1 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
3 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
5 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
8 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
11 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
12 aarch64_evpc_tbl guard.
14 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
16 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
19 2018-08-22 Martin Sebor <msebor@redhat.com>
22 * tree-pretty-print.c (pretty_print_string): Add argument.
23 (dump_generic_node): Call to pretty_print_string with string size.
25 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
27 PR rtl-optimization/86771
28 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
29 of two SETs into those two SETs, one to be placed at i2, if that SETs
30 destination is modified between i2 and i3.
32 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
34 PR tree-optimization/86725
35 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
37 (vect_analyze_scalar_cycles_1): Check it.
39 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
41 PR tree-optimization/86725
42 * tree-vect-loop.c (vect_is_simple_reduction): When treating
43 an outer loop phi as a double reduction, make sure that the
44 single user of the phi result is an inner loop phi.
46 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
48 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
49 grouped stores with gaps to a strided group.
51 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
53 * tree-vect-stmts.c (get_group_load_store_type)
54 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
55 first statement in a group.
57 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
59 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
60 the sequence used in gcc/gcc.c.
62 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
65 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
68 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
70 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
71 Darwin10-specific unwinder-shim.
72 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
73 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
74 New to cater for Darwin10 Rosetta.
76 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
78 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
79 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
81 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
86 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
87 (dwarf2out_switch_text_section): Generate a local label for the
88 second function sub-section and apply it as the second FDE start
90 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
91 second sub-section start.
93 2018-08-22 Richard Biener <rguenther@suse.de>
95 PR tree-optimization/86988
96 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
98 2018-08-22 Richard Biener <rguenther@suse.de>
100 PR tree-optimization/86945
101 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
103 2018-08-22 Alexandre Oliva <oliva@adacore.com>
105 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
106 a comment about how uses of r2 for .sdata2 come about.
108 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
110 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
112 2018-08-21 Marek Polacek <polacek@redhat.com>
114 PR c++/86981, Implement -Wpessimizing-move.
115 * doc/invoke.texi: Document -Wpessimizing-move.
117 2018-08-21 Jan Hubicka <jh@suse.cz>
119 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
120 * tree.h (is_redundant_typedef): Remove.
121 * dwarf2out.c (is_redundant_typedef): Turn into static function.
123 2018-08-21 Jan Hubicka <jh@suse.cz>
125 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
128 2018-08-21 Tamar Christina <tamar.christina@arm.com>
130 * expmed.c (extract_low_bits): Reject invalid subregs early.
132 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
135 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
138 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
140 * config/vxworks.h: Guard vxworks_asm_out_constructor and
141 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
142 * config/vxworks.c: Likewise.
144 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
146 * config/vxworks.c: Set targetm.have_ctors_dtors
147 if HAVE_INITFINI_ARRAY_SUPPORT.
148 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
149 if HAVE_INITFINI_ARRAY_SUPPORT.
151 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
153 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
154 default search path for VxWorks < 7.
156 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
158 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
159 (get_format_string): Refer to c_getstr.
161 2018-08-21 Tom de Vries <tdevries@suse.de>
163 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
164 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
165 (debuginfo_early_stop): Declare.
166 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
167 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
168 (debuginfo_early_stop): New function.
169 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
170 and debuginfo_early_stop.
171 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
172 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
173 (general_init): Call debuginfo_early_init.
174 (finalize): Call debuginfo_fini.
175 (do_compile): Call debuginfo_init.
176 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
178 (@item -fdump-debug, @item -fdump-earlydebug): Add.
180 2018-08-21 Tom de Vries <tdevries@suse.de>
182 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
183 flag_dump_noaddr and flag_dump_unnumbered.
185 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
187 * wide-int-range.cc (wide_int_range_abs): New.
188 (wide_int_range_order_set): Rename from wide_int_range_min_max.
189 * wide-int-range.h (wide_int_range_abs): New.
190 (wide_int_range_min_max): New.
191 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
192 case to call wide_int_range_abs.
193 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
194 (extract_range_from_abs_expr): Delete.
196 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
199 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
200 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
203 2018-08-20 Nathan Sidwell <nathan@acm.org>
204 Jeff Law <law@redhat.com>
206 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
207 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
209 2018-08-20 David Malcolm <dmalcolm@redhat.com>
212 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
213 (decl_attributes): Likewise.
214 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
216 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
217 * diagnostic-core.h (class auto_diagnostic_group): New class.
218 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
219 (diagnostic_report_diagnostic): Handle the first diagnostics within
221 (emit_diagnostic): Add auto_diagnostic_group instance.
223 (inform_n): Likewise.
225 (warning_at): Likewise.
226 (warning_n): Likewise.
228 (permerror): Likewise.
231 (error_at): Likewise.
233 (fatal_error): Likewise.
234 (internal_error): Likewise.
235 (internal_error_no_backtrace): Likewise.
236 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
237 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
238 * diagnostic.h (struct diagnostic_context): Add fields
239 "diagnostic_group_nesting_depth",
240 "diagnostic_group_emission_count", "begin_group_cb",
242 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
243 Add auto_diagnostic_group instance(s).
244 (find_explicit_erroneous_behavior): Likewise.
245 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
246 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
247 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
248 (gimplify_va_arg_expr): Likewise.
249 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
250 (HSA_SORRY_AT): Likewise.
251 * ipa-devirt.c (compare_virtual_tables): Likewise.
252 (warn_odr): Likewise.
253 * multiple_target.c (expand_target_clones): Likewise.
254 * opts-common.c (cmdline_handle_error): Likewise.
255 * reginfo.c (globalize_reg): Likewise.
256 * substring-locations.c (format_warning_n_va): Likewise.
257 * tree-inline.c (expand_call_inline): Likewise.
258 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
259 * tree-ssa-loop-niter.c
260 (do_warn_aggressive_loop_optimizations): Likewise.
261 * tree-ssa-uninit.c (warn_uninit): Likewise.
262 * tree.c (warn_deprecated_use): Likewise.
264 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
267 * config/i386/i386.md (eh_return): Always update EH return
268 address in word_mode.
270 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
272 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
273 TARGET_SPLIT_COMPLEX_ARG is defined.
275 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
277 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
279 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
282 * expr.c (expand_assignment): Assert that bitpos is positive.
283 (store_field): Likewise
284 (expand_expr_real_1): Make sure that bitpos is positive.
285 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
288 2018-08-20 Nathan Sidwell <nathan@acm.org>
290 * Makefile.in (CPP_ID_DATA_H): Delete.
291 (CPP_INTERNAL_H): Don't add it.
292 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
293 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
295 2018-08-20 Richard Biener <rguenther@suse.de>
297 PR tree-optimization/78655
298 * tree-vrp.c (extract_range_from_binary_expr_1): Make
299 pointer + offset nonnull if either operand is nonnull work.
301 2018-08-20 Tom de Vries <tdevries@suse.de>
303 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
304 unless the referenced die describes the added property using
305 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
306 Otherwise, add a DW_AT_location to the referenced die.
308 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
311 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
312 register_operand when calling ix86_set_reg_reg_cost.
313 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
314 Set *total to 0 for operands that satisfy x86_64_immediate_operand
315 predicate and to 1 otherwise.
317 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
319 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
320 emit a diagnostic that it is not supported and reset the option.
321 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
322 supported and consume the option. (ASM_FINAL_SPEC): New.
324 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
326 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
329 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
331 C-SKY port: Documentation
333 * doc/extend.texi (C-SKY Function Attributes): New section.
334 * doc/invoke.texi (Option Summary): Add C-SKY options.
335 (C-SKY Options): New section.
336 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
338 2018-08-17 Jojo <jijie_rong@c-sky.com>
339 Huibin Wang <huibin_wang@c-sky.com>
340 Sandra Loosemore <sandra@codesourcery.com>
341 Chung-Lin Tang <cltang@codesourcery.com>
343 C-SKY port: Backend implementation
345 * config/csky/*: New.
346 * common/config/csky/*: New.
348 2018-08-17 Jojo <jijie_rong@c-sky.com>
349 Huibin Wang <huibin_wang@c-sky.com>
350 Sandra Loosemore <sandra@codesourcery.com>
351 Chung-Lin Tang <cltang@codesourcery.com>
352 Andrew Jenner <andrew@codesourcery.com>
354 C-SKY port: Configury
356 * config.gcc (csky-*-*): New.
357 * configure.ac: Add csky to targets for dwarf2 debug_line support.
358 * configure: Regenerated.
360 2018-08-17 David Malcolm <dmalcolm@redhat.com>
362 * dump-context.h: Include "dumpfile.h".
363 (dump_context::dump_printf_va): Convert final param from va_list
364 to va_list *. Convert from ATTRIBUTE_PRINTF to
365 ATTRIBUTE_GCC_DUMP_PRINTF.
366 (dump_context::dump_printf_loc_va): Likewise.
367 * dumpfile.c: Include "stringpool.h".
368 (make_item_for_dump_printf_va): Delete.
369 (make_item_for_dump_printf): Delete.
370 (class dump_pretty_printer): New class.
371 (dump_pretty_printer::dump_pretty_printer): New ctor.
372 (dump_pretty_printer::emit_items): New member function.
373 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
375 (dump_pretty_printer::emit_item): New member function.
376 (dump_pretty_printer::stash_item): New member function.
377 (dump_pretty_printer::format_decoder_cb): New member function.
378 (dump_pretty_printer::decode_format): New member function.
379 (dump_context::dump_printf_va): Reimplement in terms of
381 (dump_context::dump_printf_loc_va): Convert final param from va_list
383 (dump_context::begin_scope): Reimplement call to
384 make_item_for_dump_printf.
385 (dump_printf): Update for change to dump_printf_va.
386 (dump_printf_loc): Likewise.
387 (selftest::test_capture_of_dump_calls): Convert "stmt" from
388 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
390 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
391 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
392 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
393 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
394 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
395 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
396 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
397 within a dump_printf_loc call to "%wu".
398 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
399 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
400 missing space after "=".
401 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
402 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
403 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
404 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
405 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
406 duplicate "vectorized" from message.
408 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
410 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
411 polyNxK_t element's TYPE_STRING_FLAG.
413 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
415 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
416 (they were unnamed before). Fix comments.
418 2018-08-17 Nathan Sidwell <nathan@acm.org>
420 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
422 2018-08-17 Richard Biener <rguenther@suse.de>
424 PR tree-optimization/86841
425 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
427 2018-08-17 Martin Liska <mliska@suse.cz>
429 * common.opt: Remove Warn, Init and Report for options with
430 Ignore/Deprecated flag. Warning is done automatically for
432 * config/i386/i386.opt: Likewise.
433 * config/ia64/ia64.opt: Likewise.
434 * config/rs6000/rs6000.opt: Likewise.
435 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
436 Remove usage of flag_check_pointer_bounds.
437 * lto-wrapper.c (merge_and_complain): Do not handle
438 OPT_fcheck_pointer_bounds.
439 (append_compiler_options): Likewise.
440 * opt-functions.awk: Do not handle Deprecated.
441 * optc-gen.awk: Check that Var, Report and Init are not
442 used for an option with Ignore/Deprecated flag.
443 * opts-common.c (decode_cmdline_option): Do not report
445 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
447 * opts.h (struct cl_option): Remove cl_deprecated flag.
448 (CL_ERR_DEPRECATED): Remove error enum value.
450 2018-08-17 Richard Biener <rguenther@suse.de>
453 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
454 across a va-arg-pack using call adjust its return value accordingly.
456 2018-08-16 Martin Sebor <msebor@redhat.com>
458 PR tree-optimization/86853
459 * gimple-ssa-sprintf.c (struct format_result): Rename member.
460 (struct fmtresult): Add member and initialize it in ctors.
461 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
462 (format_string): Handle %S the same as %ls. Set MAYFAIL.
463 (format_directive): Set POSUNDER4K when MAYFAIL is set.
464 (parse_directive): Handle %C same as %c.
465 (sprintf_dom_walker::compute_format_length): Adjust.
466 (is_call_safe): Adjust.
468 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
470 * builtins.c (c_strlen): Add new parameter eltsize. Use it
471 for determining how to count the elements.
472 * builtins.h (c_strlen): Adjust prototype.
473 * expr.c (string_constant): Add new parameter mem_size.
474 Set *mem_size appropriately.
475 * expr.h (string_constant): Adjust protoype.
476 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
477 * gimple-fold.h (get_range_strlen): Adjust prototype.
478 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
479 (format_string): Call get_string_length with eltsize.
481 2018-08-16 David Malcolm <dmalcolm@redhat.com>
483 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
484 to emit the span, rather than setting it as the prefix.
486 2018-08-16 David Malcolm <dmalcolm@redhat.com>
488 * diagnostic-show-locus.c (layout::start_annotation_line): Add
489 "margin_char" parameter, defaulting to space. Use it in place
490 of pp_space for the initial part of the margin.
491 (layout::print_leading_fixits): Use '+' when filling the margin
492 of line-insertion fix-it hints.
494 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
496 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
499 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
501 * config/rs6000/altivec.md: Don't set length attribute to the default
503 * config/rs6000/darwin.md: Ditto.
504 * config/rs6000/dfp.md: Ditto.
505 * config/rs6000/htm.md: Ditto.
506 * config/rs6000/rs6000.md: Ditto.
507 * config/rs6000/sync.md: Ditto.
508 * config/rs6000/vsx.md: Ditto.
510 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
512 * config/rs6000/altivec.md: Don't set length attribute to the default
513 value, for branch instructions.
514 * config/rs6000/darwin.md: Ditto.
515 * config/rs6000/rs6000.md: Ditto.
517 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
519 * config/rs6000/rs6000.md (length): Always define as const_int 4.
520 (unnamed conditional branch define_insn): Set length to 4 or 8
522 (<bd>_<mode>): Similar, for alternative 0.
523 (<bd>tf_<mode>): Ditto.
525 2018-08-16 Tamar Christina <tamar.christina@arm.com>
527 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
529 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
531 * doc/rtl.texi: Replace old RTX class names with new names.
534 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
536 * expmed.h (canonicalize_comparison): New declaration.
537 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
538 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
539 * optabs.c (prepare_cmp_insn): Likewise.
540 * rtl.h (unsigned_condition_p): New function which checks if a
541 comparison operator is unsigned.
543 2018-08-16 Nathan Sidwell <nathan@acm.org>
545 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
546 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
548 2018-08-16 Tamar Christina <tamar.christina@arm.com>
551 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
552 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
553 (mov<mov>): ..this and enable unconditionally.
555 2018-08-16 Tamar Christina <tamar.christina@arm.com>
557 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
559 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
561 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
562 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
563 "Common" with "Target".
565 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
567 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
568 * doc/invoke.texi (mmitigate-rop): Remove.
569 * config/i386/i386.c: Do not include "regrename.h".
570 (ix86_rop_should_change_byte_p, reg_encoded_number)
571 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
573 (ix86_reorg): Remove call to ix86_mitigate_rop.
574 * config/i386/i386.md (attr "modrm_class"): Remove.
575 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
576 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
577 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
579 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
581 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
582 allow folding of mergeh() and mergel() for the float and double types.
583 (fold_mergehl_helper): Rework to handle building a permute tree
586 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
588 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
591 2018-08-15 David Malcolm <dmalcolm@redhat.com>
593 * common.opt (fdiagnostics-show-labels): New option.
594 * diagnostic-show-locus.c (class layout_range): Add field
596 (class layout): Add field "m_show_labels_p".
597 (layout_range::layout_range): Add param "label" and use it to
599 (make_range): Pass in NULL for new "label" param of layout_range's
601 (layout::layout): Initialize m_show_labels_p.
602 (layout::maybe_add_location_range): Pass in loc_range->m_label
603 when constructing layout_range instances.
604 (struct line_label): New struct.
605 (layout::print_any_labels): New member function.
606 (layout::print_line): Call it if label-printing is enabled.
607 (selftest::test_one_liner_labels): New test.
608 (selftest::test_diagnostic_show_locus_one_liner): Call it.
609 * diagnostic.c (diagnostic_initialize): Initialize
610 context->show_labels_p.
611 * diagnostic.h (struct diagnostic_context): Add field
613 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
614 -fno-diagnostics-show-labels.
615 * dwarf2out.c (gen_producer_string): Add
616 OPT_fdiagnostics_show_labels to the ignored options.
617 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
619 (gcc_rich_location::maybe_add_expr): Likewise.
620 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
621 label" param, defaulting to NULL.
622 (gcc_rich_location::add_expr): Add "label" param.
623 (gcc_rich_location::maybe_add_expr): Likewise.
624 (class text_range_label): New class.
625 (class range_label_for_type_mismatch): New class.
626 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
627 of format_warning_va.
628 (fmtwarn_n): Likewise for new params of format_warning_n_va.
629 * lto-wrapper.c (merge_and_complain): Add
630 OPT_fdiagnostics_show_labels to the "pick one setting" options.
631 (append_compiler_options): Likewise to the dropped options.
632 (append_diag_options): Likewise to the passed-on options.
633 * opts.c (common_handle_option): Handle the new option.
634 * selftest-diagnostic.c
635 (test_diagnostic_context::test_diagnostic_context): Enable
637 * substring-locations.c: Include "gcc-rich-location.h".
638 (format_warning_n_va): Add "fmt_label" and "param_label" params
639 and use them as appropriate.
640 (format_warning_va): Add "fmt_label" and "param_label" params,
641 passing them on to format_warning_n_va.
642 (format_warning_at_substring): Likewise.
643 (format_warning_at_substring_n): Likewise.
644 * substring-locations.h (format_warning_va): Add "fmt_label" and
645 "param_label" params.
646 (format_warning_n_va): Likewise.
647 (format_warning_at_substring): Likewise.
648 (format_warning_at_substring_n): Likewise.
649 * toplev.c (general_init): Initialize global_dc->show_labels_p.
651 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
654 * builtins.c (expand_builtin_memcmp): Do not expand the call
655 when overflow is detected.
657 2018-08-15 Martin Sebor <msebor@redhat.com>
659 PR tree-optimization/71625
660 * config/aarch64/aarch64-builtins.c
661 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
663 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
665 * config/s390/s390.c (s390_reorg): Remove loop.
667 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
670 (darwin_function_switched_text_sections): Delete.
671 * gcc/config/darwin.h
672 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
674 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
677 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
678 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
679 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
681 2018-08-15 Martin Liska <mliska@suse.cz>
683 PR tree-optimization/86925
684 * predict.c (expr_expected_value_1): When taking
685 later predictor, assign also probability.
686 Use fold_build2_initializer_loc in order to fold
687 the expression in -frounding-math.
689 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
691 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
693 (expand_vec_perm_1): Try the new method.
695 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
698 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
699 Check whether lra_live_max_point is 0 before dividing.
701 2018-08-14 Martin Sebor <msebor@redhat.com>
703 PR tree-optimization/86650
704 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
705 (vrp_prop::check_mem_ref): Same.
707 2018-08-13 Liu Hao <lh_mouse@126.com>
709 * pretty-print.c (eat_esc_sequence): Swap the foreground and
710 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
711 and clear it thereafter, as it only works for DBCS.
713 2018-08-13 Liu Hao <lh_mouse@126.com>
715 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
716 handle returned by _get_osf_handle().
718 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
720 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
721 for folding vec_perm.
723 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
725 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
726 Add support for gimple-folding of vec_pack() and vec_unpack()
729 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
731 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
732 vec_xst variants to the list.
733 (rs6000_gimple_fold_builtin): Add support for folding unaligned
734 vector loads and stores.
736 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
738 * config.gcc (rs6000-ibm-aix4.x): Delete.
739 (rs6000-ibm-aix5.1): Delete.
740 (rs6000-ibm-aix5.2): Delete.
741 (rs6000-ibm-aix5.3): Delete.
742 * config/rs6000/aix43.h: Delete.
743 * config/rs6000/aix51.h: Delete.
744 * config/rs6000/aix52.h: Delete.
745 * config/rs6000/t-aix43: Delete.
747 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
749 * config/s390/s390.c (s390_decompose_constant_pool_ref):
751 (s390_decompose_address): Factor out constant pool ref
754 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
756 * config/nds32/nds32-predicates.c
757 (nds32_can_use_bclr_p): Change return type as bool.
758 (nds32_can_use_bset_p): Ditto.
759 (nds32_can_use_btgl_p): Ditto.
760 (nds32_can_use_bitci_p): Ditto.
761 * config/nds32/nds32-protos.h
762 (nds32_can_use_bclr_p): Change declaration.
763 (nds32_can_use_bset_p): Ditto.
764 (nds32_can_use_btgl_p): Ditto.
765 (nds32_can_use_bitci_p): Ditto.
767 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
769 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
770 Support -msched-prolog-epilog option.
771 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
773 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
775 * common/config/nds32/nds32-common.c
776 (nds32_option_optimization_table): Enalbe -malways-align.
778 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
780 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
782 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
783 OPT_misr_secure_ case.
784 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
785 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
786 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
788 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
789 (nds32_isr_info): New field security_level.
790 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
791 * config/nds32/nds32.md (return_internal): Consider critical attribute.
792 * config/nds32/nds32.opt (misr-secure): New option.
793 * config/nds32/nds32_init.inc: New file.
794 * config/nds32/nds32_isr.h: New file.
796 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
798 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
799 Update comment for atomic instructions.
800 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
801 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
803 (atomic_loaddi): Revise fence expansion to only emit fence prior to
804 load for __ATOMIC_SEQ_CST model.
805 (atomic_loaddi_1): Remove float register target.
806 (atomic_storedi): Handle CONST_INT values.
807 (atomic_storedi_1): Remove float register source. Add special case
809 (memory_barrier): New expander and insn.
811 2018-08-11 Jakub Jelinek <jakub@redhat.com>
813 PR tree-optimization/86835
814 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
815 new_stmt after def_gsi, make sure to insert new_square_stmt after
816 that stmt, not 2 stmts before it.
818 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
821 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
824 2018-08-10 Martin Liska <mliska@suse.cz>
827 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
829 * builtins.c (expand_builtin_expect_with_probability):
831 (expand_builtin_expect_with_probability): New function.
832 (build_builtin_expect_predicate): Add new argumnet probability
833 for BUILT_IN_EXPECT_WITH_PROBABILITY.
834 (fold_builtin_expect):
837 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
838 * builtins.h (fold_builtin_expect): Set new argument.
839 * doc/extend.texi: Document __builtin_expect_with_probability.
840 * doc/invoke.texi: Likewise.
841 * gimple-fold.c (gimple_fold_call): Pass new argument.
842 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
843 also BUILT_IN_EXPECT_WITH_PROBABILITY.
844 * predict.c (get_predictor_value): New function.
845 (expr_expected_value): Add new argument probability. Assume
846 that predictor and probability are always non-null.
847 (expr_expected_value_1): Likewise. For __builtin_expect and
848 __builtin_expect_with_probability set probability. Handle
849 combination in binary expressions.
850 (tree_predict_by_opcode): Simplify code by simply calling
852 (pass_strip_predict_hints::execute): Add handling of
853 BUILT_IN_EXPECT_WITH_PROBABILITY.
854 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
856 * tree.h (DECL_BUILT_IN_P): New function.
858 2018-08-10 Martin Liska <mliska@suse.cz>
860 PR tree-optimization/85799
861 * passes.def: Add argument for pass_strip_predict_hints.
862 * predict.c (class pass_strip_predict_hints): Add new argument
864 (strip_predictor_early): New function.
865 (pass_strip_predict_hints::execute): Call the function to
867 (strip_predict_hints): New function.
868 * predict.def: Fix comment.
870 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
872 * Makefile.in: Clarify which tm.texi to copy over to assert the
873 right to grant a GFDL license for all.
875 2018-08-09 Jeff Law <law@redhat.com>
877 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
880 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
882 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
885 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
887 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
888 reductions for variable-length vectors.
890 2018-08-09 David Malcolm <dmalcolm@redhat.com>
893 * common.opt (fdiagnostics-show-line-numbers): New option.
894 * diagnostic-show-locus.c (class layout): Add fields
895 "m_show_line_numbers_p" and "m_linenum_width";
896 (num_digits): New function.
897 (test_num_digits): New function.
898 (layout::layout): Initialize new fields. Update m_x_offset
899 logic to handle any left margin.
900 (layout::print_source_line): Print line number when requested.
901 (layout::start_annotation_line): New member function.
902 (layout::print_annotation_line): Call it.
903 (layout::print_leading_fixits): Likewise.
904 (layout::print_trailing_fixits): Likewise. Update calls to
905 move_to_column for new parameter.
906 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
907 it to potentially call start_annotation_line.
908 (layout::show_ruler): Call start_annotation_line.
909 (selftest::test_line_numbers_multiline_range): New selftest.
910 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
911 and selftest::test_line_numbers_multiline_range.
912 * diagnostic.c (diagnostic_initialize): Initialize
914 * diagnostic.h (struct diagnostic_context): Add field
915 "show_line_numbers_p".
916 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
917 -fno-diagnostics-show-line-numbers.
918 * dwarf2out.c (gen_producer_string): Add
919 OPT_fdiagnostics_show_line_numbers to the ignored options.
920 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
921 one setting" options.
922 (append_compiler_options): Likewise to the dropped options.
923 (append_diag_options): Likewise to the passed-on options.
924 * opts.c (common_handle_option): Handle the new option.
925 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
927 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
929 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
930 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
931 third argument of type "const signed char" to descriptions of
932 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
933 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
934 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
935 __builtin_bcdsub_ov functions.
937 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
939 PR tree-optimization/86858
940 * tree-vect-loop.c (vect_is_simple_reduction): Restore
941 flow_bb_inside_loop_p calls.
943 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
945 PR tree-optimization/86871
946 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
947 instead of gimple_assign_lhs.
949 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
952 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
953 register constraint to operand 0.
954 (add<mode>3_carryinC): Likewise.
955 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
957 2018-08-09 Martin Liska <mliska@suse.cz>
960 * common.opt: Remove extra line.
962 2018-08-09 Martin Liska <mliska@suse.cz>
964 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
965 at the end of a line, make first letter capital and end up
966 a sentence with a dot.
967 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
968 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
969 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
970 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
971 (PARAM_MAX_ISL_OPERATIONS): Likewise.
972 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
973 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
974 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
975 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
976 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
977 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
978 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
979 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
980 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
981 (PARAM_TREE_REASSOC_WIDTH): Likewise.
982 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
983 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
984 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
986 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
989 * config/s390/s390.c (s390_option_override_internal): Reduce the
990 stack-clash-protection-probe-interval param if it would be too big
993 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
996 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
997 * config/m68k/m68k.c (handle_move_double): Don't call
998 m68k_final_prescan_insn.
999 (m68k_adjust_decorated_operand): Renamed from
1000 m68k_final_prescan_insn, remove first and third operand and
1002 (print_operand): Call it.
1003 (print_operand_address): Call it.
1005 2018-08-08 Nathan Sidwell <nathan@acm.org>
1007 * diagnostic.c (diagnostic_report_current_module): Use
1008 linemap_included_from & linemap_included_from_linemap.
1010 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
1012 * config/aarch64/aarch64-cores.def: Add phecda core.
1013 * config/aarch64/aarch64-tune.md: Regenerate.
1014 * doc/invoke.texi: Add phecda core.
1016 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
1019 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
1021 * config/s390/s390.md ("movti"): Add more alternatives for
1022 constant to GPR copies.
1024 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
1026 * config/s390/s390.c: Fix whitespace damage throughout the file.
1027 * config/s390/s390.h: Likewise.
1028 * config/s390/tpf.h: Likewise.
1030 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
1032 * config/s390/s390.c (s390_loadrelative_operand_p):
1033 Remove TARGET_CPU_ZARCH usages.
1034 (s390_rtx_costs): Likewise.
1035 (s390_legitimate_constant_p): Likewise.
1036 (s390_cannot_force_const_mem): Likewise.
1037 (legitimate_reload_constant_p): Likewise.
1038 (s390_preferred_reload_class): Likewise.
1039 (legitimize_pic_address): Likewise.
1040 (legitimize_tls_address): Likewise.
1041 (s390_split_branches): Removed.
1042 (s390_add_execute): Removed.
1043 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
1044 (s390_mainpool_start): Likewise.
1045 (s390_mainpool_finish): Likewise.
1046 (s390_mainpool_cancel): Removed.
1047 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
1048 (s390_chunkify_cancel): Likewise.
1049 (s390_return_addr_rtx): Likewise.
1050 (s390_register_info): Remove split_branches_pending_p uages.
1051 (s390_optimize_register_info): Likewise.
1052 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
1053 split_branches_pending_p usages.
1054 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
1055 (s390_load_got): Likewise.
1056 (s390_expand_split_stack_prologue): Likewise.
1057 (output_asm_nops): Likewise.
1058 (s390_function_profiler): Likewise.
1059 (s390_emit_call): Likewise.
1060 (s390_conditional_register_usage): Likewise.
1061 (s390_optimize_prologue): Likewise.
1062 (s390_reorg): Remove TARGET_CPU_ZARCH and
1063 split_branches_pending_p usages.
1064 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
1066 (s390_output_indirect_thunk_function): Likewise.
1067 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
1068 (TARGET_CPU_ZARCH_P): Removed.
1069 (struct machine_function): Remove split_branches_pending_p.
1070 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
1072 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
1074 * common/config/s390/s390-common.c (processor_flags_table):
1076 * config.gcc: Remove with_arch/with_tune support.
1077 * config/s390/2064.md: Remove cpu attribute comparisons.
1078 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1080 * config/s390/linux.h (ASM_SPEC):
1081 Remove -march support.
1082 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
1083 Use a table to get an arch level.
1084 * config/s390/s390-opts.h (enum processor_type):
1086 * config/s390/s390.c
1087 (processor_table): Remove entries, add arch_level values.
1088 (s390_issue_rate): Remove cases.
1089 (s390_option_override): Adjust
1090 s390_option_override_internal() call.
1091 (s390_option_override_internal): Remove deprecation warning.
1092 (s390_valid_target_attribute_tree): Adjust
1093 s390_option_override_internal() call.
1094 * config/s390/s390.h (struct s390_processor):
1095 Share with s390-c.c, add arch_level field.
1096 * config/s390/s390.md:
1097 Remove occurrences in cpu attribute.
1098 * config/s390/s390.opt: Remove -march/-mtune support.
1099 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
1100 * doc/invoke.texi: Remove deprecation warning.
1102 2018-08-08 Luis Machado <luis.machado@linaro.org>
1104 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
1106 (qdf24xx_tunings): Set vector cost structure to
1107 qdf24xx_vector_cost.
1109 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
1110 <register_sextend>: Set to 3.
1112 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
1115 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
1116 * config/aarch64/aarch64-simd.md
1117 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
1118 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
1119 (aarch64_frecpx<mode>): ...this new pattern.
1120 * config/aarch64/aarch64-simd-builtins.def: Remove comment
1121 about aarch64_frecp<FRECP:frecp_suffix><mode>.
1123 2018-08-07 Martin Liska <mliska@suse.cz>
1126 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
1127 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
1128 * predict.def (PRED_MALLOC_NONNULL): New predictor.
1129 * doc/extend.texi: Document that malloc attribute adds
1132 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
1135 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1136 Define to speculation_safe_value_not_needed.
1138 2018-08-06 Jeff Law <law@redhat.com>
1140 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
1141 the vr_values instance to cprop_into_stmt.
1142 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
1143 (cprop_operand): Also query EVRP to determine if OP is a constant.
1145 2018-08-06 Nathan Sidwell <nathan@acm.org>
1147 * diagnostic.c (diagnostic_report_current_module): Reroll
1148 included-at loop. Translate text.
1150 2018-08-06 David Malcolm <dmalcolm@redhat.com>
1152 * function-tests.c (selftest::test_expansion_to_rtl): Call
1153 free_after_compilation.
1155 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1157 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
1159 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
1161 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
1162 loops with memory block operations from getting unrolled.
1164 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
1167 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1168 Define to speculation_safe_value_not_needed.
1170 2018-08-06 Jeff Law <law@redhat.com>
1172 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
1175 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1178 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
1179 with all enabled __intN types.
1181 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
1183 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1185 * alias.c (record_set): Check for clobber high.
1186 * cfgexpand.c (expand_gimple_stmt): Likewise.
1187 * combine-stack-adj.c (single_set_for_csa): Likewise.
1188 * combine.c (find_single_use_1): Likewise.
1189 (set_nonzero_bits_and_sign_copies): Likewise.
1190 (get_combine_src_dest): Likewise.
1191 (is_parallel_of_n_reg_sets): Likewise.
1192 (try_combine): Likewise.
1193 (record_dead_and_set_regs_1): Likewise.
1194 (reg_dead_at_p_1): Likewise.
1195 (reg_dead_at_p): Likewise.
1196 * dce.c (deletable_insn_p): Likewise.
1197 (mark_nonreg_stores_1): Likewise.
1198 (mark_nonreg_stores_2): Likewise.
1199 * df-scan.c (df_find_hard_reg_defs): Likewise.
1200 (df_uses_record): Likewise.
1201 (df_get_call_refs): Likewise.
1202 * dwarf2out.c (mem_loc_descriptor): Likewise.
1203 * haifa-sched.c (haifa_classify_rtx): Likewise.
1204 * ira-build.c (create_insn_allocnos): Likewise.
1205 * ira-costs.c (scan_one_insn): Likewise.
1206 * ira.c (equiv_init_movable_p): Likewise.
1207 (rtx_moveable_p): Likewise.
1208 (interesting_dest_for_shprep): Likewise.
1209 * jump.c (mark_jump_label_1): Likewise.
1210 * postreload-gcse.c (record_opr_changes): Likewise.
1211 * postreload.c (reload_cse_simplify): Likewise.
1212 (struct reg_use): Add source expr.
1213 (reload_combine): Check for clobber high.
1214 (reload_combine_note_use): Likewise.
1215 (reload_cse_move2add): Likewise.
1216 (move2add_note_store): Likewise.
1217 * print-rtl.c (print_pattern): Likewise.
1218 * recog.c (decode_asm_operands): Likewise.
1219 (store_data_bypass_p): Likewise.
1220 (if_test_bypass_p): Likewise.
1221 * regcprop.c (kill_clobbered_value): Likewise.
1222 (kill_set_value): Likewise.
1223 * reginfo.c (reg_scan_mark_refs): Likewise.
1224 * reload1.c (maybe_fix_stack_asms): Likewise.
1225 (eliminate_regs_1): Likewise.
1226 (elimination_effects): Likewise.
1227 (mark_not_eliminable): Likewise.
1228 (scan_paradoxical_subregs): Likewise.
1229 (forget_old_reloads_1): Likewise.
1230 * reorg.c (find_end_label): Likewise.
1231 (try_merge_delay_insns): Likewise.
1232 (redundant_insn): Likewise.
1233 (own_thread_p): Likewise.
1234 (fill_simple_delay_slots): Likewise.
1235 (fill_slots_from_thread): Likewise.
1236 (dbr_schedule): Likewise.
1237 * resource.c (update_live_status): Likewise.
1238 (mark_referenced_resources): Likewise.
1239 (mark_set_resources): Likewise.
1240 * rtl.c (copy_rtx): Likewise.
1241 * rtlanal.c (reg_referenced_p): Likewise.
1242 (single_set_2): Likewise.
1243 (noop_move_p): Likewise.
1244 (note_stores): Likewise.
1245 * sched-deps.c (sched_analyze_reg): Likewise.
1246 (sched_analyze_insn): Likewise.
1248 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1250 * cse.c (invalidate_reg): New function extracted from...
1251 (invalidate): ...here.
1252 (canonicalize_insn): Check for clobber high.
1253 (invalidate_from_clobbers): invalidate clobber highs.
1254 (invalidate_from_sets_and_clobbers): Likewise.
1255 (count_reg_usage): Check for clobber high.
1256 (insn_live_p): Likewise.
1257 * cselib.c (cselib_expand_value_rtx_1):Likewise.
1258 (cselib_invalidate_regno): Check for clobber in setter.
1259 (cselib_invalidate_rtx): Pass through setter.
1260 (cselib_invalidate_rtx_note_stores):
1261 (cselib_process_insn): Check for clobber high.
1262 * cselib.h (cselib_invalidate_rtx): Add operand.
1264 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1266 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
1267 (mark_not_eliminable): Likewise.
