.
16391 (vect_grouped_load_supported): Likewise.
16392 (vect_permute_load_chain): Likewise.
16393 (vect_shift_permute_load_chain): Likewise.
16394 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
16395 (vect_transform_slp_perm_load): Likewise.
16396 (vect_schedule_slp_instance): Likewise.
16397 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
16398 (vectorizable_mask_load_store): Likewise.
16399 (vectorizable_bswap): Likewise.
16400 (vectorizable_store): Likewise.
16401 (vectorizable_load): Likewise.
16402 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
16403 vec_perm_indices instead of auto_vec_perm_indices. Use
16404 tree_to_vec_perm_builder to read the vector from a tree.
16405 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
16406 vec_perm_builder instead of a vec_perm_indices.
16407 (have_whole_vector_shift): Use vec_perm_builder and
16408 vec_perm_indices instead of auto_vec_perm_indices. Leave the
16409 truncation to calc_vec_perm_mask_for_shift.
16410 (vect_create_epilog_for_reduction): Likewise.
16411 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
16412 from auto_vec_perm_indices to vec_perm_indices.
16413 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
16414 instead of changing individual elements.
16415 (aarch64_vectorize_vec_perm_const): Use new_vector to install
16416 the vector in d.perm.
16417 * config/arm/arm.c (expand_vec_perm_d::perm): Change
16418 from auto_vec_perm_indices to vec_perm_indices.
16419 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
16420 instead of changing individual elements.
16421 (arm_vectorize_vec_perm_const): Use new_vector to install
16422 the vector in d.perm.
16423 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
16424 Update vec_perm_builder constructor.
16425 (rs6000_expand_interleave): Likewise.
16426 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
16427 (rs6000_expand_interleave): Likewise.
16429 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16431 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
16432 to qimode could truncate the indices.
16433 * optabs.c (expand_vec_perm_var): Likewise.
16435 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16437 * Makefile.in (OBJS): Add vec-perm-indices.o.
16438 * vec-perm-indices.h: New file.
16439 * vec-perm-indices.c: Likewise.
16440 * target.h (vec_perm_indices): Replace with a forward class
16442 (auto_vec_perm_indices): Move to vec-perm-indices.h.
16443 * optabs.h: Include vec-perm-indices.h.
16444 (expand_vec_perm): Delete.
16445 (selector_fits_mode_p, expand_vec_perm_var): Declare.
16446 (expand_vec_perm_const): Declare.
16447 * target.def (vec_perm_const_ok): Replace with...
16448 (vec_perm_const): ...this new hook.
16449 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
16450 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
16451 * doc/tm.texi: Regenerate.
16452 * optabs.def (vec_perm_const): Delete.
16453 * doc/md.texi (vec_perm_const): Likewise.
16454 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
16455 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
16456 expand_vec_perm for constant permutation vectors. Assert that
16457 the mode of variable permutation vectors is the integer equivalent
16458 of the mode that is being permuted.
16459 * optabs-query.h (selector_fits_mode_p): Declare.
16460 * optabs-query.c: Include vec-perm-indices.h.
16461 (selector_fits_mode_p): New function.
16462 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
16463 is defined, instead of checking whether the vec_perm_const_optab
16464 exists. Use targetm.vectorize.vec_perm_const instead of
16465 targetm.vectorize.vec_perm_const_ok. Check whether the indices
16466 fit in the vector mode before using a variable permute.
16467 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
16468 vec_perm_indices instead of an rtx.
16469 (expand_vec_perm): Replace with...
16470 (expand_vec_perm_const): ...this new function. Take the selector
16471 as a vec_perm_indices rather than an rtx. Also take the mode of
16472 the selector. Update call to shift_amt_for_vec_perm_mask.
16473 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
16474 Use vec_perm_indices::new_expanded_vector to expand the original
16475 selector into bytes. Check whether the indices fit in the vector
16476 mode before using a variable permute.
16477 (expand_vec_perm_var): Make global.
16478 (expand_mult_highpart): Use expand_vec_perm_const.
16479 * fold-const.c: Includes vec-perm-indices.h.
16480 * tree-ssa-forwprop.c: Likewise.
16481 * tree-vect-data-refs.c: Likewise.
16482 * tree-vect-generic.c: Likewise.
16483 * tree-vect-loop.c: Likewise.
16484 * tree-vect-slp.c: Likewise.
16485 * tree-vect-stmts.c: Likewise.
16486 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
16488 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
16489 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
16490 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
16491 (aarch64_vectorize_vec_perm_const): ...this new function.
16492 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
16493 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
16494 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
16495 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
16496 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
16497 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
16498 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
16500 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
16501 check for NEON modes.
16502 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
16503 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
16504 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
16505 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
16507 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
16508 the old VEC_PERM_CONST conditions.