1268 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
1269 * lra-lives.c (process_bb_lives): Check for clobber high.
1270 * lra.c (new_insn_reg): Remember clobber highs.
1271 (collect_non_operand_hard_regs): Check for clobber high.
1272 (lra_set_insn_recog_data): Likewise.
1273 (add_regs_to_insn_regno_info): Likewise.
1274 (lra_update_insn_regno_info): Likewise.
1276 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1278 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
1279 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
1281 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1283 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
1284 (copy_insn_1): Likewise.
1285 (gen_hard_reg_clobber_high): New gen function.
1286 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
1287 * genemit.c (gen_exp): Likewise.
1288 (gen_emit_seq): Pass through info.
1289 (gen_insn): Check for CLOBBER_HIGH.
1290 (gen_expand): Pass through info.
1291 (gen_split): Likewise.
1292 (output_add_clobbers): Likewise.
1293 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
1294 (remove_clobbers): Likewise.
1295 * rtl.h (gen_hard_reg_clobber_high): New declaration.
1297 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1299 * doc/rtl.texi (clobber_high): Add.
1300 (parallel): Add in clobber high
1301 * rtl.c (rtl_check_failed_code3): Add function.
1302 * rtl.def (CLOBBER_HIGH): Add expression.
1303 * rtl.h (RTL_CHECKC3): Add macro.
1304 (rtl_check_failed_code3): Add declaration.
1305 (XC3EXP): Add macro.
1307 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
1310 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
1311 cfun->machine->max_used_stack_alignment if needed.
1313 2018-08-04 Martin Sebor <msebor@redhat.com>
1315 PR tree-optimization/86571
1316 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
1319 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
1322 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1325 2018-08-03 Jeff Law <law@redhat.com>
1328 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1329 Define to speculation_safe_value_not_needed.
1331 2018-08-03 David Malcolm <dmalcolm@redhat.com>
1333 * doc/gcov.texi (-x): Remove duplicate "to".
1334 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
1335 (-Wif-not-aligned): Remove duplicate "is".
1336 (-flto): Remove duplicate "the".
1337 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
1338 duplicate "v5.00.b".
1339 (MSP430 Options): Remove duplicate "and" from the description
1340 of "-mgprel-sec=regexp".
1341 (x86 Options): Remove duplicate copies of "vmldLog102" and
1342 vmlsLog104 from description of "-mveclibabi=type".
1344 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
1346 * internal-fn.h (first_commutative_argument): Declare.
1347 * internal-fn.c (first_commutative_argument): New function.
1348 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
1349 restrictions for pattern statements. Use first_commutative_argument
1350 to look for commutative operands in calls to internal functions.
1352 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
1354 * Makefile.in (wide-int-range.o): New.
1355 * tree-vrp.c: Move all the wide_int_* functions to...
1356 * wide-int-range.cc: ...here.
1357 * tree-vrp.h: Move all the wide_int_* prototypes to...
1358 * wide-int-range.h: ...here.
1360 2018-08-03 Tom de Vries <tdevries@suse.de>
1362 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
1364 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
1365 * except.c (output_function_exception_table): Do early exit if
1366 targetm_common.except_unwind_info (&global_options) == UI_NONE.
1368 2018-08-03 Martin Liska <mliska@suse.cz>
1370 * predict.c (dump_prediction): Change to 2 digits
1373 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
1375 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
1377 (falkor_am_1_gtov_gtov): ... a new insn reservation.
1379 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
1381 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
1382 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
1383 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
1384 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
1385 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
1387 2018-08-02 David Malcolm <dmalcolm@redhat.com>
1389 * diagnostic-show-locus.c (diagnostic_show_locus): Use
1390 pp_take_prefix when saving the existing prefix.
1391 * diagnostic.c (diagnostic_append_note): Likewise.
1392 * langhooks.c (lhd_print_error_function): Likewise.
1393 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
1394 param's type. Free the existing prefix.
1395 (pp_take_prefix): New function.
1396 (pretty_printer::pretty_printer): Drop the prefix parameter.
1397 Rename the length parameter to match the comment.
1398 (pretty_printer::~pretty_printer): Free the prefix.
1399 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
1401 (struct pretty_printer): Drop the "const" from "prefix" field's
1402 type and clarify memory management.
1403 (pp_set_prefix): Drop the "const" from the 2nd param.
1404 (pp_take_prefix): New decl.
1406 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
1408 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
1409 for word_mode_ok here instead of passing as argument.
1410 (expand_block_compare): Change select_block_compare_mode() call.
1411 (expand_strncmp_gpr_sequence): New function.
1412 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
1414 2018-08-02 Jeff Law <law@redhat.com>
1417 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1418 Define to speculation_safe_value_not_needed.
1421 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1422 Define to speculation_safe_value_not_needed.
1424 2018-08-02 Tom de Vries <tdevries@suse.de>
1427 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
1428 function. Return UI_TARGET unconditionally.
1429 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
1430 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
1432 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
1434 * genemit.c (print_overload_test): Fix typo.
1436 2018-08-02 Richard Biener <rguenther@suse.de>
1438 PR tree-optimization/86816
1439 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
1440 which checks for value availability before querying it.
1441 (gvn_uses_equal): Use it.
1442 (same_succ_hash): Likewise.
1443 (gimple_equal_p): Likewise.
1445 2018-08-02 Nick Clifton <nickc@redhat.com>
1448 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1449 Define to speculation_safe_value_not_needed.
1452 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1453 Define to speculation_safe_value_not_needed.
1456 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1457 Define to speculation_safe_value_not_needed.
1460 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1461 Define to speculation_safe_value_not_needed.
1464 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1465 Define to speculation_safe_value_not_needed.
1468 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1469 Define to speculation_safe_value_not_needed.
1472 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1473 Define to speculation_safe_value_not_needed.
1476 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1477 Define to speculation_safe_value_not_needed.
1480 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
1481 speculation_safe_value_not_needed.
1484 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
1485 to speculation_safe_value_not_needed.
1487 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
1489 * doc/md.texi: Expand the documentation of instruction names
1490 to mention port-local uses. Document '@' in pattern names.
1491 * read-md.h (overloaded_instance, overloaded_name): New structs.
1493 (md_reader::handle_overloaded_name): New member function.
1494 (md_reader::get_overloads): Likewise.
1495 (md_reader::m_first_overload): New member variable.
1496 (md_reader::m_next_overload_ptr): Likewise.
1497 (md_reader::m_overloads_htab): Likewise.
1498 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
1499 m_next_overload_ptr and m_overloads_htab.
1500 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
1501 (get_mode_token, get_code_token, get_int_token): New functions.
1502 (map_attr_string): Add an optional argument that passes back
1503 the associated iterator.
1504 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
1505 (md_reader::handle_overloaded_name, add_overload_instance): New
1507 (apply_iterators): Handle '@' names. Report an error if '@'
1508 is used without iterators.
1509 (initialize_iterators): Initialize the new iterator_group fields.
1510 * genopinit.c (handle_overloaded_code_for)
1511 (handle_overloaded_gen): New functions.
1512 (main): Use them to print declarations of maybe_code_for_* and
1513 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
1514 * genemit.c (print_overload_arguments, print_overload_test)
1515 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
1516 (main): Use it to print definitions of maybe_code_for_* and
1517 maybe_gen_* functions.
1518 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
1519 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
1520 instead of explicit mode checks.
1521 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
1522 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
1523 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
1524 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
1525 (aarch64_expand_compare_and_swap): Likewise
1526 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
1527 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
1528 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
1529 (aarch64_constant_pool_reload_icode): Delete.
1530 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
1531 instead of aarch64_constant_pool_reload_icode. Use
1532 code_for_aarch64_reload_mov instead of explicit mode checks.
1533 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
1534 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
1535 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
1536 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
1537 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
1538 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
1539 (aarch64_atomic_load_op_code): Delete.
1540 (aarch64_emit_atomic_load_op): Likewise.
1541 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
1542 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
1543 instead of aarch64_emit_atomic_load_op.
1544 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
1545 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
1546 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
1547 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
1548 character before the pattern name.
1549 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1550 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
1551 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
1552 (aarch64_frecps<mode>): Likewise.
1553 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
1554 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
1555 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
1556 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
1557 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
1559 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
1561 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
1562 Allow HFmode constants if TARGET_FP_F16INST.
1564 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
1567 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
1568 No longer check last store for clobber of address register.
1570 2018-08-02 Martin Liska <mliska@suse.cz>
1572 PR gcov-profile/86817
1573 * gcov.c (process_all_functions): New function.
1575 (process_file): Move functions processing to
1576 process_all_functions.
1578 2018-08-02 David Malcolm <dmalcolm@redhat.com>
1580 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
1581 "const" to the "gimple *" and "rtx_insn *" parameters.
1582 * dumpfile.h (dump_user_location_t::dump_user_location_t):
1584 (dump_location_t::dump_location_t): Likewise.
1586 2018-08-01 Martin Sebor <msebor@redhat.com>
1588 PR tree-optimization/86650
1589 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
1590 rather than a "gcall *". Directly pass the data of interest
1591 to percent_K_format, rather than building a temporary CALL_EXPR
1593 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
1594 (gimple_fold_builtin_strncat): Adjust.
1595 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
1596 gcall* argument with gimple*.
1597 * gimple-ssa-warn-restrict.c (check_call): Same.
1598 (wrestrict_dom_walker::before_dom_children): Same.
1599 (builtin_access::builtin_access): Same.
1600 (check_bounds_or_overlap): Same
1601 (maybe_diag_overlap): Same.
1602 (maybe_diag_offset_bounds): Same.
1603 * tree-diagnostic.c (default_tree_printer): Move usage of
1604 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
1606 * tree-pretty-print.c (percent_K_format): Add argument.
1607 * tree-pretty-print.h: Add argument.
1608 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
1609 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
1610 (maybe_diag_stxncpy_trunc): Same.
1611 (handle_builtin_stxncpy): Same.
1612 (handle_builtin_strcat): Same.
1614 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1616 * match.pd: Optimise pointer range checks.
1618 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1620 PR tree-optimization/86758
1621 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
1622 to remove pattern statements.
1624 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1626 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
1627 result of dfs_enumerate_from when constructing stmt_vec_infos,
1628 instead of additionally calling get_loop_body.
1630 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1632 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
1634 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
1635 When creating an iv, assert that the step is not known to be zero.
1636 (vect_setup_realignment): Update call accordingly.
1637 * tree-vect-stmts.c (vectorizable_store): Likewise.
1638 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
1640 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1642 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
1643 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
1644 (vectorizable_reduction): Likewise.
1645 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
1646 (vect_detect_hybrid_slp_stmts): Likewise.
1647 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
1649 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
1651 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
1652 (wide_int_set_zero_nonzero_bits): ...this.
1653 (zero_nonzero_bits_from_vr): Rename to...
1654 (vrp_set_zero_nonzero_bits): ...this.
1655 (extract_range_from_multiplicative_op_1): Abstract wide int
1657 (wide_int_range_multiplicative_op): ...here.
1658 (extract_range_from_binary_expr_1): Extract wide int binary
1659 operations into their own functions.
1660 (wide_int_range_lshift): New.
1661 (wide_int_range_can_optimize_bit_op): New.
1662 (wide_int_range_shift_undefined_p): New.
1663 (wide_int_range_bit_xor): New.
1664 (wide_int_range_bit_ior): New.
1665 (wide_int_range_bit_and): New.
1666 (wide_int_range_trunc_mod): New.
1667 (extract_range_into_wide_ints): New.
1668 (vrp_shift_undefined_p): New.
1669 (extract_range_from_multiplicative_op): New.
1670 (vrp_can_optimize_bit_op): New.
1671 * tree-vrp.h (value_range::dump): New.
1672 (wide_int_range_multiplicative_op): New.
1673 (wide_int_range_lshift):New.
1674 (wide_int_range_shift_undefined_p): New.
1675 (wide_int_range_bit_xor): New.
1676 (wide_int_range_bit_ior): New.
1677 (wide_int_range_bit_and): New.
1678 (wide_int_range_trunc_mod): New.
1679 (zero_nonzero_bits_from_bounds): Rename to...
1680 (wide_int_set_zero_nonzero_bits): ...this.
1681 (zero_nonzero_bits_from_vr): Rename to...
1682 (vrp_set_zero_nonzero_bits): ...this.
1683 (range_easy_mask_min_max): Rename to...
1684 (wide_int_range_can_optimize_bit_op): this.
1685 * vr-values.c (simplify_bit_ops_using_ranges): Rename
1686 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
1688 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1690 * tree-vectorizer.h (vect_orig_stmt): New function.
1691 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
1692 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
1693 (vect_create_epilog_for_reduction): Likewise.
1694 (vectorizable_live_operation): Likewise.
1695 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
1696 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
1697 * tree-vect-stmts.c (vectorizable_call): Likewise.
1698 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
1700 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1702 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
1704 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
1705 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
1706 (vect_transform_loop): Likewise.
1707 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1709 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1711 * tree-vectorizer.h (vect_schedule_slp): Return void.
1712 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1713 (vect_schedule_slp): Likewise.
1715 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1717 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
1719 (vect_transform_loop): Update calls accordingly. Schedule SLP
1720 instances before the main loop, if any exist.
1722 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
1724 PR tree-optimization/86749
1725 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
1726 If the lhs is used in a COND_EXPR, check that it is being used
1727 as the "then" or "else" value.
1729 2018-08-01 Tom de Vries <tdevries@suse.de>
1732 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
1733 speculation_safe_value_not_needed.
1735 2018-08-01 Richard Biener <rguenther@suse.de>
1737 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
1740 2018-08-01 Martin Liska <mliska@suse.cz>
1742 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
1743 format how successful transformation is dumped.
1744 (gimple_mod_pow2_value_transform): Likewise.
1745 (gimple_mod_subtract_transform): Likewise.
1746 (gimple_stringops_transform): Likewise.
1748 2018-08-01 Martin Liska <mliska@suse.cz>
1751 * value-prof.c (interesting_stringop_to_profile_p):
1752 Simplify the code and add BUILT_IN_MEMMOVE.
1753 (gimple_stringops_transform): Likewise.
1755 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
1757 * config/aarch64/aarch64-simd.md
1758 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
1759 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
1760 use GPI iterator instead of SI mode.
1762 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
1764 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
1765 rs6000_speculation_barrier.
1766 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
1767 new barrier pattern name.
1769 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
1771 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
1772 (speculation_barrier): New insn.
1774 2018-08-01 Richard Biener <rguenther@suse.de>
1777 * graphite.h: Include isl/id.h and isl/space.h to allow build
1780 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
1783 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
1784 mode for COFF targets.
1785 * defaults.h (TARGET_COFF): Define.
1786 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
1787 TARGET_COFF): Define.
1788 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
1789 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
1790 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
1792 2018-07-31 Alexandre Oliva <oliva@adacore.com>
1794 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
1795 * gimple-streamer-out.c (output_bb): Save it.
1796 * lto-streamer-in.c (input_struct_function_base): Restore
1797 instance discriminator if available. Create map on demand.
1798 * lto-streamer-out.c (output_struct_function_base): Save it if
1800 * final.c (decl_to_instance_map): Document LTO strategy.
1802 2018-07-31 Alexandre Oliva <oliva@adacore.com>
1803 Olivier Hainque <hainque@adacore.com>
1805 * debug.h (decl_to_instance_map_t): New type.
1806 (decl_to_instance_map): Declare.
1807 (maybe_create_decl_to_instance_map): New inline function.
1808 * final.c (bb_discriminator, last_bb_discriminator): New statics,
1809 to track basic block discriminators.
1810 (final_start_function_1): Initialize them.
1811 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
1813 (decl_to_instance_map): New variable.
1814 (map_decl_to_instance, maybe_set_discriminator): New functions.
1815 (notice_source_line): Set discriminator.
1817 2018-07-31 Ian Lance Taylor <iant@golang.org>
1819 * targhooks.c (default_have_speculation_safe_value): Add
1822 2018-07-31 David Malcolm <dmalcolm@redhat.com>
1824 * dump-context.h: Include "pretty-print.h".
1825 (dump_context::refresh_dumps_are_enabled): New decl.
1826 (dump_context::emit_item): New decl.
1827 (class dump_context): Add fields "m_test_pp" and
1829 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
1830 (temp_dump_context::get_dumped_text): New decl.
1831 (class temp_dump_context): Add field "m_pp".
1832 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
1833 (dump_context::refresh_dumps_are_enabled): ...and add a test for
1835 (set_dump_file): Update for above change.
1836 (set_alt_dump_file): Likewise.
1837 (dump_loc): New overload, taking a pretty_printer *.
1838 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
1839 to any test pretty-printer.
1840 (make_item_for_dump_gimple_stmt): New function, adapted from
1841 optinfo::add_gimple_stmt.
1842 (dump_context::dump_gimple_stmt): Call it, and use the result,
1843 eliminating the direct usage of dump_file and alt_dump_file in
1844 favor of indirectly using them via emit_item.
1845 (make_item_for_dump_gimple_expr): New function, adapted from
1846 optinfo::add_gimple_expr.
1847 (dump_context::dump_gimple_expr): Call it, and use the result,
1848 eliminating the direct usage of dump_file and alt_dump_file in
1849 favor of indirectly using them via emit_item.
1850 (make_item_for_dump_generic_expr): New function, adapted from
1852 (dump_context::dump_generic_expr): Call it, and use the result,
1853 eliminating the direct usage of dump_file and alt_dump_file in
1854 favor of indirectly using them via emit_item.
1855 (make_item_for_dump_printf_va): New function, adapted from
1856 optinfo::add_printf_va.
1857 (make_item_for_dump_printf): New function.
1858 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
1859 and use the result, eliminating the direct usage of dump_file and
1860 alt_dump_file in favor of indirectly using them via emit_item.
1861 (make_item_for_dump_dec): New function.
1862 (dump_context::dump_dec): Call it, and use the result,
1863 eliminating the direct usage of dump_file and alt_dump_file in
1864 favor of indirectly using them via emit_item.
1865 (make_item_for_dump_symtab_node): New function, adapted from
1866 optinfo::add_symtab_node.
1867 (dump_context::dump_symtab_node): Call it, and use the result,
1868 eliminating the direct usage of dump_file and alt_dump_file in
1869 favor of indirectly using them via emit_item.
1870 (dump_context::begin_scope): Reimplement, avoiding direct usage
1871 of dump_file and alt_dump_file in favor of indirectly using them
1873 (dump_context::emit_item): New member function.
1874 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
1875 Set up test pretty-printer on the underlying context. Call
1876 refresh_dumps_are_enabled.
1877 (temp_dump_context::~temp_dump_context): Call
1878 refresh_dumps_are_enabled.
1879 (temp_dump_context::get_dumped_text): New member function.
1880 (selftest::verify_dumped_text): New function.
1881 (ASSERT_DUMPED_TEXT_EQ): New macro.
1882 (selftest::test_capture_of_dump_calls): Run all tests twice, with
1883 and then without optinfo enabled. Add uses of
1884 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
1885 * dumpfile.h: Update comment for the dump_* API.
1886 * optinfo-emit-json.cc
1887 (selftest::test_building_json_from_dump_calls): Update for new
1888 param for temp_dump_context ctor.
1889 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
1890 and "m_owned" field.
1891 (optinfo_item::~optinfo_item): Likewise.
1892 (optinfo::add_item): New member function.
1893 (optinfo::emit): Update comment.
1894 (optinfo::add_string): Delete.
1895 (optinfo::add_printf): Delete.
1896 (optinfo::add_printf_va): Delete.
1897 (optinfo::add_gimple_stmt): Delete.
1898 (optinfo::add_gimple_expr): Delete.
1899 (optinfo::add_tree): Delete.
1900 (optinfo::add_symtab_node): Delete.
1901 (optinfo::add_dec): Delete.
1902 * optinfo.h (class dump_context): New forward decl.
1903 (optinfo::add_item): New decl.
1904 (optinfo::add_string): Delete.
1905 (optinfo::add_printf): Delete.
1906 (optinfo::add_printf_va): Delete.
1907 (optinfo::add_gimple_stmt): Delete.
1908 (optinfo::add_gimple_expr): Delete.
1909 (optinfo::add_tree): Delete.
1910 (optinfo::add_symtab_node): Delete.
1911 (optinfo::add_dec): Delete.
1912 (optinfo::add_poly_int): Delete.
1913 (optinfo_item::optinfo_item): Remove "owned" param.
1914 (class optinfo_item): Remove field "m_owned".
1916 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1919 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
1920 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
1921 requested variable alignment.
1922 (expand_one_ssa_partition): Likewise.
1923 (expand_one_var): Likewise.
1925 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1927 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1928 to speculation_safe_value_not_needed.
1930 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1932 * targhooks.h (speculation_safe_value_not_needed): New prototype.
1933 * targhooks.c (speculation_safe_value_not_needed): New function.
1934 * target.def (have_speculation_safe_value): Update documentation.
1935 * doc/tm.texi: Regenerated.
1937 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1939 * config/aarch64/iterators.md (ALLI_TI): New iterator.
1940 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
1942 (despeculate_copy<ALLI:mode>_insn): New insn.
1943 (despeculate_copyti_insn): New insn.
1944 (despeculate_simple<ALLI:mode>): New insn
1945 (despeculate_simpleti): New insn.
1946 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
1948 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
1949 aarch64_speculation_safe_value.
1950 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
1952 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1954 * config/aarch64/aarch64-speculation.cc: New file.
1955 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
1956 before pass_reorder_blocks.
1957 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
1959 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
1960 X14 and X15 when tracking speculation.
1961 * config/aarch64/aarch64.md (register name constants): Add
1962 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
1963 (unspec): Add UNSPEC_SPECULATION_TRACKER.
1964 (speculation_barrier): New insn attribute.
1965 (cmp<mode>): Allow SP in comparisons.
1966 (speculation_tracker): New insn.
1967 (speculation_barrier): Add speculation_barrier attribute.
1968 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
1969 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
1970 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
1972 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1974 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
1975 aarch64_track_speculation is true.
1976 (tb<optab><mode>1): Likewise.
1977 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
1978 generate CB[N]Z when tracking speculation.
1979 (aarch64_split_compare_and_swap): Likewise.
1980 (aarch64_split_atomic_op): Likewise.
1982 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1984 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
1986 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1988 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
1989 (speculation_barrier): New insn.
1991 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1993 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
1994 * config/arm/arm.md (speculation_barrier): New expand.
1995 (speculation_barrier_insn): New pattern.
1997 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
1999 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
2000 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
2001 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
2002 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
2004 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
2005 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
2006 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
2007 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
2008 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
2009 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
2010 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
2011 * builtins.c (expand_speculation_safe_value): New function.
2012 (expand_builtin): Call it.
2013 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
2014 * doc/extend.texi: Document __builtin_speculation_safe_value.
2015 * doc/md.texi: Document "speculation_barrier" pattern.
2016 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
2017 TARGET_HAVE_SPECULATION_SAFE_VALUE.
2018 * doc/tm.texi: Regenerated.
2019 * target.def (have_speculation_safe_value, speculation_safe_value): New
2021 * targhooks.c (default_have_speculation_safe_value): New function.
2022 (default_speculation_safe_value): New function.
2023 * targhooks.h (default_have_speculation_safe_value): Add prototype.
2024 (default_speculation_safe_value): Add prototype.
2026 2018-07-31 David Malcolm <dmalcolm@redhat.com>
2028 * dump-context.h (dump_context::dump_loc): New decl.
2029 * dumpfile.c (dump_context::dump_loc): New member function.
2030 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
2031 and dump_gimple_stmt.
2032 (dump_context::dump_gimple_expr_loc): Likewise, using
2034 (dump_context::dump_generic_expr_loc): Likewise, using
2036 (dump_context::dump_printf_loc_va): Likewise, using
2038 (dump_context::begin_scope): Explicitly using the global function
2039 "dump_loc", rather than the member function.
2041 2018-07-31 Martin Sebor <msebor@redhat.com>
2043 PR tree-optimization/86741
2044 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
2046 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
2048 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
2049 depend on whether prefetch instructions will be emitted or not.
2050 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
2051 will be emitted or not.
2052 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
2053 (TARGET_SETMEM_PFD): New macros.
2055 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2057 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
2058 (NULL_STMT_VEC_INFO): Delete.
2059 (stmt_vec_info::operator*): Likewise.
2060 (stmt_vec_info::operator gimple *): Likewise.
2061 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
2062 of NULL_STMT_VEC_INFO.
2063 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
2064 (vect_reassociating_reduction_p): Likewise.
2065 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
2066 (vectorizable_store): Likewise.
2067 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
2068 (vec_info::free_stmt_vec_infos): Likewise.
2070 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2072 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
2073 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
2075 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2077 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
2078 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
2079 (vec_info::free_stmt_vec_info): New private member functions.
2080 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
2081 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
2082 * tree-parloops.c (gather_scalar_reductions): Remove calls to
2083 set_stmt_vec_info_vec and free_stmt_vec_infos.
2084 * tree-vect-loop.c (_loop_vec_info): Remove call to
2085 set_stmt_vec_info_vec.
2086 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
2087 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
2088 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
2089 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
2090 (vec_info::free_stmt_vec_info): ...these new functions. Remove
2091 assignments in {vec_info::,}new_stmt_vec_info that are redundant
2092 with the clearing in the xcalloc.
2093 (stmt_vec_info_vec): Delete.
2094 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
2095 (vectorize_loops): Likewise.
2096 (vec_info::~vec_info): Remove argument from call to
2097 free_stmt_vec_infos.
2098 (vec_info::add_stmt): Remove vinfo argument from call to
2101 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2103 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
2104 rather than a gimple stmt.
2105 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
2106 information for pattern statements when passed the original
2107 statement; instead wait to be passed the pattern statement itself.
2108 Don't call set_vinfo_for_stmt here.
2109 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
2110 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
2111 stmt_vec_infos here.
2112 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
2113 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
2114 stmt_vec_infos entry.
2116 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2118 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
2119 * tree-vectorizer.c (vec_info::replace_stmt): New function.
2120 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
2121 * tree-vect-stmts.c (vectorizable_call): Likewise.
2122 (vectorizable_simd_clone_call): Likewise.
2124 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2126 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
2127 * tree-vectorizer.c (vec_info::remove_stmt): New function.
2128 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
2129 * tree-vect-loop.c (vect_transform_loop): Likewise.
2130 * tree-vect-slp.c (vect_schedule_slp): Likewise.
2131 * tree-vect-stmts.c (vect_remove_stores): Likewise.
2133 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2135 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
2136 (vect_dr_stmt): Delete.
2137 * tree-vectorizer.c (vec_info::lookup_dr): New function.
2138 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
2140 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
2141 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
2142 (vect_verify_datarefs_alignment, vect_peeling_supportable)
2143 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
2144 (vect_analyze_data_refs): Likewise.
2145 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
2147 (vect_find_same_alignment_drs): Likewise.
2148 (vect_slp_analyze_node_dependences): Update calls accordingly.
2149 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
2150 instead of DR_VECT_AUX.
2151 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
2152 of a vector data references. Use vec_info::lookup_dr instead of
2154 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
2155 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
2156 instead of DR_VECT_AUX.
2158 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2160 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
2162 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
2164 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
2165 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
2166 (vect_gen_prolog_loop_niters): Likewise.
2168 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2170 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
2171 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
2172 (vect_known_alignment_in_bytes, vect_dr_behavior)
2173 (vect_get_scalar_dr_size): Take references as dr_vec_infos
2174 instead of data_references. Update calls to other routines for
2175 which the same change has been made.
2176 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
2177 dr_vec_infos instead of stmt_vec_infos.
2178 (vect_analyze_data_ref_dependence): Update call accordingly.
2179 (vect_slp_analyze_data_ref_dependence)
2180 (vect_record_base_alignments): Use DR_VECT_AUX.
2181 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
2182 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
2183 (vector_alignment_reachable_p, vect_get_data_access_cost)
2184 (vect_peeling_supportable, vect_analyze_group_access_1)
2185 (vect_analyze_group_access, vect_analyze_data_ref_access)
2186 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
2187 (vect_compile_time_alias, vect_small_gap_p)
2188 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
2189 (vect_supportable_dr_alignment): Take references as dr_vec_infos
2190 instead of data_references. Update calls to other routines for
2191 which the same change has been made.
2192 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
2193 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
2194 (vect_slp_analyze_and_verify_node_alignment)
2195 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
2196 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
2197 (vect_setup_realignment): Use dr_vec_infos. Update calls after
2199 (_vect_peel_info::dr): Replace with...
2200 (_vect_peel_info::dr_info): ...this new field.
2201 (vect_peeling_hash_get_most_frequent)
2202 (vect_peeling_hash_choose_best_peeling): Update accordingly.
2203 (vect_peeling_hash_get_lowest_cost):
2204 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
2205 routines for which the same change has been made.
2206 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
2208 * tree-vect-loop-manip.c (get_misalign_in_elems)
2209 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
2211 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
2212 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
2213 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
2214 (get_group_load_store_type, get_negative_load_store_type)
2215 (vect_get_data_ptr_increment, vectorizable_store)
2216 (vectorizable_load): Likewise.
2217 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
2218 Update calls to other routines for which the same change has been made.
2220 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2222 * tree-vectorizer.h (vec_info::move_dr): New member function.
2223 (dataref_aux): Rename to...
2224 (dr_vec_info): ...this and add "dr" and "stmt" fields.
2225 (_stmt_vec_info::dr_aux): Update accordingly.
2226 (_stmt_vec_info::data_ref_info): Delete.
2227 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
2228 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
2229 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
2230 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
2232 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
2233 (STMT_VINFO_DR_INFO): New macro.
2234 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
2235 (set_dr_misalignment): Update after rename of dataref_aux.
2236 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
2237 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
2238 initialization of STMT_VINFO_DATA_REF.
2239 * tree-vectorizer.c (vec_info::move_dr): New function.
2240 * tree-vect-patterns.c (vect_recog_bool_pattern)
2241 (vect_recog_mask_conversion_pattern)
2242 (vect_recog_gather_scatter_pattern): Use it.
2243 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
2244 the "dr" and "stmt" fields of dr_vec_info instead of
2245 STMT_VINFO_DATA_REF.
2247 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2249 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
2250 (is_pattern_stmt_p): Use it.
2251 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
2252 on pattern statements.
2254 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2256 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
2257 original stmt as a stmt_vec_info rather than a gimple stmt.
2258 (vect_pattern_recog_1): Take the statement directly as a
2259 stmt_vec_info, rather than via a gimple_stmt_iterator.
2260 Update call to vect_mark_pattern_stmts.
2261 (vect_pattern_recog): Update calls accordingly.
2263 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2265 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
2266 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
2267 a vect_def_type for the first argument.
2268 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
2269 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
2270 operand if it isn't defined by a vectorized statement.
2271 (vect_build_gather_load_calls): Remove the mask_dt argument and
2272 update calls to vect_get_vec_def_for_stmt_copy.
2273 (vectorizable_bswap): Likewise the dt argument.
2274 (vectorizable_call): Update calls to vectorizable_bswap and
2275 vect_get_vec_def_for_stmt_copy.
2276 (vectorizable_simd_clone_call, vectorizable_assignment)
2277 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
2278 (vectorizable_comparison): Update calls to
2279 vect_get_vec_def_for_stmt_copy.
2280 (vectorizable_store): Likewise. Remove now-unnecessary calls to
2282 (vect_get_loop_based_defs): Remove dt argument and update call
2283 to vect_get_vec_def_for_stmt_copy.
2284 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
2285 and vect_get_vec_def_for_stmt_copy.
2286 (vectorizable_load): Update calls to vect_build_gather_load_calls
2287 and vect_get_vec_def_for_stmt_copy.
2288 * tree-vect-loop.c (vect_create_epilog_for_reduction)
2289 (vectorizable_reduction, vectorizable_live_operation): Update calls
2290 to vect_get_vec_def_for_stmt_copy.
2292 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2294 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
2295 and gimple stmt arguments with a stmt_vec_info.
2296 (vect_record_base_alignments): Update calls accordingly.
2297 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
2298 and gimple stmt arguments with a stmt_vec_info.
2299 (vect_build_slp_tree_1): Remove vinfo argument and update call
2300 to vect_record_max_nunits.
2301 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
2302 and vect_record_max_nunits.
2304 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2306 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
2307 file and take a stmt_vec_info instead of a gimple stmt.
2308 (supportable_widening_operation, vect_finish_replace_stmt)
2309 (vect_finish_stmt_generation, vect_get_store_rhs)
2310 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
2311 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
2312 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
2313 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
2314 (vect_create_data_ref_ptr, bump_vector_ptr)
2315 (vect_permute_store_chain, vect_setup_realignment)
2316 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
2317 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
2318 (vectorizable_reduction, vectorizable_induction)
2319 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
2320 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
2321 than gimple stmts as arguments.
2322 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
2323 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
2324 (can_group_stmts_p, vect_check_gather_scatter)
2325 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
2326 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
2327 (vect_permute_load_chain, vect_shift_permute_load_chain)
2328 (vect_transform_grouped_load)
2329 (vect_record_grouped_load_vectors): Likewise.
2330 * tree-vect-loop.c (vect_fixup_reduc_chain)
2331 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
2332 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
2333 (vectorizable_reduction, vectorizable_induction)
2334 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
2335 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
2336 (vect_get_load_store_mask): Likewise.
2337 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
2338 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
2339 * tree-vect-stmts.c (vect_mark_relevant)
2340 (is_simple_and_all_uses_invariant)
2341 (exist_non_indexing_operands_for_use_p, process_use)
2342 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
2343 (vect_get_vec_def_for_operand, vect_get_vec_defs)
2344 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
2345 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
2346 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
2347 (get_negative_load_store_type, get_load_store_type)
2348 (vect_check_load_store_mask, vect_check_store_rhs)
2349 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
2350 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
2351 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
2352 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
2353 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
2354 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
2355 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
2356 (supportable_widening_operation): Likewise.
2358 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2360 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
2361 a stmt_vec_info instead of a gcall.
2362 (vect_check_gather_scatter): Update call accordingly.
2363 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
2365 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
2366 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
2367 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
2368 instead of a gimple stmt.
2369 (vect_transform_loop): Update calls accordingly.
2370 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
2371 stmt_vec_infos instead of gimple stmts.
2372 (vect_analyze_slp_instance): Update use accordingly.
2373 * tree-vect-stmts.c (read_vector_array, write_vector_array)
2374 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
2375 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
2376 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
2377 (vect_gen_widened_results_half, vect_get_loop_based_defs)
2378 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
2379 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
2380 down to subroutines.
2382 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2384 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
2385 of the worklist from a vector of gimple stmts to a vector of
2387 * tree-vect-stmts.c (vect_mark_relevant, process_use)
2388 (vect_mark_stmts_to_be_vectorized): Likewise
2390 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2392 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
2393 statement before passing it to vect_analyze_stmt.
2394 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
2395 the chain of phi vector definitions. Track the exit phi via its
2397 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
2398 STMT_VINFO_REDUC_DEF.
2399 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
2400 stmt_vec_infos to handle the statement chains.
2401 (vect_get_slp_defs): Record the first statement in the node
2402 using a stmt_vec_info.
2403 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
2404 statements here and pass their stmt_vec_info down to subroutines.
2405 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
2406 down to vect_finish_stmt_generation.
2407 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
2408 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
2409 stmt_vec_infos to subroutines.
2410 (vect_remove_stores): Use stmt_vec_infos to handle the statement
2413 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2415 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
2416 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
2417 (vect_permute_store_chain, vect_setup_realignment)
2418 (vect_permute_load_chain, vect_shift_permute_load_chain)
2419 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
2420 stmts internally, and when passing values to other vectorizer routines.
2421 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
2422 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
2423 (vect_analyze_loop_operations, get_initial_def_for_reduction)
2424 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
2425 (vectorizable_reduction, vectorizable_induction)
2426 (vectorizable_live_operation, vect_transform_loop_stmt)
2427 (vect_transform_loop): Likewise.
2428 * tree-vect-patterns.c (vect_reassociating_reduction_p)
2429 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
2430 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
2431 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
2432 (vect_slp_analyze_node_operations_1): Likewise.
2433 * tree-vect-stmts.c (vect_mark_relevant, process_use)
2434 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
2435 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
2436 (vect_finish_stmt_generation_1, get_group_load_store_type)
2437 (get_load_store_type, vect_build_gather_load_calls)
2438 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
2439 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
2440 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
2441 (vectorizable_store, vectorizable_load, vectorizable_condition)
2442 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
2443 (supportable_widening_operation): Likewise.
2444 (vect_get_vector_types_for_stmt): Likewise.
2445 * tree-vectorizer.h (vect_dr_behavior): Likewise.
2447 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2449 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
2450 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
2451 (vect_permute_store_chain, vect_permute_load_chain)
2452 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
2453 repeated stmt_vec_info lookups.
2454 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
2455 (vect_update_ivs_after_vectorizer): Likewise.
2456 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
2457 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
2458 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
2459 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
2460 (vect_bb_slp_scalar_cost): Likewise.
2461 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
2463 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2465 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
2466 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
2467 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
2468 to get gassigns and gcalls, rather than operating on generc gimple
2470 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
2471 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
2472 (vectorizable_load, vect_analyze_stmt): Likewise.
2473 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
2475 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2477 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
2478 return stmt_vec_infos rather than gimple stmts. Do not accept
2480 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
2482 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
2483 Update use of get_later_stmt.
2484 (vect_get_constant_vectors): Update call accordingly.
2485 (vect_schedule_slp_instance): Likewise
2486 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
2487 (vect_slp_analyze_instance_dependence): Likewise.
2488 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
2490 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2492 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
2493 (stmt_info_for_cost::stmt_info): ...this new field.
2494 (add_stmt_costs): Update accordingly.
2495 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
2496 (vect_get_known_peeling_cost): Likewise.
2497 (vect_estimate_min_profitable_iters): Likewise.
2498 * tree-vect-stmts.c (record_stmt_cost): Likewise.
2500 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2502 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
2503 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
2504 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
2506 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
2508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2510 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
2511 a gimple stmt to a stmt_vec_info.
2512 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
2514 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2516 * tree-vectorizer.h (vec_info::grouped_stores): Change from
2517 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
2518 (_loop_vec_info::reduction_chains): Likewise.
2519 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
2521 * tree-vect-slp.c (vect_analyze_slp): Likewise.
2523 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2525 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
2526 a gimple stmt to a stmt_vec_info.
2527 (_stmt_vec_info::next_element): Likewise.
2528 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
2529 (vect_slp_analyze_and_verify_node_alignment)
2530 (vect_analyze_group_access_1, vect_analyze_group_access)
2531 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
2532 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
2533 (vect_supportable_dr_alignment): Update accordingly.
2534 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
2535 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
2536 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
2537 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
2538 * tree-vect-slp.c (vect_build_slp_tree_1)
2539 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
2540 (vect_split_slp_store_group, vect_analyze_slp_instance)
2541 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
2542 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
2543 (get_group_load_store_type, get_load_store_type)
2544 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
2545 (vect_transform_stmt, vect_remove_stores): Likewise.
2547 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2549 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
2551 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
2552 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
2553 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
2554 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
2555 (vector_alignment_reachable_p, vect_get_data_access_cost)
2556 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
2557 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
2558 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
2559 (vect_analyze_group_access_1, vect_analyze_group_access)
2560 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
2561 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
2562 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
2563 result of vect_dr_stmt and use the stmt_vec_info instead of
2564 the associated gimple stmt.
2565 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
2566 (vect_gen_prolog_loop_niters): Likewise.
2567 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
2569 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2571 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
2572 to a vec<stmt_vec_info>.
2573 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
2574 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
2576 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
2577 to a vec<stmt_vec_info>.
2578 (bst_traits::value_type, bst_traits::value_type): Likewise.
2579 (bst_traits::hash): Update accordingly.
2580 (vect_get_and_check_slp_defs): Change the stmts parameter from
2581 a vec<gimple *> to a vec<stmt_vec_info>.
2582 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
2583 (vect_build_slp_tree): Likewise.
2584 (vect_build_slp_tree_2): Likewise. Update uses of
2585 SLP_TREE_SCALAR_STMTS.
2586 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
2587 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
2588 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
2589 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
2590 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
2591 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
2592 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
2593 (vect_get_constant_vectors, vect_get_slp_defs)
2594 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
2595 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
2596 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
2597 instead of gimple stmts.
2598 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
2599 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
2600 (vect_slp_analyze_instance_dependence): Update uses of
2601 SLP_TREE_SCALAR_STMTS.
2602 (vect_slp_analyze_and_verify_node_alignment): Likewise.
2603 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
2604 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
2605 (get_initial_defs_for_reduction): Likewise.
2606 (vect_create_epilog_for_reduction): Likewise.
2607 (vectorize_fold_left_reduction): Likewise.
2608 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
2609 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
2610 (can_vectorize_live_stmts): Likewise.
2612 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2614 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
2615 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
2616 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
2618 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
2620 (gather_scalar_reductions): Update after above interface changes.
2621 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
2622 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
2624 (vect_force_simple_reduction): Likewise.
2625 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
2626 LOOP_VINFO_REDUCTIONS.
2627 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
2629 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2631 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
2632 a gimple stmt to a stmt_vec_info.
2633 * tree-vect-loop.c (vect_active_double_reduction_p)
2634 (vect_force_simple_reduction, vectorizable_reduction): Update
2637 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2639 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
2640 vec<gimple *> to a vec<stmt_vec_info>.
2641 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
2642 the reduction_phis argument from a vec<gimple *> to a
2644 (vectorizable_reduction): Likewise the phis local variable that
2645 is passed to vect_create_epilog_for_reduction. Update for new type
2646 of SLP_TREE_VEC_STMTS.
2647 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
2648 (vectorizable_live_operation): Likewise.
2649 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
2650 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
2652 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2654 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
2655 a gimple stmt to a stmt_vec_info.
2656 (vectorizable_condition, vectorizable_live_operation)
2657 (vectorizable_reduction, vectorizable_induction): Pass back the
2658 vectorized statement as a stmt_vec_info.
2659 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
2660 use of STMT_VINFO_VEC_STMT.
2661 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
2662 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
2663 as stmt_vec_infos rather than gimple stmts.
2664 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
2666 (vectorizable_live_operation): Likewise.
2667 (vectorizable_reduction, vectorizable_induction): Likewise,
2668 updating use of STMT_VINFO_VEC_STMT.
2669 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
2670 of STMT_VINFO_VEC_STMT.
2671 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
2672 (vectorizable_simd_clone_call, vectorizable_conversion)
2673 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
2674 (vectorizable_store, vectorizable_load, vectorizable_condition)
2675 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
2676 from a gimple stmt to a stmt_vec_info.
2677 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
2678 pointer to a stmt_vec_info to the vectorizable_* routines.
2680 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2682 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
2683 a gimple stmt to a stmt_vec_info.
2684 (is_pattern_stmt_p): Update accordingly.
2685 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
2686 (vect_record_grouped_load_vectors): Likewise.
2687 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
2688 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
2689 (vect_model_reduction_cost): Likewise.
2690 (vect_create_epilog_for_reduction): Likewise.
2691 (vectorizable_reduction, vectorizable_induction): Likewise.
2692 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
2693 Return the stmt_vec_info for the pattern statement.
2694 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
2695 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
2696 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
2697 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
2698 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
2699 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
2700 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
2701 (free_stmt_vec_info, vect_is_simple_use): Likewise.
2703 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2705 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
2706 (vect_finish_stmt_generation): Likewise.
2707 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
2708 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
2709 (vect_build_gather_load_calls): Use the return value of the above
2710 functions instead of a separate call to vinfo_for_stmt. Use narrow
2711 scopes for the input gimple stmt and wider scopes for the associated
2712 stmt_vec_info. Use vec_info::lookup_def when setting these
2713 stmt_vec_infos from an SSA_NAME definition.
2714 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
2715 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
2716 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
2717 (vectorizable_store, vectorizable_load, vectorizable_condition)
2718 (vectorizable_comparison): Likewise.
2719 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
2720 (vectorizable_reduction): Likewise.
2722 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2724 * tree-vectorizer.h (vect_is_simple_use): Add an optional
2725 stmt_vec_info * parameter before the optional gimple **.
2726 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
2727 (process_use, vect_get_vec_def_for_operand_1): Update callers.
2728 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
2729 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2730 (vectorizable_live_operation): Likewise.
2731 * tree-vect-patterns.c (type_conversion_p): Likewise.
2732 (vect_look_through_possible_promotion): Likewise.
2733 (vect_recog_rotate_pattern): Likewise.
2734 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
2736 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2738 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
2739 a typedef to a wrapper class.
2740 (NULL_STMT_VEC_INFO): New macro.
2741 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
2742 (stmt_vec_info::operator*): New function.
2743 (stmt_vec_info::operator gimple *): Likewise.
2744 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
2745 (add_stmt_costs): Likewise.
2746 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
2747 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
2748 (vect_get_known_peeling_cost): Likewise.
2749 (vect_estimate_min_profitable_iters): Likewise.
2750 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
2751 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
2752 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
2753 (vectorizable_store, free_stmt_vec_infos): Likewise.
2754 (new_stmt_vec_info): Change return type of xcalloc to
2757 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2759 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
2760 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
2761 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
2762 a single_imm_use-based sequence.
2763 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
2765 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2767 * tree-vectorizer.h (vec_info::lookup_def): Declare.
2768 * tree-vectorizer.c (vec_info::lookup_def): New function.
2769 * tree-vect-patterns.c (vect_get_internal_def): Use it.
2770 (vect_widened_op_tree): Likewise.
2771 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
2772 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
2773 (vectorizable_reduction): Likewise.
2774 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
2776 (vect_is_slp_reduction): Update calls accordingly. Use
2777 vec_info::lookup_def.
2778 (vect_is_simple_reduction): Likewise
2779 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
2781 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2783 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
2784 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
2785 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
2787 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
2788 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
2789 (vect_update_vf_for_slp, vect_analyze_loop_operations)
2790 (vect_is_slp_reduction, vectorizable_induction)
2791 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
2792 * tree-vect-patterns.c (vect_init_pattern_stmt):
2793 (vect_determine_min_output_precision_1, vect_determine_precisions)
2794 (vect_pattern_recog): Likewise.
2795 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
2796 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
2797 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
2798 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
2799 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
2800 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
2801 info field from a loop to a loop_vec_info.
2803 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2805 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
2806 (vec_info::add_stmt): Declare.
2807 * tree-vectorizer.c (vec_info::add_stmt): New function.
2808 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
2809 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
2810 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
2811 (vectorizable_induction): Likewise.
2812 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
2813 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
2814 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
2815 (vectorizable_load): Likewise.
2816 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
2817 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
2818 (vect_recog_gather_scatter_pattern): Likewise.
2819 (append_pattern_def_seq): Likewise. Remove a check that is
2820 performed by add_stmt itself.
2822 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2824 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
2825 which make_ssa_name was called with new_stmt before new_stmt
2828 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2830 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
2832 (vect_is_slp_reduction): ...here...
2833 (vect_is_simple_reduction): ...and here. Remove repetition of tests
2834 that are already known to be false.
2836 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2838 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
2839 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
2840 STMT_VINFO_NUM_SLP_USES when it's true.
2841 (vect_free_slp_instance): Add a final_p parameter and pass it to
2843 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
2844 (vect_analyze_slp_instance): Likewise.
2845 (vect_slp_analyze_operations): Likewise.
2846 (vect_slp_analyze_bb_1): Likewise.
2847 * tree-vectorizer.c (vec_info): Likewise.
2848 * tree-vect-loop.c (vect_transform_loop): Likewise.
2850 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2852 * tree-vect-loop.c (vectorizable_reduction): Assert that the
2853 function is not called for second and subsequent members of
2856 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2858 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
2859 cases for nested loops from here to ...
2860 (vect_create_epilog_for_reduction): ...here. Only call
2861 vect_is_simple_use for inner-loop reductions.
2863 2018-07-31 Martin Liska <mliska@suse.cz>
2865 PR gcov-profile/85338
2866 PR gcov-profile/85350
2867 PR gcov-profile/85372
2868 * profile.c (struct location_triplet): New.
2869 (struct location_triplet_hash): Likewise.
2870 (output_location): Do not output a BB that
2871 is already recorded for a line.
2872 (branch_prob): Use streamed_locations.
2874 2018-07-31 Martin Liska <mliska@suse.cz>
2876 PR gcov-profile/85370
2877 * coverage.c (coverage_begin_function): Do not mark target
2878 clones as artificial functions.
2880 2018-07-31 Martin Liska <mliska@suse.cz>
2882 PR gcov-profile/83813
2883 PR gcov-profile/84758
2884 PR gcov-profile/85217
2885 PR gcov-profile/85332
2886 * profile.c (branch_prob): Do not record GOTO expressions
2887 for GIMPLE statements which locations are already streamed.
2889 2018-07-31 Olivier Hainque <hainque@adacore.com>
2891 * gcc.c (handle_spec_function): Accept a soft_matched_part
2892 argument, as do_spec_1. Pass it down to ...
2893 (eval_spec_function): Accept a soft_matched_part argument,
2894 and pass it down to ...
2895 (do_spec_2): Accept a soft_matched_part argument, and pass
2896 it down to do_spec_1.
2897 (do_spec_1): Pass soft_matched_part to handle_spec_function.
2898 (handle_braces): Update call to handle_spec_function.
2899 (driver::set_up_specs): Update calls to do_spec_2.
2900 (compare_debug_dump_opt_spec_function): Likewise.
2901 (compare_debug_self_opt_spec_function): Likewise.
2903 2018-07-31 Olivier Hainque <hainque@adacore.com>
2905 * common.opt (nolibc): New option.
2906 * doc/invoke.texi (Link Options): Document it.
2907 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
2908 * config/alpha/linux.h: Likewise.
2909 * config/arc/elf.h: Likewise.
2910 * config/arm/uclinux-elf.h: Likewise.
2911 * config/arm/unknown-elf.h: Likewise.
2912 * config/avr/avrlibc.h: Likewise.
2913 * config/bfin/bfin.h: Likewise.
2914 * config/bfin/linux.h: Likewise.
2915 * config/bfin/uclinux.h: Likewise.
2916 * config/darwin.h: Likewise.
2917 * config/darwin10.h: Likewise.
2918 * config/darwin12.h: Likewise.
2919 * config/gnu-user.h: Likewise.
2920 * config/lm32/uclinux-elf.h: Likewise.
2921 * config/pa/pa-hpux11.h: Likewise.
2922 * config/pa/pa64-hpux.h: Likewise.
2923 * config/sparc/sparc.h: Likewise.
2925 2018-07-31 Olivier Hainque <hainque@adacore.com>
2927 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
2928 undefined variables.
2930 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
2933 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
2936 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
2938 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
2939 terminated string literal.
2941 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
2943 PR rtl-optimization/85160
2944 * combine.c (is_just_move): New function.
2945 (try_combine): Allow combining two instructions into two if neither of
2946 the original instructions was a move.
2948 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
2951 * doc/extend.texi (Global Register Variables): Discourage use of type
2953 (Local Register Variables): Likewise.
2955 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
2957 PR tree-optimization/86506
2958 * hwint.h (ceil_log2): Resync with hwint.c implementation.
2960 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
2963 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
2964 hard_regno, make sure no insn between `from` and `to` clobbers it.
2966 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
2967 Tom de Vries <tdevries@suse.de>
2969 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
2970 (PTX_DEFAULT_RUNTIME_DIM): ... this.
2971 (nvptx_goacc_validate_dims): Set default worker and gang dims to
2972 PTX_DEFAULT_RUNTIME_DIM.
2973 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
2975 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
2977 * config/pa/pa.c (pa_output_addr_vec): Align address table.
2978 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
2979 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
2981 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
2983 * config/rs6000/constraints.md (wG constraint): Delete, no longer
2985 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
2986 predicate to reflect toc fusion has been deleted.
2987 (toc_fusion_mem_raw): Delete, no longer used.
2988 (toc_fusion_mem_wrapped): Likewise.
2989 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
2991 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
2992 Delete, no longer used.
2993 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
2994 meant to be used for toc fusion.
2995 (rs6000_debug_print_mode): Delete toc fusion debugging.
2996 (rs6000_debug_reg_global): Likewise.
2997 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
2998 fusion and secondary reload support that were never used.
2999 (rs6000_option_override_internal): Delete TOC fusion, that was only
3000 partially defined, and it did not work unless you also used the
3002 (rs6000_legitimate_address_p): Delete TOC fusion support.
3003 (rs6000_opt_masks): Likewise.
3004 (fusion_wrap_memory_address): Delete function, no longer used.
3005 (fusion_split_address); Delete TOC fusion support.
3006 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
3007 longer used with toc fusion being deleted.
3008 (TARGET_TOC_FUSION_FP): Likewise.
3009 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
3011 (toc fusion spliter): Delete TOC fusion support.
3012 (toc_fusionload_<mode>): Likewise.
3013 (toc_fusionload_di): Likewise.
3014 (fusion_gpr_load_<mode>): Delete generator function, this insn no
3015 longer needs to be named. Rename predicate to delete TOC fusion.
3016 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
3017 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
3018 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
3019 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
3020 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
3022 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3024 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
3025 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
3026 __int128 in built-in function prototypes.
3027 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
3028 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
3030 2018-07-27 Martin Sebor <msebor@redhat.com>
3032 PR tree-optimization/86696
3033 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
3034 types, including enums.
3035 (handle_char_store): Be prepared for the above function to fail.
3037 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
3039 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
3040 when optimization level is lower than 2 or optimize for size.
3042 2018-07-26 Martin Sebor <msebor@redhat.com>
3044 PR tree-optimization/86043
3045 PR tree-optimization/86042
3046 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
3047 (get_string_cst_length): Rename...
3048 (get_min_string_length): ...to this. Add argument.
3049 (handle_char_store): Extend to handle multi-character stores by
3051 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
3052 * tree.h (initializer_zerop): Add argument.
3054 2018-07-26 Jakub Jelinek <jakub@redhat.com>
3057 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
3058 declare target to variables if they have always,{to,from,tofrom} map
3061 2018-07-26 Martin Liska <mliska@suse.cz>
3064 * lto-wrapper.c: Add linker_output as prefix
3065 for ltrans_output_file.
3067 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
3069 PR rtl-optimization/85805
3070 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
3071 value for hard registers if that was written in the same mode.
3073 2018-07-26 Martin Liska <mliska@suse.cz>
3075 PR gcov-profile/86536
3076 * gcov.c (format_gcov): Use printf format %.*f directly
3077 and do not handle special values.
3079 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3081 * common/config/arc/arc-common.c (arc_option_optimization_table):
3082 Update default optimizations for size.
3084 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3086 * config/arc/arc.md (movsf_insn): Add short instruction selection.
3087 * config/arc/constraints.md (CfZ): New constraint.
3088 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
3089 (subsf3_fpu): Likewise.
3090 (cmpsf_fpu): Likewise.
3091 (cmpsf_fpu_uneq): Likewise.
3093 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3095 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
3097 (arc_is_uncached_mem_p): uncached applies to both the variable and
3100 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3102 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
3105 2018-07-25 David Malcolm <dmalcolm@redhat.com>
3107 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
3108 field "m_scopes" from vec to auto_vec.
3110 2018-07-25 Martin Liska <mliska@suse.cz>
3112 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
3115 2018-07-25 Richard Biener <rguenther@suse.de>
3118 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
3119 special wrt context_die late.
3120 (gen_subprogram_die): Re-use DIEs in local scope.
3122 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
3124 PR tree-optimization/86644
3125 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
3127 2018-07-25 Martin Liska <mliska@suse.cz>
3130 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
3131 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
3133 2018-07-25 Martin Liska <mliska@suse.cz>
3136 * params.def: Explain ASan abbreviation and provide
3137 a documentation link.
3139 2018-07-24 Martin Sebor <msebor@redhat.com>
3141 PR tree-optimization/86622
3142 PR tree-optimization/86532
3143 * builtins.h (string_length): Declare.
3144 * builtins.c (c_strlen): Correct handling of non-constant offsets.
3145 (check_access): Be prepared for non-constant length ranges.
3146 (string_length): Make extern.
3147 * expr.c (string_constant): Only handle the minor non-constant
3148 array index. Use string_constant to compute the length of
3149 a generic string constant.
3151 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
3153 PR tree-optimization/86618
3154 * tree-vect-stmts.c (vectorizable_call): Don't take the address
3155 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
3157 2018-07-24 David Malcolm <dmalcolm@redhat.com>
3159 PR tree-optimization/86636
3160 * json.cc (json::object::set): Fix comment. Add assertions.
3161 (json::array::append): Move here from json.h. Add comment and an
3163 (json::string::string): Likewise.
3164 * json.h (json::array::append): Move to json.cc.
3165 (json::string::string): Likewise.
3166 * optinfo-emit-json.cc
3167 (optrecord_json_writer::impl_location_to_json): Assert that we
3168 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
3169 wrapper around it. Expand the location once, rather than three
3171 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
3172 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
3174 (optrecord_json_writer::optinfo_to_json): Likewise, in four
3175 places. Fix some overlong lines.
3177 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
3179 * config/aarch64/aarch64-simd.md
3180 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
3181 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
3182 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
3183 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
3184 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
3185 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
3186 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
3187 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
3188 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
3190 2018-07-24 Jakub Jelinek <jakub@redhat.com>
3193 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
3194 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
3195 and abs_d == d, do the power of two handling if profitable.
3197 2018-07-24 Richard Biener <rguenther@suse.de>
3199 * match.pd: Add BIT_FIELD_REF canonicalizations.
3201 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3204 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
3206 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3208 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
3209 terminated STRING_CST object.
3211 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3213 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
3215 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
3217 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
3218 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
3219 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
3220 the elements into a register.
3221 (rs6000_split_v4si_init_di_reg): Delete.
3222 (rs6000_split_v4si_init): Delete.
3223 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
3224 (vsx_init_v4si): Rewrite as a define_expand.
3226 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
3228 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
3229 zero_extend argument from memory): New.
3231 2018-07-22 Martin Sebor <msebor@redhat.com>
3234 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
3235 diagnosing calls with unknown arguments unless -Walloca-larger-than
3236 is restricted to less than PTRDIFF_MAX bytes.
3238 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
3240 * doc/gcov.texi (Invoking Gcov): Editorial changes.
3242 2018-07-20 David Malcolm <dmalcolm@redhat.com>
3244 * pretty-print.c (text_info::set_location): Remove redundant
3245 "line_table" parameter from call to rich_location::set_range.
3247 2018-07-20 Martin Sebor <msebor@redhat.com>
3250 * builtins.c (expand_builtin_alloca): Adjust.
3251 * calls.c (alloc_max_size): Simplify.
3252 * cgraphunit.c (cgraph_node::expand): Adjust.
3253 * common.opt (larger_than_size, warn_frame_larger_than): Remove
3255 (frame_larger_than_size): Same.
3256 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
3257 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
3259 * doc/invoke.texi (GCC Command Options): Document option arguments.
3260 Explain byte-size arguments and suffixes.
3261 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
3262 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
3263 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
3264 * doc/options.texi (UInteger): Expand.
3265 (Host_Wide_Int, ByteSize): Document new properties.
3266 * final.c (final_start_function_1): Include sizes in an error message.
3267 * function.c (frame_offset_overflow): Same.
3268 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
3269 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
3270 Diagnose unbounded alloca calls only for limits of less than
3272 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
3273 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
3275 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
3276 only for limits of less than PTRDIFF_MAX.
3277 * langhooks-def.h (lhd_handle_option): Change function argument
3279 * langhooks.c (lhd_handle_option): Same.
3280 * langhooks.h (handle_option): Same.
3281 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
3283 (var_type, var_type_struct): Same.
3284 (var_set): Handle ByteSize flag.
3285 * optc-gen.awk: Add comments to output to ease debugging. Make
3286 use of HOST_WIDE_INT where appropriate.
3287 * opts-gen-save.awk: Use %lx to format unsigned long.
3288 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
3289 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
3290 arguments. Parse bytes-size suffixes.
3291 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
3292 (enum_value_to_arg): Same.
3293 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
3294 (handle_option): Adjust.
3295 (generate_option): Change function argument to HOST_WIDE_INT.
3296 (cmdline_handle_error): Adjust.
3297 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
3298 (set_option): Change function argument to HOST_WIDE_INT.
3299 (option_enabled): Handle cl_host_wide_int.
3300 (get_option_state): Handle CLVC_SIZE.
3301 (control_warning_option): Same.
3302 * opts.c (common_handle_option): Change function argument to
3303 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
3304 OPT_Wvla_larger_than_.
3305 * opts.h (enum cl_var_type): Add an enumerator.
3306 * stor-layout.c (layout_decl): Print a more meaningful warning.
3307 * toplev.c (output_stack_usage): Adjust.
3309 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
3311 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
3312 call to inline_expand_builtin_string_cmp.
3313 (expand_builtin_strcmp): Likewise.
3314 (expand_builtin_strncmp): Likewise.
3315 (inline_string_cmp): Delete the last parameter, change char_type_node
3316 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
3318 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
3319 the inlining expansion on target where the type of the call has same or
3320 narrower precision than unsigned char.
3322 2018-07-20 David Malcolm <dmalcolm@redhat.com>
3324 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
3325 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
3326 * common.opt (fsave-optimization-record): New option.
3327 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
3328 * doc/invoke.texi (-fsave-optimization-record): New option.
3329 * dumpfile.c: Include "optinfo-emit-json.h".
3330 (struct kv_pair): Move to coretypes.h.
3331 (optgroup_options): Make non-static.
3332 (dump_context::end_scope): Call
3333 optimization_records_maybe_pop_dump_scope.
3334 * dumpfile.h (optgroup_options): New decl.
3335 * json.cc: New file.
3337 * optinfo-emit-json.cc: New file.
3338 * optinfo-emit-json.h: New file.
3339 * optinfo.cc: Include "optinfo-emit-json.h".
3340 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
3341 (optinfo_enabled_p): Check optimization_records_enabled_p.
3342 (optinfo_wants_inlining_info_p): Likewise.
3343 * optinfo.h: Update comment.
3344 * profile-count.c (profile_quality_as_string): New function.
3345 * profile-count.h (profile_quality_as_string): New decl.
3346 (profile_count::quality): New accessor.
3347 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
3348 and optinfo_emit_json_cc_tests.
3349 * selftest.h (selftest::json_cc_tests): New decl.
3350 (selftest::optinfo_emit_json_cc_tests): New decl.
3351 * toplev.c: Include "optinfo-emit-json.h".
3352 (compile_file): Call optimization_records_finish.
3353 (do_compile): Call optimization_records_start.
3354 * tree-ssa-live.c: Include optinfo.h.
3355 (remove_unused_scope_block_p): Retain inlining information if
3356 optinfo_wants_inlining_info_p returns true.
3358 2018-07-20 Richard Biener <rguenther@suse.de>
3361 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
3362 to cover -flto-partition=none.
3364 2018-07-20 Martin Liska <mliska@suse.cz>
3366 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
3367 (get_decl_source_range): Remove unused function.
3369 2018-07-20 Richard Biener <rguenther@suse.de>
3371 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
3372 (struct vn_phi_s): Likewise.
3373 (struct vn_reference_s): Likewise.
3374 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
3375 for searching the slot of an entry known to be in the hash itself.
3376 (vn_phi_hasher::equal): Likewise.
3377 (vn_reference_hasher::equal): Likewise.
3378 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
3380 (optimistic_info, current_info): Remove, keeping only valid_info.
3381 (vn_reference_lookup_1): Remove fallback lookup.
3382 (vn_reference_lookup_2): Likewise.
3383 (vn_nary_op_lookup_1): Likewise.
3384 (vn_phi_lookup): Likewise.
3385 (vn_nary_build_or_lookup_1): Make sure to not chain the built
3387 (vn_reference_insert): Adjust, chain the inserted hash element
3388 at last_inserted_ref.
3389 (vn_reference_insert_pieces): Likewise.
3390 (visit_reference_op_call): Likewise.
3391 (vn_nary_op_insert_into): Chain the inserted hash element at
3393 (vn_nary_op_insert_pieces): Adjust.
3394 (vn_nary_op_insert): Likewise.
3395 (vn_nary_op_insert_stmt): Likewise.
3396 (vn_phi_insert): Adjust, chain the inserted hash element at
3398 (process_scc): Remove clearing and copying the optimistic
3399 table. Instead remove elements inserted during an optimistic
3400 iteration from the single table we maintain.
3401 (init_scc_vn): Adjust.
3402 (free_scc_vn): Likewise.
3403 (sccvn_dom_walker::record_cond): Likewise.
3404 (sccvn_dom_walker::after_dom_children): Likewise.
3406 2018-07-19 Martin Sebor <msebor@redhat.com>
3408 PR tree-optimization/84047
3409 PR tree-optimization/83776
3410 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
3411 (check_array_bounds): Call it.
3413 2018-07-19 Martin Sebor <msebor@redhat.com>
3415 * align.h (align_flags): Use member initialization.
3417 2018-07-19 David Malcolm <dmalcolm@redhat.com>
3419 * Makefile.in (OBJS): Add optinfo.o.
3420 * coretypes.h (class symtab_node): New forward decl.
3421 (struct cgraph_node): New forward decl.
3422 (class varpool_node): New forward decl.
3423 * dump-context.h: New file.
3424 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
3426 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
3427 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
3428 (set_alt_dump_file): Likewise.
3429 (dump_context::~dump_context): New dtor.
3430 (dump_gimple_stmt): Move implementation to...
3431 (dump_context::dump_gimple_stmt): ...this new member function.
3432 Add the stmt to any pending optinfo, creating one if need be.
3433 (dump_gimple_stmt_loc): Move implementation to...
3434 (dump_context::dump_gimple_stmt_loc): ...this new member function.
3435 Start a new optinfo and add the stmt to it.
3436 (dump_gimple_expr): Move implementation to...
3437 (dump_context::dump_gimple_expr): ...this new member function.
3438 Add the stmt to any pending optinfo, creating one if need be.
3439 (dump_gimple_expr_loc): Move implementation to...
3440 (dump_context::dump_gimple_expr_loc): ...this new member function.
3441 Start a new optinfo and add the stmt to it.
3442 (dump_generic_expr): Move implementation to...
3443 (dump_context::dump_generic_expr): ...this new member function.
3444 Add the tree to any pending optinfo, creating one if need be.
3445 (dump_generic_expr_loc): Move implementation to...
3446 (dump_context::dump_generic_expr_loc): ...this new member
3447 function. Add the tree to any pending optinfo, creating one if
3449 (dump_printf): Move implementation to...
3450 (dump_context::dump_printf_va): ...this new member function. Add
3451 the text to any pending optinfo, creating one if need be.
3452 (dump_printf_loc): Move implementation to...
3453 (dump_context::dump_printf_loc_va): ...this new member function.
3454 Start a new optinfo and add the stmt to it.
3455 (dump_dec): Move implementation to...
3456 (dump_context::dump_dec): ...this new member function. Add the
3457 value to any pending optinfo, creating one if need be.
3458 (dump_context::dump_symtab_node): New member function.
3459 (dump_context::get_scope_depth): New member function.
3460 (dump_context::begin_scope): New member function.
3461 (dump_context::end_scope): New member function.
3462 (dump_context::ensure_pending_optinfo): New member function.
3463 (dump_context::begin_next_optinfo): New member function.
3464 (dump_context::end_any_optinfo): New member function.
3465 (dump_context::s_current): New global.
3466 (dump_context::s_default): New global.
3467 (dump_scope_depth): Delete global.
3468 (dumpfile_ensure_any_optinfo_are_flushed): New function.
3469 (dump_symtab_node): New function.
3470 (get_dump_scope_depth): Reimplement in terms of dump_context.
3471 (dump_begin_scope): Likewise.
3472 (dump_end_scope): Likewise.
3473 (selftest::temp_dump_context::temp_dump_context): New ctor.
3474 (selftest::temp_dump_context::~temp_dump_context): New dtor.
3475 (selftest::verify_item): New function.
3476 (ASSERT_IS_TEXT): New macro.
3477 (ASSERT_IS_TREE): New macro.
3478 (ASSERT_IS_GIMPLE): New macro.
3479 (selftest::test_capture_of_dump_calls): New test.
3480 (selftest::dumpfile_c_tests): Call it.
3481 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
3482 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
3483 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
3484 descriptive comment.
3485 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
3486 (dump_node, dump_bb): Move these unrelated decls.
3487 (class dump_manager): Add leading comment.
3488 * optinfo.cc: New file.
3489 * optinfo.h: New file.
3491 2018-07-19 Michael Collison <michael.collison@arm.com>
3492 Richard Henderson <rth@redhat.com>
3494 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
3495 (subti): Handle op1 zero.
3496 (subvti4, usub4ti4): New.
3497 (*sub<GPI>3_compare1_imm): New.
3498 (sub<GPI>3_carryinCV): New.
3499 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
3500 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
3502 2018-07-19 Michael Collison <michael.collison@arm.com>
3503 Richard Henderson <rth@redhat.com>
3505 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
3506 (addti3): Create simpler code if low part is already known to be 0.
3507 (addvti4, uaddvti4): New.
3508 (*add<GPI>3_compareC_cconly_imm): New.
3509 (*add<GPI>3_compareC_cconly): New.
3510 (*add<GPI>3_compareC_imm): New.
3511 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
3512 handle constants within this pattern..
3513 (*add<GPI>3_compareV_cconly_imm): New.
3514 (*add<GPI>3_compareV_cconly): New.
3515 (*add<GPI>3_compareV_imm): New.
3516 (add<GPI>3_compareV): New.
3517 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
3518 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
3519 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
3520 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
3521 with 'comparison' operator.
3522 (*add<GPI>3_compareV_cconly_imm): Ditto.
3523 (*add<GPI>3_compareV_cconly): Ditto.
3524 (*add<GPI>3_compareV_imm): Ditto.
3525 (add<GPI>3_compareV): Ditto.
3526 (add<mode>3_carryinC): Ditto.
3527 (*add<mode>3_carryinC_zero): Ditto.
3528 (*add<mode>3_carryinC): Ditto.
3529 (add<mode>3_carryinV): Ditto.
3530 (*add<mode>3_carryinV_zero): Ditto.
3531 (*add<mode>3_carryinV): Ditto.
3533 2018-07-19 Michael Collison <michael.collison@arm.com>
3534 Richard Henderson <rth@redhat.com>
3536 * config/aarch64/aarch64-modes.def (CC_V): New.
3537 * config/aarch64/aarch64-protos.h
3538 (aarch64_addti_scratch_regs): Declare
3539 (aarch64_subvti_scratch_regs): Declare.
3540 (aarch64_expand_subvti): Declare.
3541 (aarch64_gen_unlikely_cbranch): Declare
3542 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
3543 for signed overflow using CC_Vmode.
3544 (aarch64_get_condition_code_1): Handle CC_Vmode.
3545 (aarch64_gen_unlikely_cbranch): New function.
3546 (aarch64_addti_scratch_regs): New function.
3547 (aarch64_subvti_scratch_regs): New function.
3548 (aarch64_expand_subvti): New function.
3550 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
3552 * config/aarch64/aarch64-option-extensions.def: New entry for profile
3554 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
3555 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
3558 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
3561 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
3562 address check not strict.
3564 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
3566 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
3568 (store_pair_lanes<mode>): Likewise.
3569 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
3570 enum value 'ADDR_QUERY_LDP_STP_N'.
3571 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
3572 (aarch64_print_address_internal): Add declaration.
3573 (aarch64_print_ldpstp_address): Remove.
3574 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
3575 (aarch64_print_operand): Change printing of 'y'.
3576 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
3577 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
3578 'true' rather than '1'.
3579 * config/aarch64/constraints.md (Uml): Likewise.
3580 (Uml): Rename to Umn.
3583 2018-07-19 Richard Biener <rguenther@suse.de>
3585 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
3587 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
3588 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
3589 (vn_reference_hasher): Likewise.
3590 (struct vn_tables_s): Remove obstack and alloc-pool members.
3591 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
3592 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
3593 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
3594 (vn_reference_insert_pieces): Likewise.
3595 (alloc_vn_nary_op_noinit): Adjust.