16509 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
16510 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
16511 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
16512 (ia64_vectorize_vec_perm_const_ok): Merge into...
16513 (ia64_vectorize_vec_perm_const): ...this new function.
16514 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
16515 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
16516 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
16517 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
16518 * config/mips/mips.c (mips_expand_vec_perm_const)
16519 (mips_vectorize_vec_perm_const_ok): Merge into...
16520 (mips_vectorize_vec_perm_const): ...this new function.
16521 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
16522 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
16523 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
16524 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
16525 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
16526 (rs6000_expand_vec_perm_const): Delete.
16527 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
16529 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
16530 (altivec_expand_vec_perm_const_le): Take each operand individually.
16531 Operate on constant selectors rather than rtxes.
16532 (altivec_expand_vec_perm_const): Likewise. Update call to
16533 altivec_expand_vec_perm_const_le.
16534 (rs6000_expand_vec_perm_const): Delete.
16535 (rs6000_vectorize_vec_perm_const_ok): Delete.
16536 (rs6000_vectorize_vec_perm_const): New function.
16537 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
16538 an element count and rtx array.
16539 (rs6000_expand_extract_even): Update call accordingly.
16540 (rs6000_expand_interleave): Likewise.
16541 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
16542 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
16543 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
16544 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
16545 (rs6000_expand_vec_perm_const): Delete.
16546 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
16547 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
16548 (altivec_expand_vec_perm_const_le): Take each operand individually.
16549 Operate on constant selectors rather than rtxes.
16550 (altivec_expand_vec_perm_const): Likewise. Update call to
16551 altivec_expand_vec_perm_const_le.
16552 (rs6000_expand_vec_perm_const): Delete.
16553 (rs6000_vectorize_vec_perm_const_ok): Delete.
16554 (rs6000_vectorize_vec_perm_const): New function. Remove stray
16555 reference to the SPE evmerge intructions.
16556 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
16557 an element count and rtx array.
16558 (rs6000_expand_extract_even): Update call accordingly.
16559 (rs6000_expand_interleave): Likewise.
16560 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
16561 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
16563 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
16565 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16567 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
16568 vector mode and that that mode matches the mode of the data
16570 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
16571 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
16572 directly using expand_vec_perm_1 when forcing selectors into
16574 (expand_vec_perm_var): New function, split out from expand_vec_perm.
16576 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16578 * optabs-query.h (can_vec_perm_p): Delete.
16579 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
16580 * optabs-query.c (can_vec_perm_p): Split into...
16581 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
16582 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
16583 particular selector is valid.
16584 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
16585 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
16586 (vect_grouped_load_supported): Likewise.
16587 (vect_shift_permute_load_chain): Likewise.
16588 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
16589 (vect_transform_slp_perm_load): Likewise.
16590 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
16591 (vectorizable_bswap): Likewise.
16592 (vect_gen_perm_mask_checked): Likewise.
16593 * fold-const.c (fold_ternary_loc): Likewise. Don't take
16594 implementations of variable permutation vectors into account
16595 when deciding which selector to use.
16596 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
16597 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
16598 with a false third argument.
16599 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
16600 to test whether the constant selector is valid and can_vec_perm_var_p
16601 to test whether a variable selector is valid.
16603 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16605 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
16606 * optabs-query.c (can_vec_perm_p): Likewise.
16607 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
16608 instead of vec_perm_indices.
16609 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
16610 (vect_gen_perm_mask_checked): Likewise,
16611 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
16612 (vect_gen_perm_mask_checked): Likewise,
16614 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
16616 * optabs-query.h (qimode_for_vec_perm): Declare.
16617 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
16618 (qimode_for_vec_perm): ...this new function.
16619 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
16621 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16623 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
16624 does not have a conditional at the top.
16626 2018-01-02 Richard Biener <rguenther@suse.de>
16628 * ipa-inline.c (big_speedup_p): Fix expression.
16630 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
16633 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
16636 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
16640 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
16641 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
16642 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
16643 cond_taken_branch_cost 3->4.
16645 2018-01-01 Jakub Jelinek <jakub@redhat.com>
16647 PR tree-optimization/83581
16648 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
16649 TODO_cleanup_cfg if any changes have been made.
16651 PR middle-end/83608
16652 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
16653 convert_modes if target mode has the right side, but different mode
16656 PR middle-end/83609
16657 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
16658 last argument when extracting from CONCAT. If either from_real or
16659 from_imag is NULL, use expansion through memory. If result is not
16660 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
16661 the parts directly to inner mode, if even that fails, use expansion
16664 PR middle-end/83623
16665 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
16666 check for bswap in mode rather than HImode and use that in expand_unop
16669 Copyright (C) 2018 Free Software Foundation, Inc.
16671 Copying and distribution of this file, with or without modification,
16672 are permitted in any medium without royalty provided the copyright
16673 notice and this notice are preserved.