3596 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
3597 (vn_phi_eq): Adjust.
3598 (shared_lookup_phiargs): Remove.
3599 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
3600 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
3601 (visit_reference_op_call): Likewise.
3602 (copy_nary, copy_phi, copy_reference): Remove.
3603 (process_scc): Rewind the obstack when iterating. Do not
3604 copy the elements to valid_info but just move them from one
3605 hashtable to the other.
3606 (allocate_vn_table): Adjust.
3607 (free_vn_table): Likewise.
3608 (init_scc_vn): Likewise.
3609 (free_scc_vn): Likewise.
3611 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
3614 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
3615 indirect_return as function type attribute.
3616 (ix86_attribute_table): Change indirect_return to function
3618 * doc/extend.texi: Update indirect_return attribute.
3620 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
3622 * wide-int.h (widest2_int): New.
3623 * gimple-fold.c (arith_overflowed_p): Use it.
3624 * tree.h (widest2_int_cst): New.
3625 * tree-vrp.c (wide_int_binop_overflow): Rename from
3626 vrp_int_const_binop.
3627 Rewrite to work on trees.
3628 (extract_range_from_multiplicative_op_1): Abstract code to...
3629 (wide_int_range_min_max): ...here.
3630 (wide_int_range_cross_product): ...and here.
3631 (extract_range_from_binary_expr_1): Abstract overflow code to...
3632 (wide_int_range_mult_wrapping): ...here.
3633 * tree-vrp.h (wide_int_range_cross_product): New.
3634 (wide_int_range_mult_wrapping): New.
3636 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
3637 Julia Koval <julia.koval@intel.com>
3639 * config/i386/x86-tune-costs.h (skylake_memcpy,
3640 skylake_memset): Replace rep_prefix with unrolling for size 512.
3642 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
3645 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
3646 comparision with EQ_EXPR in last stmt.
3648 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
3650 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
3651 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
3652 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
3653 previously known as "PowerPC AltiVec Built-in Functions". Move
3654 some material to new subsubsections "PowerPC AltiVec Built-in
3655 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
3657 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
3658 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
3659 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
3660 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
3662 2018-07-18 Richard Biener <rguenther@suse.de>
3664 PR tree-optimization/86557
3665 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
3668 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
3670 * config/s390/s390.c (s390_function_profiler): Generate CFI.
3672 2018-07-17 Jeff Law <law@redhat.com>
3674 * config/arm/arm.c (get_label_padding): Update for recent
3675 changes to label_to_alignment.
3677 PR tree-optimization/86010
3678 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
3680 * config/mips/mips.c (vr4130_align_insns): Update for recent
3681 changes to label_to_alignment.
3683 * config/frv/frv.c (frv_label_align): Update for recent changes
3684 to label_to_alignment.
3686 * config/nios2/nios2.c (nios2_label_align): Update for recent
3687 changes which dropped ALIGN_LABELS_LOG.
3689 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
3691 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
3692 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
3694 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
3696 * config/arc/arc.c (arc_label_align): Use align_labels instead of
3697 deprecated align_labels_log.
3699 2018-07-17 Richard Biener <rguenther@suse.de>
3702 * dwarf2out.c (init_sections_and_labels): Always generate
3703 a debug_line_str_section for early LTO debug.
3704 (dwarf2out_finish): Reset debug_line_str_hash output early.
3705 Bump counter for extra dwarf5 .debug_loc labels to not conflict
3707 (dwarf2out_early_finish): Output debug_line_str.
3709 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
3711 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
3712 index register on z196 or later.
3714 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
3716 * config/s390/s390.c (s390_default_align): Set default function
3718 (s390_override_options_after_change): Call s390_default align.
3719 (s390_option_override_internal): Call s390_default_align.
3720 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
3722 2018-07-17 Jakub Jelinek <jakub@redhat.com>
3725 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
3726 to _looptemp_ clauses, other than the first two.
3728 2018-07-17 Martin Liska <mliska@suse.cz>
3730 * opts.c: Do not enable OPT_falign_* for -Os.
3732 2018-07-17 Martin Liska <mliska@suse.cz>
3734 * align.h (MAX_CODE_ALIGN): New.
3735 (MAX_CODE_ALIGN_VALUE): New.
3736 * common/config/i386/i386-common.c (ix86_handle_option):
3737 (MAX_CODE_ALIGN): Moved to align.h.
3738 * final.c (MAX_CODE_ALIGN): Likewise.
3739 * opts.c (parse_and_check_align_values):
3740 (MAX_CODE_ALIGN): Likewise.
3741 (MAX_CODE_ALIGN_VALUE): Likewise.
3743 2018-07-17 Martin Liska <mliska@suse.cz>
3745 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
3746 in order to fulfil coding style.
3747 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
3748 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
3749 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
3750 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
3751 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
3752 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
3753 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
3754 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
3755 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
3757 2018-07-17 Martin Liska <mliska@suse.cz>
3759 * align.h: New file.
3760 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
3762 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
3763 align_flags of label_to_alignment.
3764 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
3766 * config/m68k/m68k.c: Do not use removed align_labels_value and
3768 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
3769 (LOOP_ALIGN): Likewise.
3770 (LABEL_ALIGN): Likewise.
3771 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
3772 Remove not used macro.
3773 (rs6000_loop_align): Change return type to align_flags.
3774 (rs6000_loop_align_max_skip): Remove.
3775 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
3776 Change return type to align_flags.
3777 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
3778 Remove not used macro.
3779 (rs6000_loop_align): Change return type to align_flags.
3780 (rs6000_loop_align_max_skip): Remove.
3781 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
3782 * config/rx/rx-protos.h (rx_align_for_label): Make it
3784 * config/rx/rx.c (rx_align_for_label): Change return type
3786 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
3788 into align_flags class.
3789 (LABEL_ALIGN): Likewise.
3790 (LOOP_ALIGN): Likewise.
3791 * config/s390/s390.c (s390_label_align): Use align_flags
3793 (s390_asm_output_function_label): Likewise.
3794 * config/sh/sh.c (sh_override_options_after_change):
3795 Use align_flags class directly without macros.
3796 (find_barrier): Likewise.
3797 (barrier_align): Likewise.
3798 (sh_loop_align): Likewise.
3799 * config/spu/spu.c (spu_option_override):
3800 Use align_flags_tuple::get_value instead of removed macros.
3801 (spu_sched_init): Likewise.
3802 * config/spu/spu.h (GTY): Likewise.
3803 * config/visium/visium.c (visium_option_override):
3804 Set "8" as default secondary alignment.
3805 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
3806 in order to guarantee secondary alignment of 8.
3807 * coretypes.h: Include align.h header file.
3808 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
3809 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
3810 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
3811 * doc/tm.texi.in: Likewise.
3812 * final.c (struct label_alignment): Remove not used structure.
3813 (LABEL_ALIGN): Change type to align_flags.
3814 (LOOP_ALIGN): Likewise.
3815 (JUMP_ALIGN): Likewise.
3816 (default_loop_align_max_skip): Remove.
3817 (default_label_align_max_skip): Likewise.
3818 (default_jump_align_max_skip): Likewise.
3819 (default_label_align_after_barrier_max_skip):
3820 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
3821 (LABEL_TO_MAX_SKIP): Remove.
3822 (label_to_alignment): Return align_flags type instead of integer.
3823 (label_to_max_skip): Remove.
3824 (align_fuzz): Use align_flags type.
3825 (compute_alignments): Use align_flags type and use align_flags::max
3826 to combine multiple alignments.
3827 (grow_label_align): Grow vec instead of C array.
3828 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
3829 (shorten_branches): Use align_flags type and use align_flags::max
3830 to combine multiple alignments.
3831 (final_scan_insn_1): Remove usage of secondary alignment that comes
3832 from label alignment, but instead use proper secondary alignment
3833 which is computed in grow_label_align.
3834 * flags.h (struct align_flags_tuple): Move to align.h.
3835 (struct align_flags): Likewise.
3836 (state_align_loops): Rename to align_loops.
3837 (state_align_jumps): Rename to align_jumps.
3838 (state_align_labels): Rename to align_labels.
3839 (state_align_functions): Rename to align_functions.
3840 (align_loops_log): Remove.
3841 (align_jumps_log): Remove.
3842 (align_labels_log): Remove.
3843 (align_functions_log): Remove.
3844 (align_loops_max_skip): Remove.
3845 (align_jumps_max_skip): Remove.
3846 (align_labels_max_skip): Remove.
3847 (align_functions_max_skip): Remove.
3848 (align_loops_value): Remove.
3849 (align_jumps_value): Remove.
3850 (align_labels_value): Remove.
3851 (align_functions_value): Remove.
3852 * output.h (label_to_alignment): Change return type to align_flags.
3853 (label_to_max_skip): Remove.
3854 * target.def: Remove loop_align_max_skip, label_align_max_skip,
3855 jump_align_max_skip macros.
3856 * targhooks.h (default_loop_align_max_skip): Remove.
3857 (default_label_align_max_skip): Likewise.
3858 (default_jump_align_max_skip): Likewise.
3859 (default_label_align_after_barrier_max_skip): Remove.
3860 * toplev.c (read_log_maxskip): Use ::normalize function.
3861 (parse_N_M): Remove not used argument and also call ::normalize.
3862 (parse_alignment_opts): Do not pass unused arguments.
3863 * varasm.c (assemble_start_function): Use directly align_functions
3864 instead of removed macros.
3865 * system.h: Do not poison removed macros.
3867 2018-07-17 Jakub Jelinek <jakub@redhat.com>
3870 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
3871 and cond temporaries don't have reference type if iterator has
3872 pointer type. For init use &for_pre_body instead of pre_p if
3873 for_pre_body is non-empty.
3875 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
3877 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
3878 double-double modes to SFmode directly directly.
3879 (trunc<mode>sf2_fprs): Delete.
3881 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
3883 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
3884 for conversions between IFmode and the decimal floating point modes.
3885 (init_float128_ieee): Use the correct names for conversions between
3886 KFmode and the decimal floating point modes.
3888 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
3890 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
3891 for the conversions between TDmode and IFmode.
3892 (init_float128_ieee): Use more correct names for the conversions
3893 between TDmode and KFmode.
3895 2018-07-16 Jakub Jelinek <jakub@redhat.com>
3897 PR tree-optimization/86526
3898 * builtins.c (expand_builtin_memcmp): Formatting fixes.
3899 (inline_expand_builtin_string_cmp): Likewise.
3900 (inline_string_cmp): Likewise. Use c_readstr instead of
3901 builtin_memcpy_read_str. Add unit_mode temporary.
3903 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
3906 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
3907 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
3909 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
3911 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
3912 Alphabetize prototypes of built-in functions, separating out
3913 built-in functions that are listed in this section but should be
3914 described elsewhere.
3916 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
3919 * expmed.c (emit_store_flag): Do not emit setcc followed by a
3920 conditional move when trapping comparison was split to a
3921 non-trapping one (and vice versa).
3923 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
3925 * config/s390/s390.c (s390_function_profiler): Generate nops
3926 instead of profiler call sequences.
3927 * config/s390/s390.opt: Add the new option.
3929 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
3931 * config/s390/s390.c (s390_function_profiler): Generate
3932 __mcount_loc section.
3933 * config/s390/s390.opt: Add the new option.
3935 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
3937 * common.opt: Add the new warning.
3938 * config/s390/s390.c (s390_function_profiler): Emit "brasl
3939 %r0,__fentry__" when -mfentry is specified.
3940 (s390_option_override_internal): Disallow -mfentry for 31-bit
3942 * config/s390/s390.opt: Add the new option.
3944 2018-07-16 Richard Biener <rguenther@suse.de>
3947 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
3948 for function-local FUNCTION_DECL and RESULT_DECL immediately.
3950 2018-07-16 Martin Liska <mliska@suse.cz>
3953 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
3956 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
3958 * config/arc/arcHS.md: Update ARCHS scheduling rules.
3960 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
3962 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
3964 * config/arc/arc-cpus.def (hs4x): New cpu.
3966 * config/arc/arc-tables.opt: Regenerate.
3967 * config/arc/arc.c (arc_sched_issue_rate): New function.
3968 (TARGET_SCHED_ISSUE_RATE): Define.
3969 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
3970 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
3972 (attr tune): Add ARCHS4x tune values.
3973 (attr tune_dspmpy): Define.
3974 (*tst): Correct instruction type.
3975 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
3976 * config/arc/arcHS4x.md: New file.
3977 * config/arc/fpu.md: Update instruction type attributes.
3978 * config/arc/t-multilib: Regenerate.
3980 2018-07-16 Tom de Vries <tdevries@suse.de>
3983 * var-tracking.c (vt_initialize): Fix pre_dec handling.
3985 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3987 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
3990 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
3992 PR tree-optimization/86514
3993 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
3994 conversion to a boolean type from a type with greater precision.
3996 2018-07-16 Tom de Vries <tdevries@suse.de>
3998 * var-tracking.c (vt_initialize): Print adjusted insn slim if
3999 dump_flags request TDF_SLIM.
4001 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
4003 * fold-const.c (int_const_binop_1): Abstract...
4004 (wide_int_binop): ...wide int code here.
4005 (poly_int_binop): ...poly int code here.
4006 Abstract the rest of int_const_binop_1 into int_const_binop.
4007 * fold-const.h (wide_int_binop): New.
4008 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
4009 Remove useless PLUS/MINUS_EXPR case.
4010 (zero_nonzero_bits_from_vr): Move wide int code...
4011 (zero_nonzero_bits_from_bounds): ...here.
4012 (extract_range_from_binary_expr_1): Move mask optimization code...
4013 (range_easy_mask_min_max): ...here.
4014 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
4015 (range_easy_mask_min_max): New.
4017 2018-07-15 Jeff Law <law@redhat.com>
4020 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
4023 2018-07-14 Jim Wilson <jimw@sifive.com>
4025 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
4027 2018-07-14 Paul Koning <ni1d@arrl.net>
4029 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
4031 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
4033 * lto-streamer-out.c (copy_function_or_variable): Dump info about
4036 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
4037 Steve Munroe <munroesj52@gmail.com>
4039 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
4040 (_mm_andnot_si128): Likewise.
4041 (_mm_or_si128): Likewise.
4042 (_mm_xor_si128): Likewise.
4044 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
4047 * builtins.c (expand_builtin_memcmp): Inline the calls first
4048 when result_eq is false.
4049 (expand_builtin_strcmp): Inline the calls first.
4050 (expand_builtin_strncmp): Likewise.
4051 (inline_string_cmp): New routine. Expand a string compare
4052 call by using a sequence of char comparison.
4053 (inline_expand_builtin_string_cmp): New routine. Inline expansion
4054 a call to str(n)cmp/memcmp.
4055 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
4057 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
4059 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
4061 * config/arm/driver-arm.c: Include arm-native.h.
4062 (host_detect_local_cpu): Use auto-generated data tables.
4063 (vendors, arm_cpu_table): Delete. Move part information to ...
4064 * config/arm/arm-cpus.in: ... here.
4065 * config/arm/parsecpu.awk (gen_native): New function.
4066 (vendor, part): New CPU fields.
4067 (END): Add support for building the native CPU detection tables.
4068 * config/arm/t-arm (arm-native.h): Add build rule.
4069 (driver-arm.o): Add dependency on arm-native.h.
4071 2018-07-13 Richard Biener <rguenther@suse.de>
4074 * match.pd (addr1 - addr2): Allow either of the operand to
4077 2018-07-13 Tom de Vries <tdevries@suse.de>
4079 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
4082 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
4084 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
4085 arrays instead of numbered variables.
4087 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
4089 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
4090 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
4092 2018-07-13 Richard Biener <rguenther@suse.de>
4095 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
4096 instead of get_context_die.
4098 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4099 Richard Biener <rguenther@suse.de>
4102 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
4103 that the loop latch destination where phi is defined.
4105 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
4107 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
4108 (riscv_expand_epilogue): Add assertion to check interrupt mode.
4109 (riscv_set_current_function): Extract getting interrupt type to new
4111 (riscv_get_interrupt_type): New function.
4112 (riscv_merge_decl_attributes): New function, checking interrupt type is
4114 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
4116 2018-07-12 Paul Koning <ni1d@arrl.net>
4118 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
4121 2018-07-12 Paul Koning <ni1d@arrl.net>
4123 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
4124 zero reference, add doloop_end instead.
4125 * doc/md.texi (decrement_and_branch_until_zero): Remove.
4126 (Looping patterns): Remove decrement_and_branch_until_zero. Add
4127 detail for doloop_end.
4129 2018-07-12 Martin Sebor <msebor@redhat.com>
4132 * attribs.c (decl_attributes): Reject conflicting attributes before
4133 calling attribute handlers.
4135 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
4137 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
4139 (gcc::dump_manager::get_dump_file_name): likewise.
4140 (dump_begin): Likewise.
4141 * dumpfile.h (dump_begin): Update prototype.
4142 (gcc::dump_manager::get_dump_file_name,
4143 gcc::dump_manager::get_dump_file_name): Update prototype.
4145 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4147 * internal-fn.h (vectorizable_internal_fn_p): New function.
4148 * tree-vect-slp.c (compatible_calls_p): Likewise.
4149 (vect_build_slp_tree_1): Remove nops argument. Handle calls
4150 to internal functions.
4151 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
4153 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4155 * fold-const.h (inverse_conditions_p): Declare.
4156 * fold-const.c (inverse_conditions_p): New function.
4157 * match.pd: Use inverse_conditions_p. Add folds of view_converts
4158 that test the inverse condition of a conditional internal function.
4159 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
4160 * internal-fn.c (internal_fn_mask_index): Handle conditional
4162 (vectorized_internal_fn_supported_p): New function.
4163 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
4164 (any_pred_load_store): Replace with...
4165 (need_to_predicate): ...this new variable.
4166 (redundant_ssa_names): New variable.
4167 (ifcvt_can_use_mask_load_store): Move initial checks to...
4168 (ifcvt_can_predicate): ...this new function. Handle tree codes
4169 for which a conditional internal function exists.
4170 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
4171 instead of ifcvt_can_use_mask_load_store. Update after variable
4173 (predicate_load_or_store): New function, split out from
4174 predicate_mem_writes.
4175 (check_redundant_cond_expr): New function.
4176 (value_available_p): Likewise.
4177 (predicate_rhs_code): Likewise.
4178 (predicate_mem_writes): Rename to...
4179 (predicate_statements): ...this. Use predicate_load_or_store
4180 and predicate_rhs_code.
4181 (combine_blocks, tree_if_conversion): Update after above name changes.
4182 (ifcvt_local_dce): Handle redundant_ssa_names.
4183 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
4184 general conditional functions.
4185 * tree-vect-stmts.c (vectorizable_call): Likewise.
4187 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4188 Alan Hayward <alan.hayward@arm.com>
4189 David Sherwood <david.sherwood@arm.com>
4191 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
4192 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
4193 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
4194 plus and minus and convert them into IFN_COND_FMA-based sequences.
4195 (convert_mult_to_fma): Handle conditional plus and minus.
4197 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4199 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
4200 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
4201 (cond_fnms_optab): New optabs.
4202 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
4204 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
4205 * internal-fn.h (get_conditional_internal_fn): Declare.
4206 (get_unconditional_internal_fn): Likewise.
4207 * internal-fn.c (cond_ternary_direct): New macro.
4208 (expand_cond_ternary_optab_fn): Likewise.
4209 (direct_cond_ternary_optab_supported_p): Likewise.
4210 (FOR_EACH_COND_FN_PAIR): Likewise.
4211 (get_conditional_internal_fn): New function.
4212 (get_unconditional_internal_fn): Likewise.
4213 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
4214 (gimple_match_op::gimple_match_op): Add a new overload for 5
4216 (gimple_match_op::set_op): Likewise.
4217 (gimple_resimplify5): Declare.
4218 * genmatch.c (decision_tree::gen): Generate simplifications for
4220 * gimple-match-head.c (gimple_simplify): Define an overload for
4221 5 operands. Handle calls with 5 arguments in the top-level overload.
4222 (convert_conditional_op): Handle conversions from unconditional
4223 internal functions to conditional ones.
4224 (gimple_resimplify5): New function.
4225 (build_call_internal): Pass a fifth operand.
4226 (maybe_push_res_to_seq): Likewise.
4227 (try_conditional_simplification): Try converting conditional
4228 internal functions to unconditional internal functions.
4229 Handle 3-operand unconditional forms.
4230 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
4231 Define ternary equivalents of the current rules for binary conditional
4233 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
4235 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
4236 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
4237 (optab): Handle them.
4238 (SVE_COND_FP_TERNARY): New int iterator.
4239 (sve_fmla_op, sve_fmad_op): New int attributes.
4240 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
4241 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
4242 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
4244 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4246 * target.def (preferred_else_value): New target hook.
4247 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
4248 * doc/tm.texi: Regenerate.
4249 * targhooks.h (default_preferred_else_value): Declare.
4250 * targhooks.c (default_preferred_else_value): New function.
4251 * internal-fn.h (conditional_internal_fn_code): Declare.
4252 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
4253 (get_conditional_internal_fn): Use it.
4254 (conditional_internal_fn_code): New function.
4255 * gimple-match.h (gimple_match_cond): New struct.
4256 (gimple_match_op): Add a cond member function.
4257 (gimple_match_op::gimple_match_op): Update all forms to take a
4259 * genmatch.c (expr::gen_transform): Use the same condition as res_op
4260 for the suboperation, but don't specify a particular else_value.
4261 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
4262 (visit_nary_op, visit_reference_op_load): Pass
4263 gimple_match_cond::UNCOND to the gimple_match_op constructor.
4264 * gimple-match-head.c: Include tree-eh.h
4265 (convert_conditional_op): New function.
4266 (maybe_resimplify_conditional_op): Likewise.
4267 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
4268 (gimple_resimplify2): Likewise.
4269 (gimple_resimplify3): Likewise.
4270 (gimple_resimplify4): Likewise.
4271 (maybe_push_res_to_seq): Return null for conditional operations.
4272 (try_conditional_simplification): New function.
4273 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
4275 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
4277 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
4279 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
4281 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
4283 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
4285 (hash_tree): Do not hash DECL_FCONTEXT
4286 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
4287 Do not stream DECL_FCONTEXT.
4288 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
4289 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
4291 2018-07-12 Richard Biener <rguenther@suse.de>
4294 * dwarf2out.c (gen_block_die): Only output blocks when they have
4295 at least one !DECL_IGNORED_P variable.
4297 2018-07-12 Richard Biener <rguenther@suse.de>
4300 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
4301 Remove -mieee-fp handling.
4303 2018-07-12 Richard Biener <rguenther@suse.de>
4305 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
4306 left-over from last patch.
4308 2018-07-12 Jakub Jelinek <jakub@redhat.com>
4310 PR tree-optimization/86492
4311 * gimple-ssa-store-merging.c
4312 (imm_store_chain_info::coalesce_immediate_stores): Call
4313 check_no_overlap even for the merge_overlapping case. Formatting fix.
4315 2018-07-12 Richard Biener <rguenther@suse.de>
4318 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
4319 move possibly trapping operations into the conditional.
4321 2018-07-12 Richard Biener <rguenther@suse.de>
4323 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
4324 (vn_lookup_simplify_result): Remove recursion limit applied
4326 (vn_nary_build_or_lookup_1): Adjust.
4327 (try_to_simplify): Likewise.
4328 * gimple-match-head.c (gimple_resimplify1): Instead apply one
4330 (gimple_resimplify2): Likewise.
4331 (gimple_resimplify3): Likewise.
4332 (gimple_resimplify4): Likewise.
4334 2018-07-11 Jakub Jelinek <jakub@redhat.com>
4336 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
4337 Use __mmask64 type instead of __mmask8 for __M argument.
4338 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
4339 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
4341 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
4342 __mmask16 for __M argument.
4343 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
4344 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
4345 to __mmask16 instead of __mmask8.
4346 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
4347 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
4348 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
4349 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
4350 instead of __mmask16 for __U argument.
4351 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
4352 __mmask16 instead of __mmask8 for __U argument.
4353 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
4355 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
4357 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
4359 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
4361 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
4363 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
4365 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
4367 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
4369 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
4371 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
4373 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
4374 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
4375 return type as well as __M argument type and all casts from __mmask8
4377 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
4378 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
4379 return type as well as __M argument type and all casts from __mmask8
4381 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
4382 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
4383 return type as well as __M argument type and all casts from __mmask8
4385 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
4386 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
4387 return type as well as __M argument type and all casts from __mmask8
4389 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
4390 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
4393 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
4395 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
4396 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
4399 2018-07-11 Paul Koning <ni1d@arrl.net>
4401 * doc/md.texi (define_subst): Document how multiple occurrences of
4402 the same argument in the replacement pattern are handled.
4404 2018-07-11 Paul Koning <ni1d@arrl.net>
4406 * doc/extend.texi (Common Variable Attributes): Move "mode" into
4408 (Common Type Attributes): Add "mode" attribute.
4410 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
4412 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
4413 stream DECL_ORIGINAL_TYPE.
4414 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
4415 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
4416 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
4417 Do not walk original type.
4418 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
4420 (write_ts_decl_non_common_tree_pointers): Do not stream
4422 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
4423 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
4425 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
4427 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
4428 thread twice from the same starting edge.
4430 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
4432 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
4433 * gimple.c (gimple_call_nonnull_result_p): ...here...
4434 (gimple_call_nonnull_arg): ...and here.
4435 * gimple.h (gimple_call_nonnull_result_p): New.
4436 (gimple_call_nonnull_arg): New.
4438 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
4440 * config/arm/arm-cpus.in: Move information from fpu field of each
4441 cpu definition to the isa field.
4442 * config/arm/parsecpu.awk (fpu): Delete match rule.
4443 (gen_comm_data): Don't add bits from the CPU's FPU entry.
4445 2018-07-11 Richard Biener <rguenther@suse.de>
4448 * dwarf2out.c (init_sections_and_labels): Use
4449 output_asm_line_debug_info consistently.
4450 (dwarf2out_early_finish): Likewise.
4451 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
4454 2018-07-11 Richard Biener <rguenther@suse.de>
4456 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
4457 Rework father_bb setting in a way to avoid propagating constants
4458 multiple times on a loop body.
4460 2018-07-10 Mark Wielaard <mark@klomp.org>
4463 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
4465 2018-07-10 Richard Biener <rguenther@suse.de>
4467 * hash-map.h (hash_map::iterator::operator*): Return
4468 references to key and value.
4470 2018-07-10 Jakub Jelinek <jakub@redhat.com>
4473 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
4474 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
4475 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
4476 (gimplify_omp_for): For composite loops, move outer
4477 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
4478 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
4479 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
4480 TREE_LIST for both the original class iterator and the "last" helper
4481 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
4482 loop, remember has_decl_expr from outer composite loops for the
4483 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
4485 2018-07-09 Martin Sebor <msebor@redhat.com>
4489 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
4490 accessing implicitly initialized array elements.
4491 * expr.c (string_constant): Handle string initializers of
4492 character arrays within aggregates.
4493 * gimple-fold.c (fold_array_ctor_reference): Add argument.
4494 Store element offset. As a special case, handle zero size.
4495 (fold_nonarray_ctor_reference): Same.
4496 (fold_ctor_reference): Add argument. Store subobject offset.
4497 * gimple-fold.h (fold_ctor_reference): Add argument.
4499 2018-07-09 Paul Koning <ni1d@arrl.net>
4501 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
4502 (pdp11_insn_cost): New function.
4503 (pdp11_md_asm_adjust): New function.
4504 (TARGET_INVALID_WITHIN_DOLOOP): Define.
4505 (pdp11_rtx_costs): Update to match machine better.
4506 (output_addr_const_pdp11): Correct format mismatch warnings.
4507 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
4508 * config/pdp11/pdp11.md: General change to add base_cost and/or
4509 length attributes for use by new pdp11_insn_cost function.
4510 (MIN_BRANCH): Correct definition.
4512 (doloop_end): Use standard pattern name for looping pattern.
4513 (doloop_end_nocc): New.
4514 (movsf): Add another constraint alternative.
4515 (zero_extendqihi2): Add constraint alternatives for not in place
4517 (zero_extendhisi2): Remove.
4518 (shift patterns): Add CC handling variants.
4522 (define_peephole2): New peephole to recognize mov that sets CC for
4525 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4527 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
4528 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
4531 2018-07-09 Jakub Jelinek <jakub@redhat.com>
4534 * real.c (real_nextafter): Return true if result is denormal.
4536 2018-07-09 Martin Liska <mliska@suse.cz>
4538 * common.opt: Add back wrongly removed attribute.
4540 2018-07-09 Richard Biener <rguenther@suse.de>
4543 * dwarf2out.c (gen_block_die): For an early generated DIE
4544 always output high/low PC attributes.
4546 2018-07-09 Tom de Vries <tdevries@suse.de>
4548 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
4549 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
4550 onto VAR_DECL with abstract origin.
4552 2018-07-07 Jim Wilson <jimw@sifive.com>
4554 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
4556 2018-07-07 Tom de Vries <tdevries@suse.de>
4558 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
4560 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
4562 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
4564 (combine_bound): Use wide-int overflow calculation instead of
4566 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
4568 * fold-const.c (int_const_binop_2): Same.
4569 (extract_muldiv_1): Same.
4570 (fold_div_compare): Same.
4571 (fold_abs_const): Same.
4573 * poly-int.h (add): Same.
4577 * predict.c (predict_iv_comparison): Same.
4578 * profile-count.c (slow_safe_scale_64bit): Same.
4579 * simplify-rtx.c (simplify_const_binary_operation): Same.
4580 * tree-chrec.c (tree_fold_binomial): Same.
4581 * tree-data-ref.c (split_constant_offset_1): Same.
4582 * tree-if-conv.c (idx_within_array_bound): Same.
4583 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
4584 * tree-ssa-phiopt.c (minmax_replacement): Same.
4585 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
4586 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
4587 * vr-values.c (vr_values::adjust_range_with_scev): Same.
4588 * wide-int.cc (wi::add_large): Same.
4589 (wi::mul_internal): Same.
4590 (wi::sub_large): Same.
4591 (wi::divmod_internal): Same.
4592 * wide-int.h: Change overflow type to overflow_type for neg, add,
4593 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
4594 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
4595 mul_internal, divmod_internal.
4596 (overflow_type): New enum.
4597 (accumulate_overflow): New.
4599 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4601 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
4602 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
4604 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
4606 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
4607 argument is checked for zero before entering loop, avoid checking again.
4609 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
4611 * gimplify.h (generic_expr_could_trap_p): Set as global function.
4612 * gimplify.h (generic_expr_could_trap_p): Likwise.
4613 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
4615 2018-07-06 Jakub Jelinek <jakub@redhat.com>
4617 PR tree-optimization/86401
4618 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
4619 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
4620 (fold_bit_and_mask): ... here. New helper function for match.pd.
4621 * fold-const.h (fold_bit_and_mask): Declare.
4622 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
4624 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
4627 * target.def (translate_mode_attribute): New hook.
4628 * targhooks.h (default_translate_mode_attribute): Declare.
4629 * targhooks.c (default_translate_mode_attribute): New function.
4630 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
4631 * doc/tm.texi: Regenerate.
4632 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
4633 (rs6000_translate_mode_attribute): New function.
4635 2018-07-06 Paul Koning <ni1d@arrl.net>
4637 * doc/md.texi (define_split): Document DONE and FAIL.
4638 (define_peephole2): Ditto.
4640 2018-07-05 Jeff Law <law@redhat.com>
4642 PR tree-optimization/86010
4643 * tree-ssa-dse.c (compute_trims): More aggressively trim at
4644 both the head and tail of mem* and str* calls.
4646 2018-07-05 Jim Wilson <jimw@sifive.com>
4648 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
4650 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
4652 * config/aarch64/aarch64-simd.md: correct flags text for
4653 MIN_EXPR replacement.
4655 2018-07-05 James Clarke <jrtc27@jrtc27.com>
4657 * configure: Regenerated.
4659 2018-07-05 Carl Love <cel@us.ibm.com>
4661 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
4662 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
4663 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
4664 VSX_BUILTIN_DOUBLEL_V4SF.
4666 2018-07-05 Martin Sebor <msebor@redhat.com>
4669 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
4670 than its domain to compute its the upper bound of a char array.
4672 2018-07-05 Nathan Sidwell <nathan@acm.org>
4674 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
4675 * doc/cpp.texi: Update comment.
4676 * doc/tm.texi: Rebuilt.
4677 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
4678 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
4679 * doc/extend.texi (Backwards Compatibility): Clarify it is system
4680 headers affected by extern "C".
4681 * system.h: Poison NO_IMPLICIT_EXTERN_C.
4682 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
4683 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
4684 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
4685 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
4686 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
4687 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
4688 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
4689 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
4690 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
4691 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
4692 config/sparc/sp64-elf.h, config/spu/spu.h,
4693 config/stormy16/stormy16.h, config/v850/v850.h,
4694 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
4695 define NO_IMPLICIT_EXTERN_C.
4696 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
4698 2018-07-05 Tamar Christina <tamar.christina@arm.com>
4701 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
4702 instead of GET_MODE_SIZE when comparing Units.
4704 2018-07-05 Tamar Christina <tamar.christina@arm.com>
4707 * rtlanal.c (set_noop_p): Constrain on mode change,
4708 include hard-reg-set.h
4710 2018-07-05 Tamar Christina <tamar.christina@arm.com>
4712 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
4714 2018-07-05 Jakub Jelinek <jakub@redhat.com>
4717 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
4720 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
4722 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
4724 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
4727 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
4729 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
4731 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
4734 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
4736 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
4738 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
4739 neg_*_op* variables.
4741 2018-07-04 Martin Liska <mliska@suse.cz>
4743 * tree-switch-conversion.c: Define
4744 max_ratio_for_speed and max_ratio_for_size constants.
4746 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
4747 Martin Liska <mliska@suse.cz>
4752 * common.opt: Rename align options with 'str_' prefix.
4753 * common/config/i386/i386-common.c (set_malign_value): New
4755 (ix86_handle_option): Use it to set -falign-* options/
4756 * config/aarch64/aarch64-protos.h (struct tune_params): Change
4757 type from int to string.
4758 * config/aarch64/aarch64.c: Update default values from int
4760 * config/alpha/alpha.c (alpha_override_options_after_change):
4762 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
4763 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4764 max skip conditionally.
4765 * config/i386/freebsd.h (SUBALIGN_LOG): New.
4766 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4767 max skip conditionally.
4768 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4769 max skip conditionally.
4770 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
4771 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4772 max skip conditionally.
4773 * config/i386/i386.c (struct ptt): Change type from int to
4775 (ix86_default_align): Set default values.
4776 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
4777 max skip conditionally.
4778 * config/i386/iamcu.h (SUBALIGN_LOG): New.
4779 (ASM_OUTPUT_MAX_SKIP_ALIGN):
4780 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
4781 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4782 max skip conditionally.
4783 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
4784 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
4785 * config/i386/x86-64.h (SUBALIGN_LOG): New.
4786 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
4787 max skip conditionally.
4788 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
4789 * config/ia64/ia64.c (ia64_option_override): Set default values
4790 for alignment options.
4791 * config/m68k/m68k.c: Handle new str_align_* options.
4792 * config/mips/mips.c (mips_set_compression_mode): Change
4794 (mips_option_override): Set default values for options.
4795 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
4797 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4799 * config/rx/rx.c (rx_option_override): Likewise.
4800 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
4801 (LABEL_ALIGN): Use align_labels_log.
4802 (LOOP_ALIGN): Use align_loops_align.
4803 * config/s390/s390.c (s390_asm_output_function_label): Use new
4805 * config/sh/sh.c (sh_override_options_after_change):
4806 Change type of constants.
4807 * config/spu/spu.c (spu_sched_init): Likewise.
4808 * config/sparc/sparc.c (sparc_option_override): Set default
4810 * config/visium/visium.c (visium_option_override): Likewise.
4811 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
4812 emit p2align format with last argument if it's not needed.
4813 * doc/invoke.texi: Document extended format of -falign-*.
4814 * final.c: Use align_labels alignment.
4815 * flags.h (struct target_flag_state): Change type to use
4817 (struct align_flags_tuple): New.
4818 (struct align_flags): Likewise.
4819 (align_loops_log): Redefine macro to use new types.
4820 (align_loops_max_skip): Redefine macro to use new types.
4821 (align_jumps_log): Redefine macro to use new types.
4822 (align_jumps_max_skip): Redefine macro to use new types.
4823 (align_labels_log): Redefine macro to use new types.
4824 (align_labels_max_skip): Redefine macro to use new types.
4825 (align_functions_log): Redefine macro to use new types.
4826 (align_loops): Redefine macro to use new types.
4827 (align_jumps): Redefine macro to use new types.
4828 (align_labels): Redefine macro to use new types.
4829 (align_functions): Redefine macro to use new types.
4830 (align_functions_max_skip): Redefine macro to use new types.
4831 (align_loops_value): New macro.
4832 (align_jumps_value): New macro.
4833 (align_labels_value): New macro.
4834 (align_functions_value): New macro.
4835 * function.c (invoke_set_current_function_hook): Propagate
4836 alignment values from flags to global variables default in
4838 * ipa-icf.c (sem_function::equals_wpa): Use
4839 cl_optimization_option_eq instead of memcmp.
4840 * lto-streamer.h (cl_optimization_stream_out): Support streaming
4842 (cl_optimization_stream_in): Likewise.
4843 * optc-save-gen.awk: Support strings in cl_optimization.
4844 * opth-gen.awk: Likewise.
4845 * opts.c (finish_options): Remove error checking of invalid
4847 (MAX_CODE_ALIGN): Remove.
4848 (MAX_CODE_ALIGN_VALUE): Likewise.
4849 (parse_and_check_align_values): New function.
4850 (check_alignment_argument): Likewise.
4851 (common_handle_option): Use check_alignment_argument.
4852 * opts.h (parse_and_check_align_values): Declare.
4853 * toplev.c (init_alignments): Remove.
4854 (read_log_maxskip): New.
4855 (parse_N_M): Likewise.
4856 (parse_alignment_opts): Likewise.
4857 (backend_init_target): Remove usage of init_alignments.
4858 * toplev.h (parse_alignment_opts): Declare.
4859 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
4861 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
4862 * tree.c (cl_option_hasher::equal): New.
4863 * varasm.c: Use new global macros.
4865 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
4867 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
4868 Use a simpler align directive also if MAXSKIP = ALIGN-1.
4869 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4870 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4871 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4872 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
4873 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
4874 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
4875 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
4876 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4877 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4878 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4879 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4881 2018-07-04 Martin Liska <mliska@suse.cz>
4882 Jonathan Wakely <jwakely@redhat.com>
4884 * coverage.c: Use correct type.
4885 * doc/invoke.texi: Language correction.
4887 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
4890 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
4891 ENDBRANCH for non-tail call which may return via indirect branch.
4892 * doc/extend.texi: Document indirect_return attribute.
4894 2018-07-03 Martin Sebor <msebor@redhat.com>
4896 PR tree-optimization/86274
4897 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
4899 (format_floating): Correct handling of infinities and NaNs.
4901 2018-07-03 Martin Sebor <msebor@redhat.com>
4903 * print-tree.c (print_real_cst): New function.
4904 (print_node_brief): Call it.
4905 (print_node): Ditto.
4907 2018-07-03 Jeff Law <law@redhat.com>
4909 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
4910 into a single pattern.
4912 * config/h8300/h8300.md (ors code_iterator): New.
4913 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
4914 a single pattern and single splitter.
4915 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
4916 (iorqi3_1, xorqi3_1): Likewise.
4917 (iorqi3, xorqi3 expanders): Similarly.
4919 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
4920 (movmd_internal) into a single pattern using the P mode iterator.
4921 (movmd splitters): Similarly.
4922 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
4923 (movsd splitters): Similarly.
4925 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
4926 ADDB, ADDW and ADDL into a single ADD attribute which selects the
4927 right table based on the size of the operand.
4928 * config/h8300/h8300.md (length_table): Corresponding changes. All
4929 references to "addb", "addw" and "addl" changed to "add".
4930 (btst patterns): Merge two variants into a single pattern.
4931 (tstqi, tsthi): Likewise.
4932 (addhi3_incdec, addsi3_incdec): Likewise.
4933 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
4934 (mulhi3, mulsi3): Likewise.
4935 (udivhi3, udivsi3): Likewise.
4936 (divhi3, divsi3): Likewise.
4937 (andorqi3, andorhi3, andorsi3): Likewise.
4939 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
4942 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
4943 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
4945 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
4947 PR tree-optimization/85694
4948 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
4949 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
4951 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
4952 (<u>avg<mode>3_ceil): New patterns.
4954 2018-07-03 David Malcolm <dmalcolm@redhat.com>
4956 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
4957 scan-tree-dump directive.
4958 * gcc.dg/vect/slp-perm-2.c: Likewise.
4959 * gcc.dg/vect/slp-perm-3.c: Likewise.
4960 * gcc.dg/vect/slp-perm-5.c: Likewise.
4961 * gcc.dg/vect/slp-perm-6.c: Likewise.
4962 * gcc.dg/vect/slp-perm-7.c: Likewise.
4963 * gcc.dg/vect/slp-perm-8.c: Likewise.
4965 2018-07-03 Marek Polacek <polacek@redhat.com>
4968 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
4970 2018-07-03 Richard Biener <rguenther@suse.de>
4973 * tree-ssa-structalias.c (find_func_clobbers): Properly
4974 handle indirect calls.
4976 2018-07-03 Jeff Law <law@redhat.com>
4978 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
4979 (shifts): New code iterator.
4980 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
4981 expander. Fix HImode handling on H8/SX.
4982 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
4983 (subqi3, subhi3, subsi3 expanders): Likewise.
4984 (andqi3, andhi3, andsi3 expanders): Likewise.
4985 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
4986 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
4987 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
4988 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
4989 (zero_extendqihi2, zero_extendqisi2): Likewise.
4990 (extendqihi2, extendqisi2): Likewise.
4991 (rotlqi3, rotlhi3, rotlsi3): Likewise.
4992 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
4993 (rotlqi3_1, rotlhi3_1): Likewise.
4994 (logicalhi3_sn, logicalsi3_sn): Likewise.
4995 (logicalhi3, logicalsi3): Likewise.
4997 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
4999 * tree-vect-patterns.c (vect_recog_rotate_pattern)
5000 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
5001 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
5002 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
5003 type to append_pattern_def_seq instead of creating a stmt_vec_info
5005 (build_mask_conversion): Likewise. Remove vinfo argument.
5006 (vect_add_conversion_to_patterm): Likewise, renaming to...
5007 (vect_add_conversion_to_pattern): ...this.
5008 (vect_recog_mask_conversion_pattern): Update call to
5009 build_mask_conversion. Pass the vector type to
5010 append_pattern_def_seq here too.
5011 (vect_recog_gather_scatter_pattern): Update call to
5012 vect_add_conversion_to_pattern.
5014 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5016 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
5017 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
5018 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
5019 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
5020 STMT_VINFO_PATTERN_DEF_SEQ to null here.
5021 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
5022 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
5023 append_pattern_def_seq instead of new_pattern_def_seq.
5024 (vect_recog_divmod_pattern): Do both of the above.
5025 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
5028 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5030 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
5031 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
5032 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
5033 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
5034 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
5035 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
5036 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
5037 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
5038 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
5039 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
5040 parameter with a single stmt_vec_info.
5041 (vect_recog_func_ptr): Likewise.
5042 (vect_recog_gather_scatter_pattern): Likewise, folding in...
5043 (vect_try_gather_scatter_pattern): ...this.
5044 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
5045 the stmt_vec_info of the statement to be matched. Don't clear
5046 STMT_VINFO_RELATED_STMT.
5047 (vect_pattern_recog): Update call accordingly.
5049 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5051 PR tree-optimization/85694
5052 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
5053 (uavgM3_ceil): Document new optabs.
5054 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
5055 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
5057 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
5058 (savg_ceil_optab): New optabs.
5059 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
5060 (vect_vect_recog_func_ptrs): Add it.
5061 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
5062 constant directly from the associated lhs.
5064 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5066 * tree-vect-patterns.c (vect_split_statement): New function.
5067 (vect_convert_input): Use it to try to split an existing cast.
5069 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5071 * poly-int.h (print_hex): New function.
5072 * dumpfile.h (dump_dec, dump_hex): Declare.
5073 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
5074 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
5075 min_input_precision, operation_precision and operation_sign.
5076 * tree-vect-patterns.c (vect_get_range_info): New function.
5077 (vect_same_loop_or_bb_p, vect_single_imm_use)
5078 (vect_operation_fits_smaller_type): Delete.
5079 (vect_look_through_possible_promotion): Add an optional
5080 single_use_p parameter.
5081 (vect_recog_over_widening_pattern): Rewrite to use new
5082 stmt_vec_info infomration. Handle one operation at a time.
5083 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
5084 (vect_truncatable_operation_p, vect_set_operation_type)
5085 (vect_set_min_input_precision): New functions.
5086 (vect_determine_min_output_precision_1): Likewise.
5087 (vect_determine_min_output_precision): Likewise.
5088 (vect_determine_precisions_from_range): Likewise.
5089 (vect_determine_precisions_from_users): Likewise.
5090 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
5091 (vect_vect_recog_func_ptrs): Put over_widening first.
5093 (vect_pattern_recog): Call vect_determine_precisions.
5095 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5097 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
5098 statements that have been replaced by further pattern statements.
5099 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
5101 2018-07-03 Richard Biener <rguenther@suse.de>
5103 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
5104 always set *dt. Dump vectype in vectype overload.
5105 * dumpfile.h (dump_gimple_expr): New function.
5106 (dump_gimple_expr_loc): Likewise.
5107 * dumpfile.c (dump_gimple_expr): New function.
5108 (dump_gimple_expr_loc): Likewise.
5110 2018-07-02 Jeff Law <law@redhat.com>
5112 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
5113 the H8/300, H8/300H and H8/S variants into a single pattern.
5114 (movhi_h8300, movqi_h8300hs): Similarly.
5115 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
5116 (QHI mode iterator): New.
5118 * config/h8300/h8300.md: Remove trailing whitespace.
5120 2018-07-02 Jim Wilson <jimw@sifive.com>
5122 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
5123 instead of emit_insn for interrupt returns.
5124 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
5125 (riscv_sret, riscv_uret): Likewise.
5127 2018-07-02 David Malcolm <dmalcolm@redhat.com>
5129 * pretty-print.c (selftest::test_pp_format): Move save and restore
5130 of quotes to class auto_fix_quotes, and add an instance.
5131 * selftest.c: Include "intl.h".
5132 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
5133 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
5134 * selftest.h (selftest::auto_fix_quotes): New class.
5136 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
5138 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
5139 (aarch64_sve_prepare_conditional_op): Remove.
5140 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
5141 Allow aarch64_simd_reg_or_zero as select operand; remove
5142 the aarch64_sve_prepare_conditional_op call.
5143 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
5144 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
5145 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
5146 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
5147 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
5148 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
5149 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
5150 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
5151 and a splitters to match all of the *_any patterns.
5152 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
5154 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
5155 (SVE_COND_FP_BINARY_REV): Remove.
5156 (sve_int_op_rev, sve_fp_op_rev): New.
5157 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
5158 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
5159 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
5160 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
5161 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
5162 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
5163 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
5164 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
5165 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
5167 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
5168 Remove match_dup 1 from the inner unspec.
5169 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
5171 * config/aarch64/aarch64.md (movprfx): New attr.
5172 (length): Default movprfx to 8.
5173 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
5174 (*madd<SVE_I>, *msub<SVE_I): Likewise.
5175 (*<su>mul<SVE_I>3_highpart): Likewise.
5176 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
5177 (*v<ASHIFT><SVE_I>3): Likewise.
5178 (*<su><MAXMIN><SVE_I>3): Likewise.
5179 (*<su><MAXMIN><SVE_F>3): Likewise.
5180 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
5181 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
5182 (*div<SVE_F>4): Likewise.
5184 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
5186 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
5189 2018-07-02 Richard Biener <rguenther@suse.de>
5191 PR tree-optimization/86363
5192 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
5193 memset argument refers to a non-variable address.
5195 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
5197 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
5198 {PLUS,MINUS}_EXPR code to...
5199 (adjust_symbolic_bound): ...here,
5200 (combine_bound): ...here,
5201 (set_value_range_with_overflow): ...and here.
5203 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
5205 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
5207 (extract_range_from_abs_expr): ...here.
5209 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
5211 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
5212 -fno-omit-frame-pointer when not optimizing.
5214 2018-07-02 Martin Liska <mliska@suse.cz>
5217 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
5218 (propagate_nothrow): Likewise.
5220 2018-07-02 Martin Liska <mliska@suse.cz>
5223 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
5225 2018-07-02 David Malcolm <dmalcolm@redhat.com>
5227 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
5228 function in r262149, changing "loc" param from source_location to
5229 const dump_location_t &.
5230 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
5231 declaration, as above.
5233 2018-07-01 Paul Koning <ni1d@arrl.net>
5235 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
5236 -munit-asm, -mgnu-asm, -mdec-asm.
5237 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
5238 (pdp11_output_labelref): New.
5239 (pdp11_output_def): New.
5240 (pdp11_output_addr_vec_elt): New.
5241 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
5242 %# and %@ format codes.
5243 (pdp11_option_override): New.
5244 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
5245 (pdp11_output_ident): New.
5246 (pdp11_asm_named_section): New.
5247 (pdp11_asm_init_sections): New.
5248 (pdp11_file_start): New.
5249 (pdp11_file_end): New.
5250 (output_ascii): Use .ascii/.asciz for -mdec-asm.
5251 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
5252 %o, like %c but octal.
5253 (pdp11_option_override): New.
5254 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
5256 (DATA_SECTION_ASM_OP): Ditto.
5257 (READONLY_DATA_SECTION_ASM_OP): New.
5258 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
5259 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
5260 (ASM_OUTPUT_LABELREF): Ditto.
5261 (ASM_OUTPUT_DEF): Ditto.
5262 (ASM_OUTPUT_EXTERNAL): New.
5263 (ASM_OUTPUT_SOURCE_FILENAME): New.
5264 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
5265 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
5266 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
5267 %# and %@ format codes.
5268 * config/pdp11/pdp11.opt (mgnu-asm): New.
5269 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
5270 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
5271 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
5273 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
5275 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
5276 dereferencing path[] beyond its length.
5278 (debug_all_paths): New.
5279 (rewire_first_differing_edge): New.
5280 (adjust_paths_after_duplication): New.
5281 (duplicate_thread_path): Call adjust_paths_after_duplication.
5283 (thread_through_all_blocks): Add new argument to
5284 duplicate_thread_path.
5286 2018-06-30 Jim Wilson <jimw@sifive.com>
5288 * config/riscv/predicates.md (p2m1_shift_operand): New.
5289 (high_mask_shift_operand): New.
5290 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
5291 pattern using p2m1_shift_operand.
5292 (lshsi3_zero_extend_3+2): New combiner pattern using
5293 high_mask_shift_operand.
5295 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5297 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
5299 (vect_recog_rotate_pattern): ...here.
5300 (vect_convert_input): Try to insert casts of invariants in the
5302 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
5303 preheader to be empty.
5305 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5307 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
5308 vector type. If given, install it in the new statement's
5310 (vect_element_precision): New function.
5311 (vect_unpromoted_value): New struct.
5312 (vect_unpromoted_value::vect_unpromoted_value): New function.
5313 (vect_unpromoted_value::set_op): Likewise.
5314 (vect_look_through_possible_promotion): Likewise.
5315 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
5316 (vect_widened_op_tree, vect_convert_input): Likewise.
5317 (vect_convert_inputs, vect_convert_output): Likewise.
5318 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
5319 to handle the optional cast of the multiplication result and
5320 vect_widened_op_tree to detect the widened multiplication itself.
5321 Do not require the input and output of promotion casts to have
5322 the same sign, but base the signedness of the operation on the
5323 input rather than the result. If the pattern includes two
5324 promotions, check that those promotions have the same sign.
5325 Do not restrict the MULT_EXPR handling to a double-width result;
5326 handle quadruple-width results and wider. Use vect_convert_inputs
5327 to convert the inputs to the common type.
5328 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
5329 to handle the optional cast of the ABS result. Also allow a sign
5330 change or a sign extension between the ABS and MINUS.
5331 Use vect_widened_op_tree to detect the widened subtraction and use
5332 vect_convert_inputs to convert the inputs to the common type.
5333 (vect_handle_widen_op_by_const): Delete.
5334 (vect_recog_widen_op_pattern): New function.
5335 (vect_recog_widen_mult_pattern): Use it.
5336 (vect_recog_widen_shift_pattern): Likewise.
5337 (vect_recog_widen_sum_pattern): Use
5338 vect_look_through_possible_promotion to handle the promoted
5341 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5343 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
5344 the containing gimple_seq *.
5345 * gimple-iterator.h (gsi_for_stmt): Declare it.
5346 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
5347 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
5348 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
5349 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
5350 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
5352 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
5354 (vect_mark_pattern_stmts): ...here. Handle cases in which the
5355 statement being replaced is part of an existing pattern
5356 definition sequence, inserting the new pattern statements before
5358 (vect_pattern_recog_1): Don't return a bool. If the statement
5359 is already part of a pattern, instead apply pattern matching
5360 to the pattern definition statements. Don't clear the
5361 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
5362 (vect_pattern_recog): Don't break after the first match;
5363 continue processing the pattern definition statements instead.
5364 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
5366 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5368 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
5369 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
5370 (vect_recog_widen_sum_pattern): Use it.
5372 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5374 * tree-vect-loop.c (vectorizable_reduction): Assert that the
5375 phi is not a pattern statement and has not been replaced by
5376 a pattern statement.
5377 * tree-vect-patterns.c (type_conversion_p): Don't check
5378 STMT_VINFO_IN_PATTERN_P.
5379 (vect_recog_vector_vector_shift_pattern): Likewise.
5380 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
5381 the pattern statement rather than the original statement; check
5382 directly for a WIDEN_MULT_EXPR here.
5383 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
5384 vect_is_simple_use to return the pattern statement rather
5385 than the original statement; use is_pattern_stmt_p to check
5386 for such a pattern statement.
5387 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
5388 to return the pattern statement rather than the original statement;
5389 don't do the same transformation here.
5390 (vect_is_simple_use): If the defining statement has been replaced
5391 by a pattern statement, return the pattern statement instead.
5392 Remove the corresponding (local) transformation from the vectype
5395 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5397 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
5398 end and default to null.
5399 * tree-vect-loop.c (vect_create_epilog_for_reduction)
5400 (vectorizable_reduction): Update calls accordingly, dropping the
5401 gimple ** argument if the passed-back statement isn't needed.
5402 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
5403 (vect_recog_rotate_pattern): Likewise.
5404 (vect_recog_mask_conversion_pattern): Likewise.
5405 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
5406 (vect_mask_constant_operand_p): Likewise.
5407 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
5408 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
5409 (get_group_load_store_type, get_load_store_type): Likewise.
5410 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
5411 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
5412 (vectorizable_conversion, vectorizable_assignment): Likewise.
5413 (vectorizable_shift, vectorizable_operation): Likewise.
5414 (vectorizable_store, vect_is_simple_cond): Likewise.
5415 (vectorizable_condition, vectorizable_comparison): Likewise.
5416 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
5417 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
5418 and move it to the end. Cope with null def_stmt_outs.
5420 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5422 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
5424 2018-06-29 Jeff Law <law@redhat.com>
5426 * config/v850/v850.c (v850_legitimate_address_p): Handle large
5427 displacements for TARGET_V850E2V3 and newer.
5428 (TARGET_LRA_P): Remove. Defaults to LRA now.
5429 * config/v850/v850.md (sign23byte_load): Remove.
5430 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
5431 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
5433 2018-06-29 Martin Liska <mliska@suse.cz>
5436 * coverage.c (coverage_init): Mangle full path name.
5437 * doc/invoke.texi: Document the change.
5438 * gcov-io.c (mangle_path): New.
5439 * gcov-io.h (mangle_path): Likewise.
5440 * gcov.c (mangle_name): Use mangle_path for path mangling.
5442 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5444 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
5445 if starting source register is not even.
5447 2018-06-29 Martin Liska <mliska@suse.cz>
5449 PR tree-optimization/86263
5450 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
5451 Make edge redirection.
5453 2018-06-29 David Malcolm <dmalcolm@redhat.com>
5455 * dumpfile.c (dump_loc): Add indentation based on scope depth.
5456 (dump_scope_depth): New variable.
5457 (get_dump_scope_depth): New function.
5458 (dump_begin_scope): New function.
5459 (dump_end_scope): New function.
5460 * dumpfile.h (get_dump_scope_depth): New declaration.
5461 (dump_begin_scope): New declaration.
5462 (dump_end_scope): New declaration.
5463 (class auto_dump_scope): New class.
5464 (AUTO_DUMP_SCOPE): New macro.
5465 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
5468 2018-06-29 Richard Biener <rguenther@suse.de>
5470 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
5471 compute_all_dependences succeeds.
5472 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
5473 exceed --param loop-max-datarefs-for-datadeps.
5475 2018-06-29 Jakub Jelinek <jakub@redhat.com>
5477 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
5479 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
5482 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
5483 alternative 0 in preferred_for_speed attribute.
5485 2018-06-28 Paul Koning <ni1d@arrl.net>
5487 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
5488 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
5489 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
5490 * config/pdp11/pdp11.md: Correct "length" attribute calculation
5491 for shift insn patterns.
5493 2018-06-28 David Malcolm <dmalcolm@redhat.com>
5495 * cgraph.c (cgraph_node::get_body): Replace assignments to
5496 "dump_file" with calls to set_dump_file.
5497 * dumpfile.c (alt_dump_file): Make static, and group with...
5498 (alt_flags): ...this definition.
5499 (dumps_are_enabled): New variable.
5500 (refresh_dumps_are_enabled): New function.
5501 (set_dump_file): New function.
5502 (set_alt_dump_file): New function.
5503 (gcc::dump_manager::dump_start): Replace assignments to
5504 "dump_file" and "alt_dump_file" with calls to set_dump_file and
5506 (gcc::dump_manager::dump_finish): Likewise.
5507 * dumpfile.h (alt_dump_file): Delete decl.
5508 (dumps_are_enabled): New variable decl.
5509 (set_dump_file): New function decl.
5510 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
5512 * tree-nested.c (lower_nested_functions): Replace assignments to
5513 "dump_file" with calls to set_dump_file.
5515 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
5517 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
5518 goto_locus of each outgoing edge of each basic block.
5520 2018-06-28 Richard Biener <rguenther@suse.de>
5522 * dwarf2out.c (decl_scope_table): Remove.
5523 (push_decl_scope): Likewise.
5524 (pop_decl_scope): Likewise.
5525 (gen_type_die_for_member): Do not call push/pop_decl_scope.
5526 (gen_struct_or_union_type_die): Likewise.
5527 (gen_tagged_type_die): Likewise.
5528 (dwarf2out_init): Do not initialize decl_scope_table.
5529 (dwarf2out_c_finalize): Do not free it.
5531 2018-06-28 Richard Biener <rguenther@suse.de>
5533 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
5534 deciding whether to not re-use a DIE.
5536 2018-06-28 Richard Biener <rguenther@suse.de>
5538 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
5539 DW_AT_abstract_origin attribute.
5541 2018-06-28 Martin Liska <mliska@suse.cz>
5543 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
5544 Use newly introduced constants.
5545 * tree-switch-conversion.h (struct jump_table_cluster):
5546 Define max_ratio_for_size and max_ratio_for_speed.
5548 2018-06-28 Martin Liska <mliska@suse.cz>
5550 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
5551 Add new checking assert to catch invalid state.
5552 (jump_table_cluster::can_be_handled): Handle single case
5554 (jump_table_cluster::is_beneficial): Bail out for such case.
5555 (bit_test_cluster::find_bit_tests):
5556 Add new checking assert to catch invalid state.
5557 (bit_test_cluster::can_be_handled): Handle single case
5559 (bit_test_cluster::is_beneficial): Bail out for such case.
5560 (switch_decision_tree::analyze_switch_statement):
5563 2018-06-28 Martin Liska <mliska@suse.cz>
5565 * common.opt: Introduce -completion option.
5566 * gcc.c (driver_handle_option): Handle it.
5567 (driver::main): Print completions if completion
5569 * opt-suggestions.c (option_proposer::get_completions):
5571 (option_proposer::suggest_completion): Likewise.
5572 (option_proposer::find_param_completions): Likewise.
5573 (verify_autocompletions): Likewise.
5574 (test_completion_valid_options): Likewise.
5575 (test_completion_valid_params): Likewise.
5576 (in_completion_p): Likewise.
5577 (empty_completion_p): Likewise.
5578 (test_completion_partial_match): Likewise.
5579 (test_completion_garbage): Likewise.
5580 (opt_proposer_c_tests): Likewise.
5581 * opt-suggestions.h: Declare new functions.
5582 * opts.c (common_handle_option): Handle OPT__completion_.
5583 * selftest-run-tests.c (selftest::run_tests): Add
5584 opt_proposer_c_tests.
5585 * selftest.c (assert_str_startswith): New.
5586 * selftest.h (assert_str_startswith): Likewise.
5587 (opt_proposer_c_tests): New.
5588 (ASSERT_STR_STARTSWITH): Likewise.
5590 2018-06-28 Martin Liska <mliska@suse.cz>
5592 * Makefile.in: Add opt-suggestions.o.
5593 * gcc-main.c: Include opt-suggestions.h.
5594 * gcc.c (driver::driver): Likewise.
5595 (driver::~driver): Remove m_option_suggestions.
5596 (driver::build_option_suggestions): Moved to option_proposer.
5597 (driver::suggest_option): Likewise.
5598 (driver::handle_unrecognized_options): Use option_proposer.
5599 * gcc.h (class driver): Add new memver m_option_proposer.
5600 * opt-suggestions.c: New file.
5601 * opt-suggestions.h: New file.
5603 2018-06-28 Martin Liska <mliska@suse.cz>
5605 * vec.h (class auto_string_vec): New (moved from auto_argvec).
5606 (auto_string_vec::~auto_string_vec): Likewise.
5608 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
5610 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
5611 prevent_decl_creation_for_types fields up and add reset_location field.
5612 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
5613 statement if id->reset_location is true.
5614 (copy_edges_for_bb): Do not set goto_locus on the new edges if
5615 id->reset_location is true.
5616 (copy_phis_for_bb): Force input_location on the arguments if
5617 id->reset_location is true.
5618 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
5619 is set on the function to be inlined.
5621 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
5623 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
5625 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
5627 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
5628 registers for Pmode.
5629 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
5630 hard registers for the clobbered pseudo.
5632 2018-06-27 Paul Koning <ni1d@arrl.net>
5634 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
5635 mutually exclusive options.
5636 * config/pdp11/constraints.md (h): New constraint.
5637 (O): Update definition to match shift code generation.
5638 (D): New constraint.
5639 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
5641 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
5643 (output_jump): Change arguments.
5644 (pdp11_fixed_cc_regs): New function.
5645 (pdp11_cc_mode): Ditto.
5646 (pdp11_expand_shift): Ditto.
5647 (pdp11_assemble_shift): Ditto.
5648 (pdp11_small_shift): Ditto.
5649 (pdp11_branch_cost): Remove.
5650 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
5652 (pdp11_register_move_cost): Update for CC registers.
5653 (pdp11_rtx_costs): Add case for LSHIFTRT.
5654 (pdp11_output_jump): Add CCNZ mode conditional branches.
5655 (notice_update_cc_on_set): Remove.
5656 (pdp11_cc_mode): New function.
5657 (simple_memory_operand): Correct pre/post decrement case.
5658 (no_side_effect_operand): New function.
5659 (pdp11_regno_reg_class): Add CC_REGS class.
5660 (pdp11_fixed_cc_regs): New function.
5661 (pdp11_small_shift): New function.
5662 (pdp11_expand_shift): New function to expand shift insns.
5663 (pdp11_assemble_shift): New function to output shifts.
5664 (pdp11_branch_cost): Remove.
5665 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
5666 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
5667 (WCHAR_TYPE): Ditto.
5668 (PTRDIFF_TYPE): Ditto.
5669 (ADJUST_INSN_LENGTH): New macro.
5670 (FIXED_REGISTERS): Add CC registers.
5671 (CALL_USED_REGISTERS): Ditto.
5673 (REG_CLASS_NAMES): Ditto.
5674 (REG_CLASS_CONTENTS): Ditto.
5675 (SELECT_CC_MODE): Use new function.
5676 (TARGET_FLAGS_REGNUM): New macro.
5677 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
5678 (cc0_reg_rtx): Remove.
5679 (CC_STATUS_MDEP): Remove.
5680 (CC_STATUS_MDEFP_INIT): Remove.
5681 (CC_IN_FPU): Remove.
5682 (NOTICE_UPDATE_CC): Remove.
5683 (REGISTER_NAMES): Add CC registers.
5684 (BRANCH_COST): Change to constant 1.
5685 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
5687 * config/pdp11/pdp11.opt (mbcopy): Remove.
5688 (mbcopy-builtin): Remove.
5689 (mbranch-cheap): Remove.
5690 (mbranch-expensive): Remove.
5691 * config/pdp11/predicates.md (expand_shift_operand): Update to
5692 match shift code generation.
5693 (ccnz_operator): New predicate.
5694 * doc/invoke.texi (PDP-11 Options): Remove deleted options
5695 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
5696 Remove non-existent option -mabshi, -mno-abshi. Document mutually
5698 * doc/md.texi (PDP-11): Document new D and h constraints. Update
5699 description of O constraint.
5701 2018-06-27 Jeff Law <law@redhat.com>
5702 Austin Law <austinklaw@gmail.com>
5704 * config/v850/v850.md (addsi3_set_flags): New pattern.
5705 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
5706 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
5707 (zero_extendhisi2_v850_set_flags): Likewise.
5708 (zero_extendqisi2_v850_set_flags): Likewise.
5709 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
5710 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
5711 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
5713 * config/v850/v850-protos.h (notice_update_cc): Remove.
5714 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
5715 (v850_print_operand): Handle 'D' and "d".
5716 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
5717 Add handling of arithmetic/logical operations compared against zero.
5718 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
5719 Do not look at v850_compare_op, instead get mode from last argument.
5720 (v850_gen_compare): Remove
5721 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
5722 after reload for prologue insns.
5723 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
5725 (construct_save_jarl): Likewise.
5726 (TARGET_FLAGS_REGNUM): Define.
5727 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
5728 (NOTICE_UPDATE_CC): Remove.
5729 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
5730 than cc0. Conditionalize on reload_completed.
5731 (cmpsi_insn, setfcc_insn): Likewise.
5732 (tst1 splitter): Turn into define_and_split which sets the flags
5734 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
5735 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
5736 (cstoresf4, cstoredf4): Clobber the flags.
5737 (cmpsi, cmpsf, cmpdf): Remove expanders.
5738 (setf_insn): Remove pattern.
5739 (addsi3): Turn into define_and_split which clobbers the flags after
5740 reload and a suitable pattern (addsi3_clobber_flags) for use after
5742 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
5743 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
5744 (ashrsi3, ashrsi3_v850e2): Likewise.
5745 (bins): Clobber the flags.
5746 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
5747 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
5748 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
5749 (call_value_internal_short, call_value_internal_long): Likewise.
5750 (callt_save_interrupt, callt_return_interrupt): Likewise.
5751 (save_interrupt, return_interrupt): Likewise.
5752 (callt_save_all_interrupt, save_all_interrupt): Likewise.
5753 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
5754 (restore_all_interrupt, _restore_all_interrupt): Likewise.
5755 (All FP comparisons): Only allow after reload has completed.
5757 (divh, divhu): Tweak output template.
5758 (branch_z_normal, branch_z_invert): Remove
5759 (branch_nz_normal, branch_nz_invert): Likewise.
5760 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
5762 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
5763 * config/v850/v850.c (notice_update_cc): Remove.
5764 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
5765 (CC_NO_CARRY): Likewise.
5766 (NOTICE_UPDATE_CC): Define to nothing.
5767 * config/v850/v850.md: Remove block comment on cc0 handling
5768 Remove "cc" attribute from all patterns. Remove cc_status handling
5769 from all patterns. Minor formatting fixes.
5771 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5773 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
5774 (cortex-a76.cortex-a55): Likewise.
5775 * config/aarch64/aarch64-tune.md: Regenerate.
5776 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
5777 cortex-a76.cortex-a55.
5779 2018-06-27 Jeff Law <law@redhat.com>
5781 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
5782 (MULTILIB_DIRNAMES): Similarly.
5784 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
5786 * gimple.h (gimple_return_retbnd): Delete.
5787 (gimple_return_set_retbnd): Likewise.
5788 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
5789 gimple_return_set_retbnd.
5790 * gimple-pretty-print.c (dump_gimple_return): Remove call to
5791 gimple_return_retbnd and adjust.
5792 * tree-inline.h (struct copy_body_data): Remove retbnd field.
5793 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
5794 Explicitly return NULL in a couple more cases. Move assertion
5795 on debug statements and remove unreachable code.
5796 (reset_debug_binding): Do not test id->retbnd.
5797 (expand_call_inline): Do not set it.
5799 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5801 * configure.ac: Add --disable-gcov option.
5802 * configure: Regenerate.
5803 * Makefile.in: Honour @enable_gcov@.
5804 * doc/install.texi: Document --disable-gcov.
5806 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5808 * config/arm/arm-cpus.in (cortex-a76): New entry.
5809 (cortex-a76.cortex-a55): Likewise.
5810 * config/arm/arm-tables.opt: Regenerate.
5811 * config/arm/arm-tune.md: Likewise.
5812 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
5813 * doc/invoke.texi (ARM Options): Document cortex-a76 and
5814 cortex-a76.cortex-a55.
5816 2018-06-27 Tamar Christina <tamar.christina@arm.com>
5819 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
5821 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
5823 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
5825 (EPILOGUE_USES): Likewise.
5827 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
5829 * tree-inline.c (remap_location): New function extracted from...
5830 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
5831 (copy_phis_for_bb): ...here. Call remap_location.
5832 (copy_cfg_body): Adjust call to copy_edges_for_bb.
5834 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
5836 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
5837 unaligned vsx for 16B memset.
5839 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
5842 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
5843 ieee128_float_type_node to long_double_type_node unless
5844 TARGET_LONG_DOUBLE_128 is set.
5846 2018-06-26 David Malcolm <dmalcolm@redhat.com>
5848 * cfgloop.c (get_loop_location): Convert return type from
5849 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
5850 by implicit construction from rtx_insn *, and using
5851 dump_user_location_t::from_function_decl for the fallback case.
5852 * cfgloop.h (get_loop_location): Convert return type from
5853 location_t to dump_user_location_t.
5854 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
5855 dump_printf_loc to pass in a dump_location_t rather than a
5856 location_t, via the gimple stmt.
5857 * coverage.c (get_coverage_counts): Update calls to
5858 dump_printf_loc to pass in dump_location_t rather than a
5860 * doc/optinfo.texi (Dump types): Convert example of
5861 dump_printf_loc from taking "locus" to taking "insn". Update
5862 description of the "_loc" calls to cover dump_location_t.
5863 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
5865 (dump_user_location_t::dump_user_location_t): New constructors,
5866 from gimple *stmt and rtx_insn *.
5867 (dump_user_location_t::from_function_decl): New function.
5868 (dump_loc): Make static.
5869 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
5870 const dump_location_t &.
5871 (dump_generic_expr_loc): Delete.
5872 (dump_printf_loc): Convert param "loc" from location_t to
5873 const dump_location_t &.
5874 (selftest::test_impl_location): New function.
5875 (selftest::dumpfile_c_tests): New function.
5876 * dumpfile.h: Include "profile-count.h".
5877 (class dump_user_location_t): New class.
5878 (struct dump_impl_location_t): New struct.
5879 (class dump_location_t): New class.
5880 (dump_printf_loc): Convert 2nd param from source_location to
5881 const dump_location_t &.
5882 (dump_generic_expr_loc): Delete.
5883 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
5884 const dump_location_t &.
5885 * gimple-fold.c (fold_gimple_assign): Update call to
5886 dump_printf_loc to pass in a dump_location_t rather than a
5887 location_t, via the gimple stmt.
5888 (gimple_fold_call): Likewise.
5889 * gimple-loop-interchange.cc
5890 (loop_cand::analyze_iloop_reduction_var): Update for change to
5891 check_reduction_path.
5892 (tree_loop_interchange::interchange): Update for change to
5894 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
5895 change in return-type of find_loop_location.
5896 (graphite_regenerate_ast_isl): Likewise.
5897 * graphite-optimize-isl.c (optimize_isl): Likewise.
5898 * graphite.c (graphite_transform_loops): Likewise.
5899 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
5900 pass in a dump_location_t rather than a location_t, via the
5902 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
5903 * ipa.c (walk_polymorphic_call_targets): Likewise.
5904 * loop-unroll.c (report_unroll): Convert "locus" param from
5905 location_t to dump_location_t.
5906 (decide_unrolling): Update for change to get_loop_location's
5908 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
5909 location_t to dump_user_location_t.
5910 (grid_find_single_omp_among_assignments_1): Updates calls to
5911 dump_printf_loc to pass in a dump_location_t rather than a
5912 location_t, via the gimple stmt.
5913 (grid_parallel_clauses_gridifiable): Convert "tloc" from
5914 location_t to dump_location_t. Updates calls to dump_printf_loc
5915 to pass in a dump_location_t rather than a location_t, via the
5917 (grid_inner_loop_gridifiable_p): Likewise.
5918 (grid_dist_follows_simple_pattern): Likewise.
5919 (grid_gfor_follows_tiling_pattern): Likewise.
5920 (grid_target_follows_gridifiable_pattern): Likewise.
5921 (grid_attempt_target_gridification): Convert initialization
5922 of local "grid" from memset to zero-initialization; FIXME: does
5923 this require C++11? Update call to dump_printf_loc to pass in a
5924 optinfo_location rather than a location_t, via the gimple stmt.
5925 * profile.c (read_profile_edge_counts): Updates call to
5926 dump_printf_loc to pass in a dump_location_t rather than a
5928 (compute_branch_probabilities): Likewise.
5929 * selftest-run-tests.c (selftest::run_tests): Call
5931 * selftest.h (dumpfile_c_tests): New decl.
5932 * tree-loop-distribution.c (pass_loop_distribution::execute):
5933 Update for change in return type of find_loop_location.
5934 * tree-parloops.c (parallelize_loops): Likewise.
5935 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
5936 "locus" from location_t to dump_user_location_t.
5937 (canonicalize_loop_induction_variables): Likewise.
5938 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
5939 for change in return type of find_loop_location.
5940 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
5941 to dump_printf_loc to pass in a dump_location_t rather than a
5942 location_t, via the stmt.
5943 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
5945 * tree-vect-loop-manip.c (find_loop_location): Convert return
5946 type from source_location to dump_user_location_t.
5947 (vect_do_peeling): Update for above change.
5948 (vect_loop_versioning): Update for change in type of
5950 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
5951 from location_t to dump_user_location_t.
5952 (vect_estimate_min_profitable_iters): Update for change in type
5954 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
5955 location_t to dump_location_t.
5956 (vect_slp_bb): Update for change in type of vect_location.
5957 * tree-vectorizer.c (vect_location): Convert from source_location
5958 to dump_user_location_t.
5959 (try_vectorize_loop_1): Update for change in vect_location's type.
5960 (vectorize_loops): Likewise.
5961 (increase_alignment): Likewise.
5962 * tree-vectorizer.h (vect_location): Convert from source_location
5963 to dump_user_location_t.
5964 (find_loop_location): Convert return type from source_location to
5965 dump_user_location_t.
5966 (check_reduction_path): Convert 1st param from location_t to
5967 dump_user_location_t.
5968 * value-prof.c (check_counter): Update call to dump_printf_loc to
5969 pass in a dump_user_location_t rather than a location_t; update
5970 call to error_at for change in type of "locus".
5971 (check_ic_target): Update call to dump_printf_loc to
5972 pass in a dump_user_location_t rather than a location_t, via the
5975 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
5977 * config/s390/s390.h (enum processor_flags): Do not use
5978 default tune parameter when -march was specified.
5980 2018-06-26 Jakub Jelinek <jakub@redhat.com>
5983 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
5984 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
5987 2018-06-26 Richard Biener <rguenther@suse.de>
5989 PR tree-optimization/86287
5991 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
5992 (vect_analyze_loop): Initialize n_stmts.
5994 2018-06-26 Richard Biener <rguenther@suse.de>
5997 * fold-const.c (fold_convertible_p): Pointer extension
6000 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
6003 * dwarf2out.c (loc_list_has_views): Adjust comments.
6004 (dw_loc_list): Split single cross-partition range with
6007 2018-06-25 Jeff Law <law@redhat.com>
6009 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
6010 on -mbig-switch by default.
6012 * config/v850/predicates.md (const_float_1_operand): Fix match_code
6014 (const_float_0_operand): Remove unused predicate.
6015 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
6016 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
6017 (recipsf2): New expander. Original pattern now called
6019 (recipdf2, recipdf2_insn): Similarly.
6020 (rsqrtsf2, rsqrtsf2_insn): Similarly
6021 (rsqrtdf2, rsqrtdf2_insn): Similarly
6023 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
6025 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
6026 Simplify logic for FreeBSD (twice).
6028 2018-06-25 Martin Sebor <msebor@redhat.com>
6030 PR tree-optimization/86204
6031 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
6032 a strnlen result if it's less than the length of the string.
6034 2018-06-25 Martin Sebor <msebor@redhat.com>
6036 PR tree-optimization/85700
6037 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
6038 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
6039 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
6041 2018-06-25 Martin Sebor <msebor@redhat.com>
6043 * doc/extend.texi (Zero-length arrays): Update and clarify.
6045 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
6047 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
6048 added IEEE/IBM long double multilib support on PowerPC little
6049 endian Linux systems.
6050 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
6051 (MULTILIB_DEFAULTS): Likewise.
6052 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6054 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
6055 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
6056 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
6058 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
6061 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
6062 (REORDER_45): Likewise.
6064 2018-06-25 Jeff Law <law@redhat.com>
6066 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
6067 dividend to 32 bits. Adjust length.
6068 (udivmodhi4): Cleanup output template. Fix length.
6070 2018-06-25 Carl Love <cel@us.ibm.com>
6072 * config/rs6000/vsx.md: Change word selector to prefered location.
6074 2018-06-25 Richard Biener <rguenther@suse.de>
6076 PR tree-optimization/86304
6077 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
6078 epilogue-if-converted loops as well.
6080 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
6082 * lto-section-out.c (lto_begin_section): Do not print section
6083 name for noaddr and unnumbered dumps.
6085 2018-06-25 Richard Biener <rguenther@suse.de>
6087 * tree-vectorizer.h (struct vec_info_shared): New structure
6088 with parts split out from struct vec_info and loop_nest from
6089 struct _loop_vec_info.
6090 (struct vec_info): Adjust accordingly.
6091 (struct _loop_vec_info): Likewise.
6092 (LOOP_VINFO_LOOP_NEST): Adjust.
6093 (LOOP_VINFO_DATAREFS): Likewise.
6094 (LOOP_VINFO_DDRS): Likewise.
6095 (struct _bb_vec_info): Likewise.
6096 (BB_VINFO_DATAREFS): Likewise.
6097 (BB_VINFO_DDRS): Likewise.
6098 (struct _stmt_vec_info): Add dr_aux member.
6099 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
6100 (DR_MISALIGNMENT_UNINITIALIZED): New.
6101 (set_dr_misalignment): Adjust.
6102 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
6103 (vect_analyze_loop): Adjust prototype.
6104 (vect_analyze_loop_form): Likewise.
6105 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
6106 Compute dependences lazily.
6107 (vect_record_base_alignments): Use shared datarefs/ddrs.
6108 (vect_verify_datarefs_alignment): Likewise.
6109 (vect_analyze_data_refs_alignment): Likewise.
6110 (vect_analyze_data_ref_accesses): Likewise.
6111 (vect_analyze_data_refs): Likewise.
6112 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
6113 constructor parameter for shared part.
6114 (vect_analyze_loop_form): Pass in shared part and adjust.
6115 (vect_analyze_loop_2): Pass in storage for the number of
6116 stmts. Move loop nest finding to the caller. Compute
6118 (vect_analyze_loop): Pass in shared part.
6119 (vect_transform_loop): Verify shared datarefs are unchanged.
6120 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
6121 constructor parameter for shared part.
6122 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
6123 (vect_slp_bb): Verify shared datarefs are unchanged before
6125 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
6127 (new_stmt_vec_info): Initialize DR_AUX misalignment to
6128 DR_MISALIGNMENT_UNINITIALIZED.
6129 * tree-vectorizer.c (vec_info::vec_info): Add constructor
6130 parameter for shared part.
6131 (vec_info::~vec_info): Adjust.
6132 (vec_info_shared::vec_info_shared): New.
6133 (vec_info_shared::~vec_info_shared): Likewise.
6134 (vec_info_shared::save_datarefs): Likewise.
6135 (vec_info_shared::check_datarefs): Likewise.
6136 (try_vectorize_loop_1): Construct shared part live for analyses
6137 of a single loop for multiple vector sizes.
6138 * tree-parloops.c (gather_scalar_reductions): Adjust.
6140 2018-06-25 Richard Biener <rguenther@suse.de>
6142 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
6143 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
6144 (vect_analyze_data_refs): Remove similar code from here and
6145 simplify accordingly.
6147 2018-06-25 Richard Biener <rguenther@suse.de>
6149 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
6150 for reverse storage order accesses rather than asserting
6151 they cannot happen here.
6153 2018-06-25 Tom de Vries <tdevries@suse.de>
6156 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
6157 Use data16 instead of .byte for insn prefix.
6159 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
6162 * parser.c (make_char_string_pack): Pass this literal chars
6163 through cpp_interpret_string.
6164 (cp_parser_userdef_numeric_literal): Check the result of
6165 make_char_string_pack.
6167 2018-06-24 Maya Rashish <coypu@sdf.org>
6169 * ginclude/stddef.h: Simplify conditions around avoiding
6170 re-definition of __size_t.
6172 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6174 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
6175 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
6177 2018-06-22 Maya Rashish <coypu@sdf.org>
6179 * doc/invoke.texi (mno-fancy-math-387): Update for changes
6180 made to OpenBSD and NetBSD through the years.
6182 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
6184 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
6185 behavior of vec_pack (vector double, vector double) to match
6186 behavior of vec_float2 (vector double, vector double).
6188 2018-06-22 Olivier Hainque <hainque@adacore.com>
6190 * gimplify.c (gimplify_function_tree): Prevent creation
6191 of a trampoline for the address of the current function
6192 passed to entry/exit instrumentation hooks.
6194 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
6197 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
6200 2018-06-22 Martin Liska <mliska@suse.cz>
6202 PR tree-optimization/86263
6203 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
6204 Bail out if is_enabled is false.
6205 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
6207 (jump_table_cluster::is_enabled): New function.
6209 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6211 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6212 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
6213 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
6214 (lto_input_ts_binfo_tree_pointers): Likewise.
6215 * tree-streamer-out.c (streamer_write_tree_bitfields,
6216 write_ts_binfo_tree_pointers): Likewise.
6217 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
6219 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6221 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
6223 2018-06-22 Martin Liska <mliska@suse.cz>
6225 * symbol-summary.h (get): Make it pure and inline move
6226 functionality from ::get function.
6227 (get): Remove and inline into ::get and ::get_create.
6228 (get_create): Move code from ::get function.
6230 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6233 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
6234 -x assembler-with-cpp.
6236 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6238 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
6239 _FILE_OFFSET_BITS=64 for C++.
6241 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
6243 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
6244 conversion insn that shows up when pr85657-3.c is compiled using
6245 IEEE 128-bit long double.
6246 (neg<mode>2_internal): Use the correct mode to check whether the
6247 mode is IBM extended.
6248 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
6249 multiply and divide external functions from being created more
6252 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6254 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
6256 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
6257 the edge can be forwarded.
6258 (cfg_layout_merge_blocks): Likewise.
6260 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6262 * except.c (finish_eh_generation): Commit edge insertions only after
6263 the EH edges have been redirected from post-landing to landing pads.
6265 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6267 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
6268 create_tmp_var_for to create the FRAME decl.
6269 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
6271 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6273 * tree-inline.c (copy_edges_for_bb): Minor tweak.
6274 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
6275 debug statement when resetting its value.
6276 (expand_call_inline): Copy the locus of the call onto the assignment
6277 of the return value, if any. Use local variable in more cases.
6279 2018-06-21 Martin Liska <mliska@suse.cz>
6281 * ipa-pure-const.c (propagate_nothrow): Use
6282 funct_state_summaries->get.
6283 (dump_malloc_lattice): Likewise.
6284 (propagate_malloc): Likewise.
6286 2018-06-21 Richard Biener <rguenther@suse.de>
6288 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
6289 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
6290 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
6292 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
6293 BLOCK_ABSTRACT_ORIGIN unconditionally.
6295 2018-06-21 David Malcolm <dmalcolm@redhat.com>
6297 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
6299 * ipa-reference.c (ipa_reference_c_finalize): Delete
6300 ipa_ref_opt_sum_summaries and set it to NULL.
6302 2018-06-21 Tom de Vries <tdevries@suse.de>
6304 PR tree-optimization/85859
6305 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
6306 test with comment from bb_no_side_effects_p.
6308 2018-06-21 Richard Biener <rguenther@suse.de>
6310 PR tree-optimization/86232
6311 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
6312 max for constant niter.
6314 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
6316 * config/aarch64/aarch64-simd.md
6317 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
6319 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
6321 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
6322 Make opernads of the unspec commutative.
6324 2018-06-21 Richard Biener <rguenther@suse.de>
6326 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
6327 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6328 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
6329 (vect_analyze_data_ref_dependence): Re-order checks to deal with
6331 (vect_record_base_alignments): Do not record base alignment
6332 for gathers or scatters.
6333 (vect_compute_data_ref_alignment): Drop return value that is always
6334 true. Bail out early for gathers or scatters.
6335 (vect_enhance_data_refs_alignment): Bail out early for gathers
6337 (vect_find_same_alignment_drs): Likewise.
6338 (vect_analyze_data_refs_alignment): Remove dead code.
6339 (vect_slp_analyze_and_verify_node_alignment): Likewise.
6340 (vect_analyze_data_refs): For possible gathers or scatters do
6341 not create an alternate DR, just check their possible validity
6342 and mark them. Adjust DECL_NONALIASED handling to not rely
6344 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
6345 update inits of gathers or scatters.
6346 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
6347 Also copy gather/scatter flag to pattern vinfo.
6349 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
6351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
6352 behavior of vec_packsu (vector unsigned long long, vector unsigned
6353 long long) to match behavior of vec_packs with same signature.
6355 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
6356 Thomas Schwinge <thomas@codesourcery.com>
6357 Cesar Philippidis <cesar@codesourcery.com>
6359 * gimplify.c (gimplify_scan_omp_clauses): Add support for
6360 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
6361 (gimplify_adjust_omp_clauses): Likewise.
6362 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
6363 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
6364 (gimplify_omp_target_update): Update handling of acc update and
6366 * omp-low.c (install_var_field): Remove unused parameter
6367 base_pointers_restrict.
6368 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
6369 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
6371 (omp_target_base_pointers_restrict_p): Delete.
6372 (scan_omp_target): Update call to scan_sharing_clauses.
6373 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
6375 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
6376 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
6377 (convert_local_omp_clauses): Likewise.
6378 * tree-pretty-print.c (dump_omp_clause): Likewise.
6379 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
6381 (omp_clause_code_name): Likewise.
6383 2018-06-20 Jakub Jelinek <jakub@redhat.com>
6386 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
6389 PR tree-optimization/86231
6390 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
6391 anti-range don't overwrite *vr0min before using it to compute *vr0max.
6393 2018-06-20 Tom de Vries <tdevries@suse.de>
6395 PR tree-optimization/86097
6396 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
6397 iv type if signedness of iv type is not the same as that of *nit.
6399 2018-06-20 Jakub Jelinek <jakub@redhat.com>
6401 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
6402 EDGE_EH edges, verify they are all EDGE_EH.
6404 2018-06-20 Maya Rashish <coypu@sdf.org>
6406 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
6408 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6410 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
6411 * config/aarch64/aarch64.c (xgene1_tunings): Add
6412 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
6413 (aarch64_mode_valid_for_sched_fusion_p):
6414 Allow 16-byte modes.
6415 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
6416 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
6418 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
6420 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
6421 * config/aarch64/iterators.md (VQ2): New mode iterator.
6423 2018-06-20 Martin Liska <mliska@suse.cz>
6425 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
6426 Change default ratio from 10 to 8.
6428 2018-06-20 Martin Liska <mliska@suse.cz>
6430 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
6432 (bit_test_cluster::find_bit_tests): Likewise.
6433 (switch_decision_tree::analyze_switch_statement): Find clusters.
6434 * tree-switch-conversion.h (struct jump_table_cluster): Document
6437 2018-06-20 Martin Liska <mliska@suse.cz>
6439 * tree-switch-conversion.c (switch_conversion::collect):
6440 Record m_uniq property.
6441 (switch_conversion::expand): Bail out for special conditions.
6442 (group_cluster::~group_cluster): New.
6443 (group_cluster::group_cluster): Likewise.
6444 (group_cluster::dump): Likewise.
6445 (jump_table_cluster::emit): New.
6446 (switch_decision_tree::fix_phi_operands_for_edges): New.
6447 (struct case_node): Remove struct.
6448 (jump_table_cluster::can_be_handled): New.
6449 (case_values_threshold): Moved to header.
6450 (reset_out_edges_aux): Likewise.
6451 (jump_table_cluster::is_beneficial): New.
6452 (bit_test_cluster::can_be_handled): Likewise.
6453 (add_case_node): Remove.
6454 (bit_test_cluster::is_beneficial): New.
6455 (case_bit_test::cmp): New.
6456 (bit_test_cluster::emit): New.
6457 (expand_switch_as_decision_tree_p): Remove.
6458 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
6459 (fix_phi_operands_for_edge): Likewise.
6460 (switch_decision_tree::analyze_switch_statement): New.
6461 (compute_cases_per_edge): Move ...
6462 (switch_decision_tree::compute_cases_per_edge): ... here.
6463 (try_switch_expansion): Likewise.
6464 (switch_decision_tree::try_switch_expansion): Likewise.
6465 (record_phi_operand_mapping): Likewise.
6466 (switch_decision_tree::record_phi_operand_mapping): Likewise.
6467 (emit_case_decision_tree): Likewise.
6468 (switch_decision_tree::emit): Likewise.
6469 (balance_case_nodes): Likewise.
6470 (switch_decision_tree::balance_case_nodes): Likewise.
6471 (dump_case_nodes): Likewise.
6472 (switch_decision_tree::dump_case_nodes): Likewise.
6473 (emit_jump): Likewise.
6474 (switch_decision_tree::emit_jump): Likewise.
6475 (emit_cmp_and_jump_insns): Likewise.
6476 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
6477 (emit_case_nodes): Likewise.
6478 (switch_decision_tree::emit_case_nodes): Likewise.
6479 (conditional_probability): Remove.
6480 * tree-switch-conversion.h (enum cluster_type): New.
6482 (struct cluster): Likewise.
6483 (cluster::cluster): Likewise.
6484 (struct simple_cluster): Likewise.
6485 (simple_cluster::simple_cluster): Likewise.
6486 (struct group_cluster): Likewise.
6487 (struct jump_table_cluster): Likewise.
6488 (struct bit_test_cluster): Likewise.
6489 (struct min_cluster_item): Likewise.
6490 (struct case_tree_node): Likewise.
6491 (case_tree_node::case_tree_node): Likewise.
6492 (jump_table_cluster::case_values_threshold): Likewise.
6493 (struct case_bit_test): Likewise.
6494 (struct switch_decision_tree): Likewise.
6495 (struct switch_conversion): Likewise.
6496 (switch_decision_tree::reset_out_edges_aux): Likewise.
6498 2018-06-20 Martin Liska <mliska@suse.cz>
6500 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
6501 (hoist_edge_and_branch_if_true): Likewise.
6502 (expand_switch_using_bit_tests_p): Likewise.
6503 (struct case_bit_test): Likewise.
6504 (case_bit_test_cmp): Likewise.
6505 (emit_case_bit_tests): Likewise.
6506 (switch_conversion::switch_conversion): New class.
6507 (struct switch_conv_info): Remove old struct.
6508 (collect_switch_conv_info): More to ...
6509 (switch_conversion::collect): ... this.
6510 (check_range): Likewise.
6511 (switch_conversion::check_range): Likewise.
6512 (check_all_empty_except_final): Likewise.
6513 (switch_conversion::check_all_empty_except_final): Likewise.
6514 (check_final_bb): Likewise.
6515 (switch_conversion::check_final_bb): Likewise.
6516 (create_temp_arrays): Likewise.
6517 (switch_conversion::create_temp_arrays): Likewise.
6518 (free_temp_arrays): Likewise.
6519 (gather_default_values): Likewise.
6520 (switch_conversion::gather_default_values): Likewise.
6521 (build_constructors): Likewise.
6522 (switch_conversion::build_constructors): Likewise.
6523 (constructor_contains_same_values_p): Likewise.
6524 (switch_conversion::contains_same_values_p): Likewise.
6525 (array_value_type): Likewise.
6526 (switch_conversion::array_value_type): Likewise.
6527 (build_one_array): Likewise.
6528 (switch_conversion::build_one_array): Likewise.
6529 (build_arrays): Likewise.
6530 (switch_conversion::build_arrays): Likewise.
6531 (gen_def_assigns): Likewise.
6532 (switch_conversion::gen_def_assigns): Likewise.
6533 (prune_bbs): Likewise.
6534 (switch_conversion::prune_bbs): Likewise.
6535 (fix_phi_nodes): Likewise.
6536 (switch_conversion::fix_phi_nodes): Likewise.
6537 (gen_inbound_check): Likewise.
6538 (switch_conversion::gen_inbound_check): Likewise.
6539 (process_switch): Use the newly created class.
6540 (switch_conversion::expand): New.
6541 (switch_conversion::~switch_conversion): New.
6542 * tree-switch-conversion.h: New file.
6544 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6546 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
6547 tree-vect-patterns.c.
6548 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
6549 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
6550 (vect_recog_sad_pattern): Likewise.
6551 (vect_recog_widen_sum_pattern): Likewise.
6552 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
6553 (vect_recog_widen_shift_pattern): Remove the type_in argument.
6554 (vect_recog_rotate_pattern): Likewise.
6555 (vect_recog_mult_pattern): Likewise.
6556 (vect_recog_vector_vector_shift_pattern): Likewise.
6557 (vect_recog_divmod_pattern): Likewise.
6558 (vect_recog_mixed_size_cond_pattern): Likewise.
6559 (vect_recog_bool_pattern): Likewise.
6560 (vect_recog_mask_conversion_pattern): Likewise.
6561 (vect_try_gather_scatter_pattern): Likewise.
6562 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
6563 (vect_recog_over_widening_pattern): Likewise.
6564 (vect_recog_gather_scatter_pattern): Likewise.
6565 (vect_recog_func_ptr): Move from tree-vectorizer.h
6566 (vect_vect_recog_func_ptrs): Move further down the file.
6567 (vect_recog_func): Likewise. Remove the third argument.
6568 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
6569 (vect_pattern_recog_1): Expect the pattern function to do any
6570 necessary target tests. Also expect it to provide a vector type.
6571 Remove the type_in handling.
6573 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6575 * tree-vect-patterns.c (vect_pattern_detected): New function.
6576 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
6577 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
6578 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
6579 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
6580 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6581 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6582 (vect_recog_mask_conversion_pattern)
6583 (vect_try_gather_scatter_pattern): Likewise.
6585 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6587 * tree-vect-patterns.c (vect_get_internal_def): New function.
6588 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6589 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
6590 (search_type_for_mask_1): Use it.
6592 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6594 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
6595 redundant WIDEN_SUM_EXPR handling.
6596 (vect_recog_sad_pattern): Likewise.
6598 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6600 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
6601 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
6602 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
6603 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
6604 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
6606 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6608 * tree-vect-stmts.c (vectorizable_call): Make sure that we
6609 use the stmt_vec_info of the original bb statement for the
6610 new zero assignment, even if the call is part of a pattern.
6612 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6614 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
6615 that the sequence is attached to the original statement rather
6616 than the pattern statement.
6617 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
6618 PATTERN_DEF_SEQ from the original statement rather than
6619 the main pattern statement.
6620 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
6621 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
6622 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
6624 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
6626 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
6627 definition statements before the early exit for statements that aren't
6629 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
6631 (vect_transform_loop): ...here. Process pattern definition
6632 statements without first checking whether the main pattern
6633 statement is live or relevant.
6635 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
6637 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
6638 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
6640 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
6642 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
6643 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
6644 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
6645 (expand_block_compare): Change select_block_compare_mode call.
6646 (expand_strncmp_align_check): Use new functions, fix comment.
6647 (emit_final_str_compare_gpr): New function.
6648 (expand_strn_compare): Refactor and clean up code.
6649 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
6651 2018-06-19 Tony Reix <tony.reix@atos.com>
6652 Damien Bergamini <damien.bergamini@atos.com>
6653 David Edelsohn <dje.gcc@gmail.com>
6655 * collect2.c (static_obj): New variable.
6656 (static_libs): New variable.
6657 (is_in_list): Uncomment declaration.
6658 (main): Track AIX libraries linked statically.
6659 (is_in_list): Uncomment definition.
6660 (scan_prog_file): Don't add AIX shared libraries initializer
6661 to constructor list if linking statically.
6663 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
6665 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
6667 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
6669 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
6671 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
6674 2018-06-19 Martin Liska <mliska@suse.cz>
6676 * config/i386/i386.c (ix86_can_inline_p): Do not use
6677 ipa_fn_summaries::get_create.
6678 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
6680 (devirtualization_time_bonus): Likewise.
6681 (ipcp_propagate_stage): Likewise.
6682 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
6683 (edge_set_predicate): Likewise.
6684 (evaluate_conditions_for_known_args): Likewise.
6685 (evaluate_properties_for_edge): Likewise.
6686 (ipa_call_summary::reset): Tranform to ...
6687 (ipa_call_summary::~ipa_call_summary): ... this.
6688 (ipa_fn_summary::reset): Transform to ...
6689 (ipa_fn_summary::~ipa_fn_summary): ... this.
6690 (ipa_fn_summary_t::remove): Rename to ...
6691 (ipa_fn_summary_t::remove_callees): ... this.
6692 (ipa_fn_summary_t::duplicate): Use placement new
6693 instead of memory copy.
6694 (ipa_call_summary_t::duplicate): Likewise.
6695 (ipa_call_summary_t::remove): Remove.
6696 (dump_ipa_call_summary): Change get_create to get.
6697 (ipa_dump_fn_summary): Dump only when summary exists.
6698 (analyze_function_body): Use symbol_summary::get instead
6700 (compute_fn_summary): Likewise.
6701 (estimate_edge_devirt_benefit): Likewise.
6702 (estimate_edge_size_and_time): Likewise.
6703 (inline_update_callee_summaries): Likewise.
6704 (remap_edge_change_prob): Likewise.
6705 (remap_edge_summaries): Likewise.
6706 (ipa_merge_fn_summary_after_inlining): Likewise.
6707 (write_ipa_call_summary): Likewise.
6708 (ipa_fn_summary_write): Likewise.
6709 (ipa_free_fn_summary): Likewise.
6710 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
6711 (struct ipa_call_summary): Likewise.
6712 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
6714 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
6715 (estimate_size_after_inlining): Likewise.
6716 (estimate_growth): Likewise.
6717 (growth_likely_positive): Likewise.
6718 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
6719 (inline_call): Likewise.
6720 * ipa-inline.c (caller_growth_limits): Likewise.
6721 (can_inline_edge_p): Likewise.
6722 (can_inline_edge_by_limits_p): Likewise.
6723 (compute_uninlined_call_time): Likewise.
6724 (compute_inlined_call_time): Likewise.
6725 (want_inline_small_function_p): Likewise.
6726 (edge_badness): Likewise.
6727 (update_caller_keys): Likewise.
6728 (update_callee_keys): Likewise.
6729 (inline_small_functions): Likewise.
6730 (inline_to_all_callers_1): Likewise.
6731 (dump_overall_stats): Likewise.
6732 (early_inline_small_functions): Likewise.
6733 (early_inliner): Likewise.
6734 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
6735 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
6736 * ipa-pure-const.c (malloc_candidate_p): Likewise.
6737 * ipa-split.c (execute_split_functions): Likewise.
6738 * symbol-summary.h: Likewise.
6739 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
6741 2018-06-19 Richard Biener <rguenther@suse.de>
6743 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
6744 (vectorize_loops): ... here. Fix dbgcnt handling.
6745 (try_vectorize_loop): Wrap try_vectorize_loop_1.
6747 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
6750 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
6751 ieee128 argument takes up only one (vector) register, not two (floating
6754 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
6756 * gimplify.c (gimplify_init_constructor): Really never clear for an
6757 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
6759 2018-06-19 Richard Biener <rguenther@suse.de>
6761 PR tree-optimization/86179
6762 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
6763 after failed recognition.
6765 2018-06-18 Martin Sebor <msebor@redhat.com>
6768 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
6769 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
6770 Handle integer subtraction.
6771 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
6772 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
6774 2018-06-18 David Malcolm <dmalcolm@redhat.com>
6776 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
6777 param from rtx to rtx_insn *.
6778 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
6780 (frv_ifcvt_modify_insn): Likwise.
6781 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
6782 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
6783 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
6784 as_a <rtx_insn *> cast to local "unprotected_region" once
6785 it's been established that it's not NULL or pc_rtx.
6786 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
6787 param "sethi" from rtx to rtx_insn *.
6788 (nds32_group_float_insns): Likewise for param "insn".
6789 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
6791 (vax_output_int_subtract): Likewise.
6792 * config/vax/vax.c (vax_output_int_add): Likewise for param
6794 (vax_output_int_subtract): Likewise.
6795 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
6796 (emit_pattern_after): Likewise for param "after".
6797 (emit_insn_after): Likewise.
6798 (emit_jump_insn_after): Likewise.
6799 (emit_call_insn_after): Likewise.
6800 (emit_debug_insn_after): Likewise.
6801 (emit_pattern_before): Likewise for param "before".
6802 (emit_insn_before): Likewise.
6803 (emit_jump_insn_before): Likewise.
6804 * final.c (get_insn_template): Likewise for param "insn", removing
6806 * output.h (get_insn_template): Likewise for 2nd param.
6807 * rtl.h (emit_insn_before): Likewise.
6808 (emit_jump_insn_before): Likewise.
6809 (emit_debug_insn_before_noloc): Likewise.
6810 (emit_insn_after): Likewise.
6811 (emit_jump_insn_after): Likewise.
6812 (emit_call_insn_after): Likewise.
6813 (emit_debug_insn_after): Likewise.
6814 (set_insn_deleted): Likewise for param.
6816 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
6819 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
6820 floating point modes, so that IFmode is numerically greater than
6821 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
6822 to declare the ordering. This prevents IFmode from being
6823 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
6824 machine. Include rs6000-modes.h to share the fractional values
6825 between genmodes* and the rest of the compiler.
6829 * config/rs6000/rs6000-modes.h: New file.
6830 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
6831 meaning of rs6000_long_double_size so that 126..128 selects an
6832 appropriate 128-bit floating point type.
6833 (rs6000_option_override_internal): Likewise.
6834 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
6835 (TARGET_LONG_DOUBLE_128): Change the meaning of
6836 rs6000_long_double_size so that 126..128 selects an appropriate
6837 128-bit floating point type.
6838 (LONG_DOUBLE_TYPE_SIZE): Update comment.
6839 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
6840 source and destination to match the standard usage.
6841 (truncifkf2): Likewise.
6842 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
6843 ISA 2.07 to use an explicit clobber, instead of passing in a
6845 (copysign<mode>3_soft): Likewise.
6847 2018-06-18 David Malcolm <dmalcolm@redhat.com>
6849 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
6850 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
6851 (vect_slp_analyze_instance_dependence): Likewise.
6852 (vect_enhance_data_refs_alignment): Likewise.
6853 (vect_analyze_data_refs_alignment): Likewise.
6854 (vect_slp_analyze_and_verify_instance_alignment
6855 (vect_analyze_data_ref_accesses): Likewise.
6856 (vect_prune_runtime_alias_test_list): Likewise.
6857 (vect_analyze_data_refs): Likewise.
6858 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
6859 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
6860 (vect_analyze_scalar_cycles_1): Likewise.
6861 (vect_get_loop_niters): Likewise.
6862 (vect_analyze_loop_form_1): Likewise.
6863 (vect_update_vf_for_slp): Likewise.
6864 (vect_analyze_loop_operations): Likewise.
6865 (vect_analyze_loop): Likewise.
6866 (vectorizable_induction): Likewise.
6867 (vect_transform_loop): Likewise.
6868 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
6869 * tree-vect-slp.c (vect_analyze_slp): Likewise.
6870 (vect_make_slp_decision): Likewise.
6871 (vect_detect_hybrid_slp): Likewise.
6872 (vect_slp_analyze_operations): Likewise.
6873 (vect_slp_bb): Likewise.
6874 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
6875 (vectorizable_bswap): Likewise.
6876 (vectorizable_call): Likewise.
6877 (vectorizable_simd_clone_call): Likewise.
6878 (vectorizable_conversion): Likewise.
6879 (vectorizable_assignment): Likewise.
6880 (vectorizable_shift): Likewise.
6881 (vectorizable_operation): Likewise.
6882 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
6884 2018-06-18 Martin Sebor <msebor@redhat.com>
6886 PR tree-optimization/81384
6887 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
6888 * builtins.c (expand_builtin_strnlen): New function.
6889 (expand_builtin): Call it.
6890 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
6891 * builtins.def (BUILT_IN_STRNLEN): New.
6892 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
6893 Warn for bounds in excess of maximum object size.
6894 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
6895 single-value ranges. Handle strnlen.
6896 (handle_builtin_strlen): Handle strnlen.
6897 (strlen_check_and_optimize_stmt): Same.
6898 * doc/extend.texi (Other Builtins): Document strnlen.
6900 2018-06-18 Maya Rashish <coypu@sdf.org>
6902 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
6903 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
6904 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
6906 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
6908 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
6910 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6912 * tree.c (escaped_string::escape): Replace cast to char * by
6913 const_cast<char *> (unescaped).
6915 2018-06-18 Nick Clifton <nickc@redhat.com>
6918 * tree.c (escaped_string): New class. Converts an unescaped
6919 string into its escaped equivalent.
6920 (warn_deprecated_use): Use the new class to convert the
6921 deprecation message, if present.
6922 (test_escaped_strings): New self test.
6923 (test_c_tests): Add test_escaped_strings.
6924 * doc/extend.texi (deprecated): Add a note that the
6925 deprecation message is affected by the -fmessage-length
6926 option, and that control characters will be escaped.
6927 (#pragma GCC error): Document this pragma.
6928 (#pragma GCC warning): Likewise.
6929 * doc/invoke.texi (-fmessage-length): Document this option's
6930 effect on the #warning and #error preprocessor directives and
6931 the deprecated attribute.
6933 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
6935 * tree.c (decl_value_expr_lookup): Revert latest change.
6936 (decl_value_expr_insert): Likewise.
6938 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
6940 * gimplify.c (nonlocal_vlas): Delete.
6941 (nonlocal_vla_vars): Likewise.
6942 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
6944 (gimplify_body): Do not create and destroy nonlocal_vlas.
6945 * tree-nested.c: Include diagnostic.h.
6946 (use_pointer_in_frame): Tweak.
6947 (lookup_field_for_decl): Add assertion and declare the transformation.
6948 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
6949 internal error when the reference is in a wrong context. Do not
6950 create a debug decl by default.
6951 (note_nonlocal_block_vlas): Delete.
6952 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
6953 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
6954 create a debug decl by default.
6955 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
6956 call is in a wrong context.
6957 (fixup_vla_decls): New function.
6958 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
6959 debug variables were created.
6960 * tree.c (decl_value_expr_lookup): Add checking assertion.
6961 (decl_value_expr_insert): Likewise.
6963 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
6966 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
6967 * tree-scalar-evolution.c (interpret_expr): Likewise.
6968 (expression_expensive_p): Likewise.
6969 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
6970 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
6971 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
6972 (ssa_defined_by_minus_one_stmt_p): New.
6974 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
6977 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
6978 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
6979 * dojump.c (do_jump): Likewise.
6980 * expr.c (expand_expr_real_2): Check operand type's sign.
6981 * fold-const.c (const_unop): Handle ABSU_EXPR.
6982 (fold_abs_const): Likewise.
6983 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
6984 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
6985 (strip_sign_op_1): Likesise.
6986 * match.pd: Add new pattern to generate ABSU_EXPR.
6987 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
6988 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6989 * tree-eh.c (operation_could_trap_helper_p): Likewise.
6990 * tree-inline.c (estimate_operator_cost): Likewise.
6991 * tree-pretty-print.c (dump_generic_node): Likewise.
6992 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
6993 * tree.def (ABSU_EXPR): New.
6995 2018-06-16 Jakub Jelinek <jakub@redhat.com>
6998 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
6999 documented as preserved for backward compatibility only.
7000 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
7002 PR rtl-optimization/86108
7003 * bb-reorder.c (create_forwarder_block): Renamed to ...
7004 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
7005 jump from new landing pad to the second part.
7006 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
7009 2018-06-15 Jakub Jelinek <jakub@redhat.com>
7012 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
7013 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
7014 Only call store_expr for halves if the mode is the same.
7017 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
7018 Fix up comment formatting.
7020 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7022 * typed-splay-tree.h (typed_splay_tree::remove): New function.
7023 (typed_splay_tree::closure,
7024 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
7025 (typed_splay_tree::typed_splay_tree,
7026 typed_splay_tree::operator =): Declared private.
7027 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
7028 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
7029 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
7030 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
7031 typed_splay_tree::splay_tree_splay,
7032 typed_splay_tree::splay_tree_foreach_helper,
7033 typed_splay_tree::splay_tree_insert,
7034 typed_splay_tree::splay_tree_remove,
7035 typed_splay_tree::splay_tree_lookup,
7036 typed_splay_tree::splay_tree_predecessor,
7037 typed_splay_tree::splay_tree_successor,
7038 typed_splay_tree::splay_tree_min,
7039 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
7040 (typed_splay_tree::root, typed_splay_tree::comp,
7041 typed_splay_tree::delete_key,
7042 typed_splay_tree::delete_value): New data members.
7043 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
7044 typed_splay_tree::remove.
7046 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
7048 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
7049 -mginv and -mno-ginv to the assembler.
7050 * config/mips/mips.opt (-mcrc): New option.
7052 * doc/invoke.text (-mcrc): Document.
7055 2018-06-15 Nick Clifton <nickc@redhat.com>
7058 * tree.c (escaped_string): New class. Converts an unescaped
7059 string into its escaped equivalent.
7060 (warn_deprecated_use): Use the new class to convert the
7061 deprecation message, if present.
7062 (test_escaped_strings): New self test.
7063 (test_c_tests): Add test_escaped_strings.
7064 * doc/extend.texi (deprecated): Add a note that the
7065 deprecation message is affected by the -fmessage-length
7066 option, and that control characters will be escaped.
7067 (#pragma GCC error): Document this pragma.
7068 (#pragma GCC warning): Likewise.
7069 * doc/invoke.texi (-fmessage-length): Document this option's
7070 effect on the #warning and #error preprocessor directives and
7071 the deprecated attribute.
7073 2018-06-15 Richard Biener <rguenther@suse.de>
7075 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
7076 here, also noting vector size used.
7077 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
7078 size used in MSG_OPTIMIZED_LOCATIONS dump.
7079 (pass_slp_vectorize::execute): Adjust.
7081 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
7084 * config/arc/arc.c (arc_return_address_register): Fix
7087 2018-06-15 Richard Biener <rguenther@suse.de>
7090 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
7091 leave useless conversion stripping to force_gimple_operand_gsi.
7092 (gimplify_build2): Likewise.
7093 (gimplify_build1): Likewise.
7095 2018-06-15 Richard Biener <rguenther@suse.de>
7098 * tree-cfg.c (move_stmt_op): unshare invariant addresses
7099 before adjusting their block.
7101 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
7103 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
7104 multilibs for *-*-rtems*.
7105 * config/riscv/t-rtems: New file.
7107 2018-06-14 Jakub Jelinek <jakub@redhat.com>
7110 * match.pd ((A +- CST1) +- CST2): Punt if last resort
7111 unsigned_type_for returns NULL.
7114 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
7115 subregs of multi-word pseudos unless the float mode has word size.
7117 2018-06-14 Richard Biener <rguenther@suse.de>
7120 * tree-vect-generic.c (build_word_mode_vector_type): Remove
7121 duplicate and harmful type_hash_canon.
7122 * tree.c (type_hash_canon): Assert we didn't find ourselves.
7124 2018-06-14 Richard Biener <rguenther@suse.de>
7127 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
7130 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
7132 * config/rtems.h (STDINT_LONG32): Define.
7134 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
7135 Prachi Godbole <prachi.godbole@imgtec.com>
7137 * config/mips/mips-cpus.def: Define P6600.
7138 * config/mips/mips-tables.opt: Regenerate.
7139 * config/mips/mips.c (mips_ucbranch_type): New enum.
7140 (mips_rtx_cost_data): Add support for P6600.
7141 (mips_issue_rate): Likewise.
7142 (mips_multipass_dfa_lookahead): Likewise.
7143 (mips_avoid_hazard): Likewise.
7144 (mips_reorg_process_insns): Likewise.
7145 (mips_classify_branch_p6600): New function.
7146 * config/mips/mips.h (TUNE_P6600): New define.
7147 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
7148 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
7149 * config/mips/mips.md: Include p6600.md.
7150 (processor): Add p6600.
7151 * config/mips/p6600.md: New file.
7152 * doc/invoke.texi: Add p6600 to supported architectures.
7154 2018-06-13 Martin Sebor <msebor@redhat.com>
7156 PR tree-optimization/86114
7157 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
7159 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
7161 2018-06-13 Richard Biener <rguenther@suse.de>
7163 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
7164 Properly set vector type of the intermediate stmt.
7165 * tree-vect-stmts.c (vectorizable_operation): The destination
7166 var always has vectype_out type.
7168 2018-06-13 Jeff Law <law@redhat.com>
7170 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
7171 integer 0 for argument to print_rtl_with_bb.
7172 (rl78_reorg): Likewise.
7174 2018-06-13 David Malcolm <dmalcolm@redhat.com>
7176 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
7177 from rtx to rtx_insn *.
7178 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
7180 (add_sched_insns_for_speculation): Likewise for local "target",
7181 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
7182 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
7183 from rtx_insn ** to rtx_code_label **.
7184 (reorg_emit_nops): Likewise.
7185 (c6x_reorg): Likewise for local "call_labels".
7186 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
7188 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
7189 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
7190 the loops over LABEL_REFS.
7191 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
7193 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
7194 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
7195 (split_branches): Strengthen local "olabel" from rtx to
7196 rtx_insn *, adding a safe_as_a cast.
7197 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
7199 (add_insn_after): Likewise for first two params.
7200 (add_insn_before): Likewise.
7201 (remove_insn): Likewise for param.
7202 (emit_pattern_before_noloc): Likewise for second and third params.
7203 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
7204 (emit_call_insn_before_noloc): Likewise.
7205 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
7207 (emit_barrier_before): Likewise.
7208 (emit_label_before): Strengthen "label" param from "rtx" to
7209 "rtx_code_label *". Strengthen "before" param from "rtx" to
7211 (emit_insn_after_1): Strengthen "after" param from "rtx" to
7213 (emit_pattern_after_noloc): Likewise.
7214 (emit_insn_after_noloc): Likewise.
7215 (emit_jump_insn_after_noloc): Likewise.
7216 (emit_call_insn_after_noloc): Likewise.
7217 (emit_debug_insn_after_noloc): Likewise.
7218 (emit_barrier_after): Likewise.
7219 (emit_label_after): Likewise for both params.
7220 (emit_pattern_after_setloc): Likewise for "after" param. Convert
7221 "loc" param from "int" to "location_t".
7222 (emit_insn_after_setloc): Likewise.
7223 (emit_jump_insn_after_setloc): Likewise.
7224 (emit_call_insn_after_setloc): Likewise.
7225 (emit_debug_insn_after_setloc): Likewise.
7226 (emit_pattern_before_setloc): Likewise for "before" param. Convert
7227 "loc" param from "int" to "location_t".
7228 (emit_pattern_before): Convert NULL_RTX to NULL.
7229 (emit_insn_before_setloc): Convert "loc" param from "int" to
7231 (emit_jump_insn_before_setloc): Likewise.
7232 (emit_call_insn_before_setloc): Likewise.
7233 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
7234 rtx_insn *. Convert "loc" param from "int" to "location_t".
7235 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
7236 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
7237 Convert 3rd param from "int" to "location_t".
7238 (emit_barrier_before, emit_barrier_after, next_real_insn):
7239 Strengthen param from rtx to rtx_insn *.
7240 (emit_label_before): Strengthen 1st param from "rtx" to
7241 "rtx_code_label *". Strengthen 2nd param from "rtx" to
7243 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
7244 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
7245 Strengthen 2nd param from "rtx" to "rtx_insn *".
7246 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
7247 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
7248 Likewise. Convert 3rd param from "int" to "location_t".
7249 (emit_label_after): Strengthen 1st param from "rtx" to
7251 (next_real_insn, remove_insn): Strengthen param from "rtx" to
7253 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
7254 from "rtx" to "rtx_insn *".
7256 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
7258 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
7259 bodies streamed in with -Q.
7260 * dumpfile.c (dump_files): Add lto-stream-out dump file.
7261 * dumpfile.h (tree_dump_index): Add lto_stream_out.
7262 * gimple-streamer-out.c: Include gimple-pretty-print.h
7263 (output_bb): Dump stmts streamed.
7264 * lto-section-out.c: Include print-tree.h
7265 (lto_begin_section): Dump sections created.
7266 (lto_output_decl_index): Dump decl encoded.
7267 * lto-streamer-out.c: Include print-tree.h
7268 (create_output_block): Dump output block created.
7269 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
7270 (output_function): Dump function output.
7271 (output_constructor): Dump constructor streamed.
7272 (write_global_stream): Output indexes encoded.
7273 (produce_asm_for_decls): Dump streams encoded.
7274 * lto-streamer.c (streamer_dump_file): New global var.
7275 * lto-streamer.h (streamer_dump_file): Declare.
7276 * passes.c (ipa_write_summaries): Initialize streamer dump.
7277 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
7280 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
7283 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
7284 offsets for register save directives. Emit a second batch of save
7285 directives, if need be, when the function accesses prior frames.
7287 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7289 * config/arc/fpu.md (fmasf4): Force operand to register.
7290 (fnmasf4): Likewise.
7292 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7294 * config/arc/arc-protos.h (arc_pad_return): Remove.
7295 * config/arc/arc.c (machine_function): Remove force_short_suffix
7297 (arc_print_operand): Adjust printing of '&'.
7298 (arc_verify_short): Remove conditional printing of short suffix.
7299 (arc_final_prescan_insn): Remove reference to size_reason.
7300 (pad_return): New function.
7301 (arc_reorg): Call pad_return.
7302 (arc_pad_return): Remove.
7303 (arc_init_machine_status): Remove reference to force_short_suffix.
7304 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
7305 (attr length): When attribute iscompact is true force to 2
7306 regardless; in the case of maybe check if we want to force the
7307 instruction to have 4 bytes length.
7308 (nopv): Change it to generate 4 byte long nop as well.
7309 (blockage): New pattern.
7310 (simple_return): Remove call to arc_pad_return.
7311 (p_return_i): Likewise.
7313 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7315 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
7317 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7319 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
7322 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7324 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
7325 for ARC700 and ARCv2.
7327 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
7330 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
7331 operands[2] instead of operands[1].
7334 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
7336 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
7337 case, check whether the outer register overlaps an unallocatable
7338 register, not just whether it fits the required class.
7340 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
7342 * poly-int.h (can_div_trunc_p): Add new overload in which all values
7344 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
7345 (memrefs_conflict_p): Likewise.
7346 (init_alias_analysis): Likewise.
7347 * cfgexpand.c (expand_debug_expr): Likewise.
7348 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
7349 * cse.c (fold_rtx): Likewise.
7350 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
7351 * expr.c (emit_block_move_hints): Likewise.
7352 (clear_storage_hints, push_block, emit_push_insn): Likewise.
7353 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
7354 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
7355 (emit_group_store): Likewise.
7356 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
7357 to read the PRE/POST_MODIFY increment.
7358 * calls.c (store_one_arg): Use strip_offset.
7359 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
7361 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
7363 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
7364 (simplify_binary_operation_1): Extend CONST_INT handling to
7366 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
7367 than a HOST_WIDE_INT.
7368 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
7370 (adjust_mems, add_stores): Update accodingly.
7371 (vt_canonicalize_addr): Track polynomial offsets.
7372 (emit_note_insn_var_location): Likewise.
7373 (vt_add_function_parameter): Likewise.
7374 (vt_initialize): Likewise.
7376 2018-06-12 Jeff Law <law@redhat.com>
7378 * config.gcc (alpha*-*-freebsd*): Remove.
7379 * config/alpha/freebsd.h: Remove.
7381 2018-06-12 David Malcolm <dmalcolm@redhat.com>
7384 * spellcheck-tree.c (levenshtein_distance): Rename to...
7385 (get_edit_distance): ...this, and update for underlying renaming.
7386 * spellcheck-tree.h (levenshtein_distance): Rename to...
7387 (get_edit_distance): ...this.
7388 * spellcheck.c (levenshtein_distance): Rename to...
7389 (get_edit_distance): ...this. Convert from Levenshtein distance
7390 to Damerau-Levenshtein distance by supporting transpositions of
7391 adjacent characters. Rename "v1" to "v_next" and "v0" to
7393 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
7394 (selftest::test_edit_distance_unit_test_oneway): ...this, and
7395 update for underlying renaming.
7396 (selftest::levenshtein_distance_unit_test): Rename to...
7397 (selftest::test_get_edit_distance_unit): ...this, and update for
7398 underlying renaming.
7399 (selftest::test_find_closest_string): Add example from PR 69968
7400 where transposition helps
7401 (selftest::test_metric_conditions): Update for renaming.
7402 (selftest::test_metric_conditions): Likewise.
7403 (selftest::spellcheck_c_tests): Likewise.
7404 * spellcheck.h (levenshtein_distance): Rename both overloads to...
7405 (get_edit_distance): ...this.
7406 (best_match::consider): Update for renaming.
7408 2018-06-12 Martin Sebor <msebor@redhat.com>
7410 PR tree-optimization/85259
7411 * builtins.c (compute_objsize): Handle constant offsets.
7412 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
7413 true iff a warning has been issued.
7414 * gimple.h (gimple_nonartificial_location): New function.
7415 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
7416 gimple_nonartificial_location and handle -Wno-system-headers.
7417 (handle_builtin_stxncpy): Same.
7419 2018-06-12 Martin Sebor <msebor@redhat.com>
7422 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
7424 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
7426 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7427 BUILTIN_VEC_XST entries for pointer to double and long long.
7429 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
7432 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
7434 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
7437 2018-06-12 Martin Liska <mliska@suse.cz>
7439 * doc/options.texi: Document IntegerRange.
7441 2018-06-12 Martin Liska <mliska@suse.cz>
7443 * config/i386/i386.opt: Make MPX-related options as Deprecated.
7444 * opt-functions.awk: Handle Deprecated flag.
7445 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
7447 (read_cmdline_option): Report warning for a deprecated option.
7448 * opts.h (struct cl_option): Add new field cl_deprecated.
7449 (CL_ERR_DEPRECATED): New.
7451 2018-06-12 Martin Liska <mliska@suse.cz>
7453 * doc/options.texi: Document Deprecated option flag.
7455 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7457 * config/arc/arc-arch.h (arc_extras): New enum.
7458 (arc_cpu_t):Add field extra.
7459 (arc_cpu_types): Consider the extras.
7460 * config/arc/arc-cpus.def: Add extras info.
7461 * config/arc/arc-opts.h (processor_type): Consider extra field.
7462 * config/arc/arc.c (arc_override_options): Handle extra field.
7464 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7466 * config/arc/arc-arch.h: Update ARC_OPTX macro.
7467 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
7469 * config/arc/arc.c (arc_init): Update pic warning.
7470 (irq_range): Update irq range parsing warnings.
7471 (arc_override_options): Update various warning messages.
7472 (arc_handle_aux_attribute): Likewise.
7474 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
7476 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
7478 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7480 * doc/sourcebuild.texi: Document usage of line number 0 in verify
7481 compiler messages directives.
7483 2018-06-12 Matthew Fortune <mfortune@gmail.com>
7485 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
7486 * config/mips/mips-tables.opt: Regenerate.
7487 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
7489 * doc/invoke.texi: Document -march=i6500.
7491 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
7493 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
7494 (i6400_gpmul): Add cpu_unit.
7495 (i6400_gpdiv): Likewise.
7496 (i6400_msa_add_d): Update reservations.
7497 (i6400_msa_int_add) Likewise.
7498 (i6400_msa_short_logic3) Likewise.
7499 (i6400_msa_short_logic2) Likewise.
7500 (i6400_msa_short_logic) Likewise.
7501 (i6400_msa_move) Likewise.
7502 (i6400_msa_cmp) Likewise.
7503 (i6400_msa_short_float2) Likewise.
7504 (i6400_msa_div_d) Likewise.
7505 (i6400_msa_long_logic1) Likewise.
7506 (i6400_msa_long_logic2) Likewise.
7507 (i6400_msa_mult) Likewise.
7508 (i6400_msa_long_float2) Likewise.
7509 (i6400_msa_long_float4) Likewise.
7510 (i6400_msa_long_float5) Likewise.
7511 (i6400_msa_long_float8) Likewise.
7512 (i6400_fpu_fadd): Include frint type.
7513 (i6400_fpu_store): New define_insn_reservation.
7514 (i6400_fpu_load): Likewise.
7515 (i6400_fpu_move): Likewise.
7516 (i6400_fpu_fcmp): Likewise.
7517 (i6400_fpu_fmadd): Likewise.
7518 (i6400_int_mult): Include imul3nc type and update reservation.
7519 (i6400_int_div): Include idiv3 type and update reservation.
7520 (i6400_int_load): Update to check type not move_type.
7521 (i6400_int_store): Likewise.
7522 (i6400_int_prefetch): Set zero latency.
7524 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
7526 * gcc.c: Document new %@{...} sequence.
7527 (LINK_COMMAND_SPEC): Use it for the -L switches.
7528 (cpp_unique_options): Use it for the -I switches.
7529 (at_file_argbuf): New global variable.
7530 (in_at_file): Likewise.
7531 (alloc_args): Create at_file_argbuf.
7532 (clear_args): Truncate at_file_argbuf.
7533 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
7534 (open_at_file): New function.
7535 (close_at_file): Likewise.
7536 (create_at_file): Delete.
7537 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
7540 (validate_switches_from_spec): Deal with %@{...} sequence.
7541 (validate_switches): Likewise.
7542 (driver::finalize): Call clear_args.
7544 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
7546 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
7548 2018-06-11 Martin Sebor <msebor@redhat.com>
7550 * doc/invoke.texi (-Wall): List -Wc++17-compat.
7551 (Wno-class-memaccess): Add @opindex.
7552 (Wno-templates, Wno-multiple-inheritance): Same.
7553 (Wno-virtual-inheritance, Wno-namespaces): Same.
7554 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
7555 (Wno-format-overflow, Wno-format-truncation): Same.
7556 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
7557 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
7558 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
7559 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
7560 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
7561 (Wno-misspelled-isr): Same.
7563 2018-06-11 Martin Sebor <msebor@redhat.com>
7565 * PR tree-optimization/86083
7566 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
7568 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
7570 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
7572 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
7575 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
7576 on the correct operand.
7577 (*movdi_internal64): Ditto.
7579 2018-06-11 Martin Liska <mliska@suse.cz>
7581 PR tree-optimization/86089
7582 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
7584 2018-06-11 Julia Koval <julia.koval@intel.com>
7586 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
7587 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
7588 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
7590 2018-06-11 Olivier Hainque <hainque@adacore.com>
7592 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
7593 for Ada with strict dwarf2.
7595 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
7598 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
7601 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
7603 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
7605 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
7607 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
7610 2018-06-08 Martin Liska <mliska@suse.cz>
7612 * tree-cfg.h (debug_function): Fix argument type to match
7615 2018-06-08 Martin Liska <mliska@suse.cz>
7617 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
7618 Remove usage of MPX-related (and removed) fields.
7619 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
7621 2018-06-08 David Malcolm <dmalcolm@redhat.com>
7623 * cfg.c (debug): Use TDF_NONE rather than 0.
7624 * cfghooks.c (debug): Likewise.
7625 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
7626 (struct dump_option_value_info): Convert to...
7627 (struct kv_pair): ...this template type.
7628 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
7630 (optinfo_verbosity_options): Likewise.
7631 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
7633 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
7634 than int for "optgroup_flags" param.
7635 (dump_generic_expr_loc): Use dump_flags_t rather than int for
7637 (dump_dec): Likewise.
7638 (dump_finish): Use TDF_NONE rather than 0.
7639 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
7640 rather than int for "optgroup_flags" param. Use TDF_NONE rather
7641 than 0. Update for change to option_ptr.
7642 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
7643 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
7644 0. Update for changes to optinfo_verbosity_options and
7646 (opt_info_switch_p): Convert optgroup_flags from int to
7648 (dump_basic_block): Use dump_flags_t rather than int
7649 for "dump_kind" param.
7650 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
7651 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
7652 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
7653 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
7654 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
7655 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
7656 TDF_NONE): Convert from macros to...
7657 (enum dump_flag): ...this new enum.
7658 (dump_flags_t): Update to use enum.
7659 (operator|, operator&, operator~, operator|=, operator&=):
7660 Implement for dump_flags_t.
7661 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
7662 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
7663 Convert from macros to...
7664 (enum optgroup_flag): ...this new enum.
7665 (optgroup_flags_t): New typedef.
7666 (operator|, operator|=): Implement for optgroup_flags_t.
7667 (struct dump_file_info): Convert field "alt_flags" to
7668 dump_flags_t. Convert field "optgroup_flags" to
7670 (dump_basic_block): Use dump_flags_t rather than int for param.
7671 (dump_generic_expr_loc): Likewise.
7672 (dump_dec): Likewise.
7673 (dump_register): Convert param "optgroup_flags" to
7675 (opt_info_enable_passes): Likewise.
7676 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
7678 * gimple-pretty-print.c (debug): Likewise.
7679 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
7680 (merged_store_group::apply_stores): Likewise.
7681 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
7682 * gimple.c (verify_gimple_pp): Likewise.
7683 * graphite-poly.c (print_pbb_body): Likewise.
7684 * passes.c (pass_manager::register_one_dump_file): Convert
7685 local "optgroup_flags" to optgroup_flags_t.
7686 * print-tree.c (print_node): Use TDF_NONE rather than 0.
7688 (debug_body): Likewise.
7689 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
7690 to optgroup_flags_t.
7691 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
7693 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
7694 (convert_mult_to_fma): Likewise.
7695 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
7696 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
7697 * tree-vect-data-refs.c (dump_lower_bound): Convert param
7698 "dump_kind" to dump_flags_t.
7700 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
7702 * config/rs6000/rs6000.c (min, max): Delete.
7704 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
7706 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
7707 -mabi=spe and -mabi=no-spe.
7709 2018-06-08 Martin Liska <mliska@suse.cz>
7711 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
7712 where we expect an existing summary.
7714 2018-06-08 Martin Liska <mliska@suse.cz>
7716 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
7717 * ipa-inline.h (estimate_edge_growth): Likewise.
7719 2018-06-08 Martin Liska <mliska@suse.cz>
7721 * cgraph.c (function_version_hasher::hash): Use
7722 cgraph_node::get_uid ().
7723 (function_version_hasher::equal):
7724 * cgraph.h (cgraph_node::get_uid): New method.
7725 * ipa-inline.c (update_caller_keys): Use
7726 cgraph_node::get_uid ().
7727 (update_callee_keys): Likewise.
7728 * ipa-utils.c (searchc): Likewise.
7729 (ipa_reduced_postorder): Likewise.
7730 * lto-cgraph.c (input_node): Likewise.
7731 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
7732 * symbol-summary.h (symtab_insertion): Likewise.
7733 (symtab_removal): Likewise.
7734 (symtab_duplication): Likewise.
7735 * tree-pretty-print.c (dump_function_header): Likewise.
7736 * tree-sra.c (convert_callers_for_node): Likewise.
7738 2018-06-08 Martin Liska <mliska@suse.cz>
7740 * cgraph.c (symbol_table::create_edge): Always assign a new
7742 (symbol_table::free_edge): Do not recycle numbers.
7743 * cgraph.h (cgraph_edge::get): New method.
7744 * symbol-summary.h (symtab_removal): Use it.
7745 (symtab_duplication): Likewise.
7746 (call_summary::hashable_uid): Remove.
7748 2018-06-08 Martin Liska <mliska@suse.cz>
7750 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
7751 (initialize_growth_caches): Remove.
7752 (free_growth_caches): Likewise.
7753 (do_estimate_edge_time): Use edge_growth_cache.
7754 (do_estimate_edge_size): Likewise.
7755 (do_estimate_edge_hints): Likewise.
7756 * ipa-inline.c (reset_edge_caches): Likewise.
7757 (recursive_inlining): Likewise.
7758 (inline_small_functions): Likewise.
7759 * ipa-inline.h (initialize_growth_caches): Remove.
7760 (estimate_edge_size): Likewise.
7761 (estimate_edge_time): Likewise.
7762 (estimate_edge_hints): Likewise.
7763 (reset_edge_growth_cache): Likewise.
7764 * symbol-summary.h (call_summary::remove): New method.
7766 2018-06-08 Martin Liska <mliska@suse.cz>
7768 * ipa-cp.c (class edge_clone_summary): New summary.
7769 (grow_edge_clone_vectors): Remove.
7770 (ipcp_edge_duplication_hook): Remove.
7771 (class edge_clone_summary_t): New call_summary class.
7772 (ipcp_edge_removal_hook): Remove.
7773 (edge_clone_summary_t::duplicate): New function.
7774 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
7775 (create_specialized_node): Likewise.
7776 (ipcp_driver): Initialize edge_clone_summaries and do not
7779 2018-06-08 Martin Liska <mliska@suse.cz>
7781 * symbol-summary.h (get): New function.
7782 (call_summary::m_initialize_when_cloning): New class member.
7784 2018-06-08 Martin Liska <mliska@suse.cz>
7786 * cgraph.c (cgraph_node::remove): Do not recycle uid.
7787 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
7788 (symbol_table::allocate_cgraph_symbol): Do not set uid.
7789 * passes.c (uid_hash_t): Record removed_nodes by their uids.
7790 (remove_cgraph_node_from_order): Use the removed_nodes set.
7791 (do_per_function_toporder): Likwise.
7792 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
7793 instead of summary_uid.
7794 (symtab_removal): Likewise.
7795 (symtab_duplication): Likewise.
7797 2018-06-08 Martin Liska <mliska@suse.cz>
7799 * ipa-cp.c (ipcp_store_bits_results): Use
7800 ipcp_transformation_sum.
7801 (ipcp_store_vr_results): Likewise.
7802 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
7804 (ipcp_transformation_initialize): ... this.
7805 (ipa_set_node_agg_value_chain):
7806 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
7807 (write_ipcp_transformation_info): Likewise.
7808 (read_ipcp_transformation_info): Likewise.
7809 (ipcp_update_bits): Likewise.
7810 (ipcp_update_vr): Likewise.
7811 (ipcp_transform_function): Likewise.
7812 * ipa-prop.h: Rename ipcp_transformation_summary to
7813 ipcp_transformation.
7814 (class ipcp_transformation_t): New function summary.
7815 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
7816 (ipa_get_agg_replacements_for_node): Likewise.
7818 2018-06-08 Martin Liska <mliska@suse.cz>
7820 * ipa-pure-const.c (struct funct_state_d): Do it class instead
7822 (class funct_state_summary_t): New function_summary class.
7823 (has_function_state): Remove.
7824 (get_function_state): Likewise.
7825 (set_function_state): Likewise.
7826 (add_new_function): Likewise.
7827 (funct_state_summary_t::insert): New function.
7828 (duplicate_node_data): Remove.
7829 (remove_node_data): Remove.
7830 (funct_state_summary_t::duplicate): New function.
7831 (register_hooks): Create new funct_state_summaries.
7832 (pure_const_generate_summary): Use it.
7833 (pure_const_write_summary): Likewise.
7834 (pure_const_read_summary): Likewise.
7835 (propagate_pure_const): Likewise.
7836 (propagate_nothrow): Likewise.
7837 (dump_malloc_lattice): Likewise.
7838 (propagate_malloc): Likewise.
7839 (execute): Do not register hooks, just remove summary
7841 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
7844 2018-06-08 Martin Liska <mliska@suse.cz>
7846 * ipa-reference.c (remove_node_data): Remove.
7847 (duplicate_node_data): Likewise.
7848 (class ipa_ref_var_info_summary_t): New class.
7849 (class ipa_ref_opt_summary_t): Likewise.
7850 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
7851 (get_reference_optimization_summary): Use
7852 ipa_ref_opt_sum_summaries.
7853 (set_reference_vars_info): Remove.
7854 (set_reference_optimization_summary): Likewise.
7855 (ipa_init): Create summaries.
7856 (init_function_info): Use function summary.
7857 (ipa_ref_opt_summary_t::duplicate): New function.
7858 (ipa_ref_opt_summary_t::remove): New function.
7859 (get_read_write_all_from_node): Fix GNU coding style.
7860 (propagate): Use function summary.
7861 (write_node_summary_p): Fix GNU coding style.
7862 (stream_out_bitmap): Likewise.
7863 (ipa_reference_read_optimization_summary): Use function summary.
7864 (ipa_reference_c_finalize): Do not release hooks.
7866 2018-06-08 Martin Liska <mliska@suse.cz>
7868 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
7869 (analyze_function_body): Extract multiple calls of get_create.
7870 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
7871 * ipa-inline.c (recursive_inlining): Use ::get method.
7872 * ipa-inline.h (estimate_edge_growth): Likewise.
7874 2018-06-08 Martin Liska <mliska@suse.cz>
7876 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
7878 (hsa_function_summary::hsa_function_summary): Use the new enum
7880 (hsa_gpu_implementation_p): Use hsa_summaries::get.
7881 * hsa-gen.c (hsa_get_host_function): Likewise.
7882 (get_brig_function_name): Likewise.
7883 * ipa-hsa.c (process_hsa_functions): Likewise.
7884 (ipa_hsa_write_summary): Likewise.
7885 * symbol-summary.h (symtab_duplication): Use ::get function/
7886 (get): New function.
7888 2018-06-08 Martin Liska <mliska@suse.cz>
7890 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
7892 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
7893 (hsa_register_kernel): Likewise.
7894 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
7895 * hsa-gen.c (hsa_get_host_function): Likewise.
7896 (get_brig_function_name): Likewise.
7897 (generate_hsa): Likewise.
7898 (pass_gen_hsail::execute): Likewise.
7899 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
7900 (devirtualization_time_bonus): Likewise.
7901 (ipcp_propagate_stage): Likewise.
7902 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
7903 (edge_set_predicate): Likewise.
7904 (evaluate_conditions_for_known_args): Likewise.
7905 (evaluate_properties_for_edge): Likewise.
7906 (ipa_fn_summary::reset): Likewise.
7907 (ipa_fn_summary_t::duplicate): Likewise.
7908 (dump_ipa_call_summary): Likewise.
7909 (ipa_dump_fn_summary): Likewise.
7910 (analyze_function_body): Likewise.
7911 (compute_fn_summary): Likewise.
7912 (estimate_edge_devirt_benefit): Likewise.
7913 (estimate_edge_size_and_time): Likewise.
7914 (estimate_calls_size_and_time): Likewise.
7915 (estimate_node_size_and_time): Likewise.
7916 (inline_update_callee_summaries): Likewise.
7917 (remap_edge_change_prob): Likewise.
7918 (remap_edge_summaries): Likewise.
7919 (ipa_merge_fn_summary_after_inlining): Likewise.
7920 (ipa_update_overall_fn_summary): Likewise.
7921 (read_ipa_call_summary): Likewise.
7922 (inline_read_section): Likewise.
7923 (write_ipa_call_summary): Likewise.
7924 (ipa_fn_summary_write): Likewise.
7925 (ipa_free_fn_summary): Likewise.
7926 * ipa-hsa.c (process_hsa_functions): Likewise.
7927 (ipa_hsa_write_summary): Likewise.
7928 (ipa_hsa_read_section): Likewise.
7929 * ipa-icf.c (sem_function::merge): Likewise.
7930 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
7931 (do_estimate_edge_time): Likewise.
7932 (estimate_size_after_inlining): Likewise.
7933 (estimate_growth): Likewise.
7934 (growth_likely_positive): Likewise.
7935 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
7936 (inline_call): Likewise.
7937 * ipa-inline.c (caller_growth_limits): Likewise.
7938 (can_inline_edge_p): Likewise.
7939 (can_inline_edge_by_limits_p): Likewise.
7940 (compute_uninlined_call_time): Likewise.
7941 (compute_inlined_call_time): Likewise.
7942 (want_inline_small_function_p): Likewise.
7943 (edge_badness): Likewise.
7944 (update_caller_keys): Likewise.
7945 (update_callee_keys): Likewise.
7946 (recursive_inlining): Likewise.
7947 (inline_small_functions): Likewise.
7948 (inline_to_all_callers_1): Likewise.
7949 (dump_overall_stats): Likewise.
7950 (early_inline_small_functions): Likewise.
7951 (early_inliner): Likewise.
7952 * ipa-inline.h (estimate_edge_growth): Likewise.
7953 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
7954 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7955 * ipa-prop.h (IPA_NODE_REF): Likewise.
7956 (IPA_EDGE_REF): Likewise.
7957 * ipa-pure-const.c (malloc_candidate_p): Likewise.
7958 (propagate_malloc): Likewise.
7959 * ipa-split.c (execute_split_functions): Likewise.
7960 * symbol-summary.h: Rename get to get_create.
7962 (get_create): Likewise.
7963 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
7965 2018-06-08 Martin Liska <mliska@suse.cz>
7967 * symbol-summary.h (release): Move definition out of class
7969 (symtab_removal): Likewise.
7970 (symtab_duplication): Likewise.
7972 2018-06-08 Martin Liska <mliska@suse.cz>
7974 * symbol-summary.h (function_summary): Move constructor
7975 implementation out of class declaration.
7976 (release): Likewise.
7977 (symtab_insertion): Likewise.
7978 (symtab_removal): Likewise.
7979 (symtab_duplication): Likewise.
7982 2018-06-08 Martin Liska <mliska@suse.cz>
7984 * Makefile.in: Remove support for MPX (macros, related functions,
7985 fields in cgraph_node, ...).
7986 * builtin-types.def (BT_BND): Likewise.
7987 (BT_FN_BND_CONST_PTR): Likewise.
7988 (BT_FN_CONST_PTR_BND): Likewise.
7989 (BT_FN_VOID_PTR_BND): Likewise.
7990 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
7991 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
7992 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
7993 (expand_builtin_mempcpy_with_bounds): Likewise.
7994 (expand_builtin_memset_with_bounds): Likewise.
7995 (expand_builtin_memset_args): Likewise.
7996 (std_expand_builtin_va_start): Likewise.
7997 (expand_builtin): Likewise.
7998 (expand_builtin_with_bounds): Likewise.
7999 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
8000 (DEF_LIB_BUILTIN_CHKP): Likewise.
8001 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
8002 (DEF_CHKP_BUILTIN): Likewise.
8003 (BUILT_IN_MEMCPY): Likewise.
8004 (BUILT_IN_MEMMOVE): Likewise.
8005 (BUILT_IN_MEMPCPY): Likewise.
8006 (BUILT_IN_MEMSET): Likewise.
8007 (BUILT_IN_STPCPY): Likewise.
8008 (BUILT_IN_STRCAT): Likewise.
8009 (BUILT_IN_STRCHR): Likewise.
8010 (BUILT_IN_STRCPY): Likewise.
8011 (BUILT_IN_STRLEN): Likewise.
8012 (BUILT_IN_MEMCPY_CHK): Likewise.
8013 (BUILT_IN_MEMMOVE_CHK): Likewise.
8014 (BUILT_IN_MEMPCPY_CHK): Likewise.
8015 (BUILT_IN_MEMSET_CHK): Likewise.
8016 (BUILT_IN_STPCPY_CHK): Likewise.
8017 (BUILT_IN_STRCAT_CHK): Likewise.
8018 (BUILT_IN_STRCPY_CHK): Likewise.
8019 * calls.c (store_bounds): Likewise.
8020 (emit_call_1): Likewise.
8021 (special_function_p): Likewise.
8022 (maybe_warn_nonstring_arg): Likewise.
8023 (initialize_argument_information): Likewise.
8024 (finalize_must_preallocate): Likewise.
8025 (compute_argument_addresses): Likewise.
8026 (expand_call): Likewise.
8027 * cfgexpand.c (expand_call_stmt): Likewise.
8028 (expand_return): Likewise.
8029 (expand_gimple_stmt_1): Likewise.
8030 (pass_expand::execute): Likewise.
8031 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
8032 (cgraph_node::remove): Likewise.
8033 (cgraph_node::dump): Likewise.
8034 (cgraph_node::verify_node): Likewise.
8035 * cgraph.h (chkp_function_instrumented_p): Likewise.
8036 (symtab_node::get_alias_target): Likewise.
8037 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
8038 (cgraph_local_p): Likewise.
8039 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
8040 (cgraph_edge::rebuild_references): Likewise.
8041 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
8042 (walk_polymorphic_call_targets): Likewise.
8043 (cgraph_node::expand_thunk): Likewise.
8044 (symbol_table::output_weakrefs): Likewise.
8045 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
8046 (ix86_handle_option): Likewise.
8047 * config/i386/constraints.md: Likewise.
8048 * config/i386/i386-builtin-types.def (BND): Likewise.
8052 * config/i386/i386-builtin.def (BDESC_END): Likewise.
8053 (BDESC_FIRST): Likewise.
8055 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
8056 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
8057 * config/i386/i386.c (enum reg_class): Likewise.
8058 (ix86_target_string): Likewise.
8059 (ix86_option_override_internal): Likewise.
8060 (ix86_conditional_register_usage): Likewise.
8061 (ix86_valid_target_attribute_inner_p): Likewise.
8062 (ix86_set_indirect_branch_type): Likewise.
8063 (ix86_set_current_function): Likewise.
8064 (ix86_function_arg_regno_p): Likewise.
8065 (init_cumulative_args): Likewise.
8066 (ix86_function_arg_advance): Likewise.
8067 (ix86_function_arg): Likewise.
8068 (ix86_pass_by_reference): Likewise.
8069 (ix86_function_value_regno_p): Likewise.
8070 (ix86_function_value_1): Likewise.
8071 (ix86_function_value_bounds): Likewise.
8072 (ix86_return_in_memory): Likewise.
8073 (ix86_setup_incoming_vararg_bounds): Likewise.
8074 (ix86_va_start): Likewise.
8075 (indirect_thunk_need_prefix): Likewise.
8076 (print_reg): Likewise.
8077 (ix86_print_operand): Likewise.
8078 (ix86_expand_call): Likewise.
8079 (ix86_output_function_return): Likewise.
8080 (reg_encoded_number): Likewise.
8081 (BDESC_VERIFYS): Likewise.
8082 (ix86_init_mpx_builtins): Likewise.
8083 (ix86_init_builtins): Likewise.
8084 (ix86_emit_cmove): Likewise.
8085 (ix86_emit_move_max): Likewise.
8086 (ix86_expand_builtin): Likewise.
8087 (ix86_builtin_mpx_function): Likewise.
8088 (ix86_get_arg_address_for_bt): Likewise.
8089 (ix86_load_bounds): Likewise.
8090 (ix86_store_bounds): Likewise.
8091 (ix86_load_returned_bounds): Likewise.
8092 (ix86_store_returned_bounds): Likewise.
8093 (ix86_class_likely_spilled_p): Likewise.
8094 (ix86_hard_regno_mode_ok): Likewise.
8095 (x86_order_regs_for_local_alloc): Likewise.
8096 (ix86_mitigate_rop): Likewise.
8097 (ix86_bnd_prefixed_insn_p): Likewise.
8098 (ix86_mpx_bound_mode): Likewise.
8099 (ix86_make_bounds_constant): Likewise.
8100 (ix86_initialize_bounds): Likewise.
8101 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
8102 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
8103 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
8104 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
8105 (TARGET_CHKP_BOUND_MODE): Likewise.
8106 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
8107 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
8108 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
8109 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
8110 * config/i386/i386.h (TARGET_MPX): Likewise.
8111 (TARGET_MPX_P): Likewise.
8112 (VALID_BND_REG_MODE): Likewise.
8113 (FIRST_BND_REG): Likewise.
8114 (LAST_BND_REG): Likewise.
8115 (enum reg_class): Likewise.
8116 (BND_REG_P): Likewise.
8117 (BND_REGNO_P): Likewise.
8118 (BNDmode): Likewise.
8119 (ADJUST_INSN_LENGTH): Likewise.
8120 * config/i386/i386.md: Likewise.
8121 * config/i386/i386.opt: Likewise.
8122 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
8123 (defined): Likewise.
8124 (LINK_MPX): Likewise.
8125 (MPX_SPEC): Likewise.
8126 (LIBMPX_SPEC): Likewise.
8127 (LIBMPXWRAPPERS_SPEC): Likewise.
8128 (CHKP_SPEC): Likewise.
8129 * config/i386/predicates.md: Likewise.
8130 * dbxout.c (dbxout_type): Likewise.
8131 * doc/extend.texi: Likewise.
8132 * doc/invoke.texi: Likewise.
8133 * doc/md.texi: Likewise.
8134 * doc/tm.texi: Likewise.
8135 * doc/tm.texi.in: Likewise.
8136 * dwarf2out.c (is_base_type): Likewise.
8137 (gen_formal_types_die): Likewise.
8138 (gen_subprogram_die): Likewise.
8139 (gen_type_die_with_usage): Likewise.
8140 (gen_decl_die): Likewise.
8141 (dwarf2out_late_global_decl): Likewise.
8142 * expr.c (expand_assignment): Likewise.
8143 (emit_storent_insn): Likewise.
8144 (store_expr_with_bounds): Likewise.
8145 (store_expr): Likewise.
8146 (expand_expr_real_1): Likewise.
8147 * expr.h (store_expr_with_bounds): Likewise.
8148 * function.c (use_register_for_decl): Likewise.
8149 (struct bounds_parm_data): Likewise.
8150 (assign_parms_augmented_arg_list): Likewise.
8151 (assign_parm_find_entry_rtl): Likewise.
8152 (assign_parm_is_stack_parm): Likewise.
8153 (assign_parm_load_bounds): Likewise.
8154 (assign_bounds): Likewise.
8155 (assign_parms): Likewise.
8156 (expand_function_start): Likewise.
8157 * gcc.c (CHKP_SPEC): Likewise.
8158 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8159 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
8160 (wrestrict_dom_walker::check_call): Likewise.
8161 * gimple.c (gimple_build_call_from_tree): Likewise.
8162 * gimple.h (enum gf_mask): Likewise.
8163 (gimple_call_with_bounds_p): Likewise.
8164 (gimple_call_set_with_bounds): Likewise.
8165 * gimplify.c (gimplify_init_constructor): Likewise.
8166 * ipa-cp.c (initialize_node_lattices): Likewise.
8167 (propagate_constants_across_call): Likewise.
8168 (find_more_scalar_values_for_callers_subset): Likewise.
8169 * ipa-hsa.c (process_hsa_functions): Likewise.
8170 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
8171 * ipa-icf.c (sem_function::merge): Likewise.
8172 * ipa-inline.c (early_inliner): Likewise.
8173 * ipa-pure-const.c (warn_function_noreturn): Likewise.
8174 (warn_function_cold): Likewise.
8175 (propagate_pure_const): Likewise.
8176 * ipa-ref.h (enum GTY): Likewise.
8177 * ipa-split.c (find_retbnd): Likewise.
8178 (consider_split): Likewise.
8179 (split_function): Likewise.
8180 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
8181 * ipa.c (walk_polymorphic_call_targets): Likewise.
8182 (symbol_table::remove_unreachable_nodes): Likewise.
8183 (process_references): Likewise.
8184 (cgraph_build_static_cdtor_1): Likewise.
8185 * lto-cgraph.c (lto_output_node): Likewise.
8186 (output_refs): Likewise.
8187 (compute_ltrans_boundary): Likewise.
8188 (input_overwrite_node): Likewise.
8189 (input_node): Likewise.
8190 (input_cgraph_1): Likewise.
8191 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
8192 * passes.c (pass_manager::execute_early_local_passes): Likewise.
8193 (class pass_chkp_instrumentation_passes): Likewise.
8194 (make_pass_chkp_instrumentation_passes): Likewise.
8195 * passes.def: Likewise.
8196 * rtl.h (struct GTY): Likewise.
8197 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
8198 * stor-layout.c (layout_type): Likewise.
8199 * symtab.c: Likewise.
8200 * target.def: Likewise.
8201 * targhooks.c (default_chkp_bound_type): Likewise.
8202 (default_chkp_bound_mode): Likewise.
8203 (default_builtin_chkp_function): Likewise.
8204 (default_chkp_function_value_bounds): Likewise.
8205 (default_chkp_make_bounds_constant): Likewise.
8206 (default_chkp_initialize_bounds): Likewise.
8207 * targhooks.h (default_chkp_bound_type): Likewise.
8208 (default_chkp_bound_mode): Likewise.
8209 (default_builtin_chkp_function): Likewise.
8210 (default_chkp_function_value_bounds): Likewise.
8211 (default_chkp_make_bounds_constant): Likewise.
8212 (default_chkp_initialize_bounds): Likewise.
8213 * toplev.c (compile_file): Likewise.
8214 (process_options): Likewise.
8215 * tree-core.h (DEF_BUILTIN): Likewise.
8216 (DEF_BUILTIN_CHKP): Likewise.
8217 * tree-inline.c (declare_return_variable): Likewise.
8218 (remap_gimple_stmt): Likewise.
8219 (copy_bb): Likewise.
8220 (initialize_inlined_parameters): Likewise.
8221 (expand_call_inline): Likewise.
8222 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
8223 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
8224 (make_pass_ipa_chkp_produce_thunks): Likewise.
8225 (make_pass_chkp): Likewise.
8226 (make_pass_chkp_opt): Likewise.
8227 (make_pass_chkp_instrumentation_passes): Likewise.
8228 * tree-pretty-print.c (dump_generic_node): Likewise.
8229 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
8230 * tree-ssa-dce.c (propagate_necessity): Likewise.
8231 (eliminate_unnecessary_stmts): Likewise.
8232 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
8233 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
8234 * tree-ssa-sccvn.h: Likewise.
8235 * tree-ssa-strlen.c (get_string_length): Likewise.
8236 (valid_builtin_call): Likewise.
8237 (adjust_last_stmt): Likewise.
8238 (handle_builtin_strchr): Likewise.
8239 (handle_builtin_strcpy): Likewise.
8240 (handle_builtin_stxncpy): Likewise.
8241 (handle_builtin_memcpy): Likewise.
8242 (handle_builtin_strcat): Likewise.
8243 (strlen_check_and_optimize_stmt): Likewise.
8244 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
8245 * tree-streamer-in.c: Likewise.
8246 * tree-streamer.c (record_common_node): Likewise.
8247 * tree.c (tree_code_size): Likewise.
8248 (wide_int_to_tree_1): Likewise.
8249 (type_contains_placeholder_1): Likewise.
8250 (build_common_tree_nodes): Likewise.
8251 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
8252 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
8253 (POINTER_BOUNDS_P): Likewise.
8254 (BOUNDED_TYPE_P): Likewise.
8255 (BOUNDED_P): Likewise.
8256 (CALL_WITH_BOUNDS_P): Likewise.
8257 (pointer_bounds_type_node): Likewise.
8258 * value-prof.c (gimple_ic): Likewise.
8259 * var-tracking.c (vt_add_function_parameters): Likewise.
8260 * varasm.c (make_decl_rtl): Likewise.
8261 (assemble_start_function): Likewise.
8262 (output_constant): Likewise.
8263 (maybe_assemble_visibility): Likewise.
8264 * varpool.c (ctor_for_folding): Likewise.
8265 * chkp-builtins.def: Remove.
8266 * ipa-chkp.c: Remove.
8267 * ipa-chkp.h: Remove.
8268 * rtl-chkp.c: Remove.
8269 * rtl-chkp.h: Remove.
8270 * tree-chkp-opt.c: Remove.
8271 * tree-chkp.c: Remove.
8272 * tree-chkp.h: Remove.
8274 2018-06-07 Carl Love <cel@us.ibm.com>
8276 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8277 vextract_fp_from_shortl): Add BE support.
8279 2018-06-07 Paul Koning <ni1d@arrl.net>
8281 * compare-elim.c (try_merge_compare): Don't merge compare if
8282 address contains a side effect.
8283 (try_eliminate_compare): Likewise.
8285 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
8287 * config.gcc: Support "tremont".
8288 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
8289 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8291 * config/i386/i386.c (m_TREMONT): Define.
8292 (processor_target_table): Add "tremont".
8293 (PTA_TREMONT): Define.
8294 (ix86_lea_outperforms): Add TARGET_TREMONT.
8295 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
8296 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
8297 and M_INTEL_GOLDMONT_PLUS.
8298 (fold_builtin_cpu): Add "tremont".
8299 (ix86_add_stmt_cost): Add TARGET_TREMONT.
8300 (ix86_option_override_internal): Add "tremont".
8301 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
8302 (processor_type): Add PROCESSOR_TREMONT.
8303 * config/i386/x86-tune.def: Add m_TREMONT.
8304 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
8306 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8308 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
8309 symbol defined for msp430i* devices to be lower case.
8311 2018-06-07 Richard Biener <rguenther@suse.de>
8313 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
8314 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
8315 Properly wrap signed arithmetic if overflow wraps.
8317 2018-06-07 Jakub Jelinek <jakub@redhat.com>
8319 PR tree-optimization/69615
8320 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
8321 of a cast from a same precision integral SSA_NAME in a bb dominated
8322 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
8323 cast to utype if rhs2 has already a compatible type.
8325 2018-06-07 Richard Biener <rguenther@suse.de>
8327 PR tree-optimization/85935
8328 * graphite-scop-detection.c (find_params_in_bb): Analyze
8329 condition operands with respect to the correct loop. Assert
8330 the analysis doesn't fail.
8332 2018-06-04 Carl Love <cel@us.ibm.com>
8334 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
8335 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
8336 as it is slightly cheaper.
8337 (first_match_or_eos_index_<mode>):
8338 Calculate index using natural element order.
8339 (first_match_index_<mode>):
8340 Calculate index using natural element order.
8341 (first_match_or_eos_index_<mode>):
8342 Calculate index using natural order.
8343 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
8344 for BE and LE modes.
8345 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
8346 P9V_BUILTIN_VCLZLSBB_V16QI.
8347 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
8350 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
8352 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
8353 indentation and line wrap for many prototypes. Add missing
8354 @smallexample directives around block of prototypes for vec_xl and
8357 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
8359 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
8360 track if we pass or return IEEE 128-bit floating point.
8361 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
8362 C++ mangling that is compatible with GCC 8.1.
8363 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
8364 (init_cumulative_args): Note if we pass or return IEEE 128-bit
8365 floating point types.
8366 (rs6000_function_arg_advance_1): Likewise.
8367 (rs6000_mangle_type): Optionally generate mangled names that match
8368 what GCC 8.1 generated for IEEE 128-bit floating point types.
8369 (rs6000_globalize_decl_name): If we have an external function that
8370 passes or returns IEEE 128-bit types, generate a weak reference
8371 from the mangled name used in GCC 8.1 to the current mangled
8373 (rs6000_init_builtins): Make __ibm128 use the long double type if
8374 long double is IBM extended double. Make __float128 use the long
8375 double type if long double is IEEE 128-bit.
8378 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
8379 macro for __ibm128 built-in functions.
8380 (PACK_IF): Add __ibm128 pack/unpack functions.
8381 (UNPACK_IF): Likewise.
8382 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
8383 enable long double built-in functions if long double is IEEE
8384 128-bit floating point.
8385 (rs6000_invalid_builtin): Update long double built-in function
8387 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
8388 functions, adjust the built-in function to use the long double
8389 built-in function if __ibm128 and long double are the same type.
8390 * doc/extend.texi (PowerPC builtins): Update documention for
8391 __builtin_{,un}pack_longdouble. Add documentation for
8392 __builtin_{,un}pack_ibm128.
8394 2018-06-06 Jim Wilson <jimw@sifive.com>
8396 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
8397 (struct machine_function): New field interrupt_mode.
8398 (riscv_handle_type_attribute): New function. Add forward declaration.
8399 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
8400 (riscv_expand_epilogue): Check interrupt_mode field.
8401 (riscv_set_current_function): Check interrupt attribute args and
8402 set interrupt_mode field.
8403 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
8404 (riscv_sret, riscv_uret): New.
8405 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
8406 new arguments to interrupt attribute.
8408 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
8411 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
8412 Don't handle -mcpu=power8 if -mpower9-vector is also used.
8414 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
8416 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
8417 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
8418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8419 several redundant entries.
8421 2018-06-06 David Malcolm <dmalcolm@redhat.com>
8423 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
8424 type from "rtx" to "rtx_insn *".
8425 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
8426 for local "call_insn", removing cast.
8427 (ix86_expand_call): Likewise, introducing a "call_insn" local.
8429 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
8431 PR tree-optimization/86066
8432 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
8433 for BIT_INSERT_EXPR stores.
8435 2018-06-06 Richard Biener <rguenther@suse.de>
8437 PR tree-optimization/86062
8438 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
8439 component refs ontop
8440 of to be offsetted base.
8442 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8444 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
8445 to be static and remove check on interrupt attribute name.
8447 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
8449 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
8450 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
8452 2018-06-05 Steve Ellcey <sellcey@cavium.com>
8455 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
8457 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
8458 Remove second argument, change how error is called.
8459 (aarch64_layout_arg): Remove second argument from
8460 aarch64_err_no_fpadvsimd call.
8461 (aarch64_init_cumulative_args): Ditto.
8462 (aarch64_gimplify_va_arg_expr): Ditto.
8463 * config/aarch64/aarch64.md (mov<mode>): Ditto.
8465 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
8467 * config/i386/i386.md (simple_return_indirect_internal): New expander.
8468 (*simple_return_indirect_internal<mode>): Rename from
8469 simple_return_indirect_internal. Use W mode iterator.
8470 (rstorssp): New expander.
8471 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
8472 (clrssbsy): New expander.
8473 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
8475 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
8477 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
8479 (cmse_check_pointed_object): Likewise.
8481 2018-06-05 Martin Liska <mliska@suse.cz>
8483 PR gcov-profile/47618
8484 * doc/invoke.texi: Document how -fprofile-dir format
8487 2018-06-05 Richard Biener <rguenther@suse.de>
8489 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
8490 removal pretend DOM info isn't available so we do not update
8491 it and only remove edges, not dominated blocks. Actually free
8492 DOM info in case we removed something. Remove unreachable blocks.
8493 (mfb_keep_latches): Work with either DOM info or marked backedges.
8494 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
8495 first. Mark backedges if DOM info isn't available.
8496 (Re-)compute DOM info after cleanup_control_flow_pre.
8498 2018-06-05 Richard Biener <rguenther@suse.de>
8500 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
8501 (locus_discrim_hasher::hash): Adjust.
8502 (locus_discrim_hasher::equal): Likewise.
8503 (next_discriminator_for_locus): Work on line directly.
8504 (same_line_p): Pass in expanded locus1 as well.
8505 (assign_discriminators): Avoid redundant location expansions.
8507 2018-06-05 Richard Biener <rguenther@suse.de>
8509 PR tree-optimization/86046
8510 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
8511 if required after clearing TREE_ADDRESSABLE.
8513 2018-06-05 Richard Biener <rguenther@suse.de>
8515 PR tree-optimization/86047
8516 * tree-ssa-loop.c (for_each_index): Glob handling of all
8517 decls and constants and really handle all of them.
8519 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8522 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
8523 qualifier_void_pointer and qualifier_const_void_pointer.
8524 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
8525 (arm_init_builtins): Handle the above.
8526 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
8527 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
8528 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
8531 2018-06-05 Martin Liska <mliska@suse.cz>
8533 * auto-profile.c (read_autofdo_file): Do not use
8534 gcov_ctr_summary struct.
8535 (afdo_callsite_hot_enough_for_early_inline): Likewise.
8536 * coverage.c (struct counts_entry): Likewise.
8537 (read_counts_file): Read just single summary entry.
8538 (get_coverage_counts): Use gcov_summary struct.
8539 * coverage.h (get_coverage_counts): Likewise.
8540 * gcov-dump.c (dump_working_sets): Likewise.
8541 (tag_summary): Dump just single summary.
8542 * gcov-io.c (gcov_write_summary): Write just histogram
8544 (gcov_read_summary): Read just single summary.
8545 (compute_working_sets): Use gcov_summary struct.
8546 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
8547 of GCOV_COUNTERS_SUMMABLE.
8548 (GCOV_COUNTERS_SUMMABLE): Remove.
8549 (GCOV_FIRST_VALUE_COUNTER): Replace with
8550 GCOV_COUNTER_V_INTERVAL.
8551 (struct gcov_ctr_summary): Remove.
8552 (struct gcov_summary): Directly use fields of former
8554 (compute_working_sets): Use gcov_summary struct.
8555 * gcov.c (read_count_file): Do not use ctrs fields.
8556 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
8558 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
8560 * profile.c: Likewise.
8561 * profile.h: Likewise.
8563 2018-06-05 Martin Liska <mliska@suse.cz>
8565 PR gcov-profile/84846
8566 * gcov.c (output_lines): Print working directory only
8567 in intermediate format.
8569 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
8571 * config/s390/s390-builtin-types.def: Add void function type.
8572 * config/s390/s390-builtins.def: Use the function type for the
8575 2018-06-04 Jim Wilson <jimw@sifive.com>
8577 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
8579 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
8580 and maybe_eh_return. Change regno to unsigned int. Use new args to
8581 handle EH_RETURN_DATA_REGNO registers properly.
8582 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
8583 (riscv_expand_epilogue): Update comment. Change argument name and
8584 type. Update code to use new name and type. Pass new args to
8585 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
8587 * config/riscv/riscv.md (NORMAL_RETURN): New.
8588 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
8589 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
8590 (eh_return): Call gen_eh_return_internal and emit barrier.
8591 (eh_return_internal): Call riscv_expand_epilogue.
8593 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
8595 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
8596 bit_insertion field and declare can_be_merged_into method.
8597 (merged_store_group::can_be_merged_into): New method.
8598 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
8599 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
8600 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
8602 2018-06-04 Richard Biener <rguenther@suse.de>
8604 PR tree-optimization/85955
8605 * builtins.c (fold_builtin_sincos): Convert pointers to
8606 destination to appropriate type before dereferencing.
8608 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
8610 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
8612 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
8614 * expr.c (expand_expr_real_1): Force the operand into memory if
8615 its TYPE_MODE is BLKmode and if there is no integer mode for
8616 the number of bits being extracted.
8618 2018-06-04 Jakub Jelinek <jakub@redhat.com>
8622 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
8623 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
8625 2018-06-04 Richard Biener <rguenther@suse.de>
8627 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
8628 (cleanup_tree_cfg_noloop): ... single caller. Do
8629 start_recording_case_labels later.
8631 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
8633 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
8634 to _IMMINTRIN_H_INCLUDED.
8635 * config/i386/pconfigintrin.h: Ditto.
8636 * config/i386/waitpkgintrin.h: Ditto.
8637 * config/i386/immintrin.h: Add includes for sgxintrin.h,
8638 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
8639 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
8640 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
8641 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
8642 waitpkgintrin.h and cldemoteintrin.h.
8644 2018-06-04 Richard Biener <rguenther@suse.de>
8646 PR tree-optimization/86038
8647 * tracer.c (find_best_successor): Check probability for
8648 being initialized, bail out if not.
8650 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
8653 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
8654 of bits to ignore when comparing architectures.
8656 2018-06-04 Jakub Jelinek <jakub@redhat.com>
8658 PR tree-optimization/69615
8659 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
8660 maximum or minimum of the type, try to merge it also as if
8661 range1 is + [-, x - 1] or + [x + 1, -].
8664 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
8666 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
8668 PR tree-optimization/86034
8669 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
8670 the unsigned bitfield type in a bit insertion sequence if it does not
8671 have a larger precision than the bitfield size.
8672 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
8674 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
8676 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
8678 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
8680 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
8681 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
8682 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
8683 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
8685 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
8687 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
8688 Disable -fdelete-null-pointer-checks for ELF toolchain.
8690 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
8691 Kito Cheng <kito.cheng@gmail.com>
8693 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
8694 (nds32le-*-*, nds32be-*-*): Integrate checking process.
8695 (nds32*-*-*): Add glibc and uclibc conditions.
8696 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
8697 (TARGET_EXCEPT_UNWIND_INFO): Define.
8698 * config/nds32/elf.h: New file.
8699 * config/nds32/linux.h: New file.
8700 * config/nds32/nds32-elf.opt: New file.
8701 * config/nds32/nds32-linux.opt: New file.
8702 * config/nds32/nds32-fp-as-gp.c
8703 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
8704 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
8706 (nds32_asm_file_end): Ditto.
8707 (nds32_print_operand): Ditto.
8708 (nds32_insert_attributes): Ditto.
8709 (nds32_init_libfuncs): New function.
8710 (TARGET_HAVE_TLS): Define.
8711 (TARGET_INIT_LIBFUNCS): Define.
8712 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
8714 (TARGET_ELF): Apply different mcmodel setting.
8715 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
8716 been migrated into elf.h and linux.h files.
8717 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
8718 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
8719 (mcmodel): The content has been migrated into nds32-elf.opt and
8720 nds32-linux.opt files.
8721 * config/nds32/t-elf: New file.
8722 * config/nds32/t-linux: New file.
8724 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
8725 Shiva Chen <shiva0217@gmail.com>
8727 * config/nds32/constants.md (unspec_volatile_element): Add
8728 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
8729 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
8731 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
8732 (make_pass_nds32_fp_as_gp): Declare.
8733 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
8735 (nds32_asm_function_end_prologue): Remove unused asm output.
8736 (nds32_asm_function_begin_epilogue): Remove unused asm output.
8737 (nds32_asm_file_start): Output necessary fp_as_gp information.
8738 (nds32_option_override): Adjust register usage.
8739 (nds32_expand_prologue): Consider fp_as_gp situation.
8740 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
8741 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
8744 (simple_return): Ditto.
8745 (omit_fp_begin): Output special directive for fp_as_gp.
8746 (omit_fp_end): Output special directive for fp_as_gp.
8747 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
8748 mforbid-fp-as-gp): New options.
8750 2018-06-01 Mark Wielaard <mark@klomp.org>
8752 * dwarf2out.c (dwarf2out_finish): Remove generation of
8753 DW_AT_loclists_base.
8755 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
8757 * gimple-ssa-store-merging.c: Include gimple-fold.h.
8758 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
8759 (struct merged_store_group): Add bit_insertion field.
8760 (dump_char_array): Use standard hexadecimal format.
8761 (merged_store_group::merged_store_group): Set bit_insertion to false.
8762 (merged_store_group::apply_stores): Use optimal buffer size. Deal
8763 with BIT_INSERT_EXPR stores. Move up code updating the mask and
8764 also print the mask in the dump file.
8765 (pass_store_merging::gate): Minor tweak.
8766 (imm_store_chain_info::coalesce_immediate): Fix wrong association
8767 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
8768 stores with INTEGER_CST stores.
8769 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
8770 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
8771 and use it throughout. Generate bit insertion sequences if need be.
8772 (pass_store_merging::process_store): Remove redundant condition.
8773 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
8775 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
8777 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
8778 the 128-bit floating point types. Fix function comment.
8780 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8782 * config/aarch64/aarch64-simd.md
8783 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
8785 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
8788 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
8790 PR tree-optimization/85989
8791 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
8793 (backprop::intersect_uses): Check it when deciding whether this
8794 is a backedge reference.
8795 (backprop::process_block): Add each phi to m_visited_phis
8796 after visiting it, then clear it at the end.
8798 2018-06-01 Richard Biener <rguenther@suse.de>
8800 * tree-vectorizer.h (vect_dr_stmt): New function.
8801 (vect_get_load_cost): Adjust.
8802 (vect_get_store_cost): Likewise.
8803 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
8804 Use vect_dr_stmt instead of DR_SMTT.
8805 (vect_record_base_alignments): Likewise.
8806 (vect_calculate_target_alignment): Likewise.
8807 (vect_compute_data_ref_alignment): Likewise and make static.
8808 (vect_update_misalignment_for_peel): Likewise.
8809 (vect_verify_datarefs_alignment): Likewise.
8810 (vector_alignment_reachable_p): Likewise.
8811 (vect_get_data_access_cost): Likewise. Pass down
8812 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
8813 (vect_get_peeling_costs_all_drs): Likewise.
8814 (vect_peeling_hash_get_lowest_cost): Likewise.
8815 (vect_enhance_data_refs_alignment): Likewise.
8816 (vect_find_same_alignment_drs): Likewise.
8817 (vect_analyze_data_refs_alignment): Likewise.
8818 (vect_analyze_group_access_1): Likewise.
8819 (vect_analyze_group_access): Likewise.
8820 (vect_analyze_data_ref_access): Likewise.
8821 (vect_analyze_data_ref_accesses): Likewise.
8822 (vect_vfa_segment_size): Likewise.
8823 (vect_small_gap_p): Likewise.
8824 (vectorizable_with_step_bound_p): Likewise.
8825 (vect_prune_runtime_alias_test_list): Likewise.
8826 (vect_analyze_data_refs): Likewise.
8827 (vect_supportable_dr_alignment): Likewise.
8828 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8829 (vect_gen_prolog_loop_niters): Likewise.
8830 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8831 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
8833 (vect_recog_mask_conversion_pattern): Likewise.
8834 (vect_try_gather_scatter_pattern): Likewise.
8835 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
8836 to vect_get_store_cost.
8837 (vect_get_store_cost): Get stmt_info instead of DR.
8838 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
8839 (vect_get_load_cost): Get stmt_info instead of DR.
8841 2018-06-01 Richard Biener <rguenther@suse.de>
8844 * gimple-fold.c (var_decl_component_p): Also allow offsetted
8845 vars wrapped in MEM_REFs.
8847 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
8849 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8850 Fix subreg tests so that we only return a choice between
8851 GENERAL_REGS and FP_REGS if the original classes included both.
8853 2018-06-01 Richard Biener <rguenther@suse.de>
8856 * tree-ssa-structalias.c (get_function_part_constraint):
8857 Handle NULL fi->decl.
8858 (find_func_aliases_for_call): Properly handle indirect
8859 fi from direct call.
8860 (find_func_clobbers): Likewise.
8861 (ipa_pta_execute): Likewise.
8862 (create_variable_info_for): For functions that are ifunc_resolver
8863 resolve to a varinfo that contains the result of the resolver call.
8864 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
8867 2018-05-31 Michael Collison <michael.collison@arm.com>
8869 * config/aarch64/aarch64.md:
8870 (*fix_to_zero_extenddfdi2): New pattern.
8871 * gcc.target/aarch64/fix_extend1.c: New testcase.
8873 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
8877 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
8878 and BUILT_IN_STRNCMP_EQ.
8879 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
8880 BUILT_IN_STRNCMP_EQ.
8881 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
8882 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
8883 (gimple_fold_builtin): Likewise.
8884 * tree-ssa-strlen.c (compute_string_length): New function.
8885 (determine_min_obsize): New function.
8886 (handle_builtin_string_cmp): New function to handle calls to
8887 string compare functions.
8888 (strlen_optimize_stmt): Add handling to builtin string compare
8890 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
8891 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
8892 * tree.c (build_common_builtin_nodes): Add new defines of
8893 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
8895 2018-05-31 Jakub Jelinek <jakub@redhat.com>
8898 * bb-reorder.c (pass_partition_blocks::gate): Return false for
8899 functions with naked attribute.
8901 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
8903 * config/i386/sse.md (avx_vec_concat<mode>):
8904 Substitute concat_tg_mode mode attribute with xtg_mode.
8905 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
8906 (concat_tg_mode): Remove mode attribute.
8908 2018-05-31 Martin Sebor <msebor@redhat.com>
8911 * calls.c (alloc_max_size): Correct a logic error/typo.
8912 Treat excessive arguments as infinite. Warn for invalid arguments.
8913 * doc/invoke.texi (-Walloc-size-larger-than): Update.
8915 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
8918 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
8919 and movx for Haswell.
8921 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
8922 Cesar Philippidis <cesar@codesourcery.com>
8925 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
8926 when emitting error on private/firstprivate reductions.
8927 * omp-low.c (lower_omp_target): Avoid reference-type processing
8928 on pointers for firstprivate clause.
8930 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
8932 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
8935 * config/aarch64/aarch64-simd.md
8936 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
8937 (aarch64_ld1_x3_<mode>): Likewise
8938 (aarch64_st1x2<VALLDIF:mode>): Likewise
8939 (aarch64_st1_x2_<mode>): Likewise
8940 (aarch64_st1x3<VALLDIF:mode>): Likewise
8941 (aarch64_st1_x3_<mode>): Likewise
8942 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
8943 (vld1_s8_x3): Likewise.
8944 (vld1_u16_x3): Likewise.
8945 (vld1_s16_x3): Likewise.
8946 (vld1_u32_x3): Likewise.
8947 (vld1_s32_x3): Likewise.
8948 (vld1_u64_x3): Likewise.
8949 (vld1_s64_x3): Likewise.
8950 (vld1_f16_x3): Likewise.
8951 (vld1_f32_x3): Likewise.
8952 (vld1_f64_x3): Likewise.
8953 (vld1_p8_x3): Likewise.
8954 (vld1_p16_x3): Likewise.
8955 (vld1_p64_x3): Likewise.
8956 (vld1q_u8_x3): Likewise.
8957 (vld1q_s8_x3): Likewise.
8958 (vld1q_u16_x3): Likewise.
8959 (vld1q_s16_x3): Likewise.
8960 (vld1q_u32_x3): Likewise.
8961 (vld1q_s32_x3): Likewise.
8962 (vld1q_u64_x3): Likewise.
8963 (vld1q_s64_x3): Likewise.
8964 (vld1q_f16_x3): Likewise.
8965 (vld1q_f32_x3): Likewise.
8966 (vld1q_f64_x3): Likewise.
8967 (vld1q_p8_x3): Likewise.
8968 (vld1q_p16_x3): Likewise.
8969 (vld1q_p64_x3): Likewise.
8970 (vst1_s64_x2): Likewise.
8971 (vst1_u64_x2): Likewise.
8972 (vst1_f64_x2): Likewise.
8973 (vst1_s8_x2): Likewise.
8974 (vst1_p8_x2): Likewise.
8975 (vst1_s16_x2): Likewise.
8976 (vst1_p16_x2): Likewise.
8977 (vst1_s32_x2): Likewise.
8978 (vst1_u8_x2): Likewise.
8979 (vst1_u16_x2): Likewise.
8980 (vst1_u32_x2): Likewise.
8981 (vst1_f16_x2): Likewise.
8982 (vst1_f32_x2): Likewise.
8983 (vst1_p64_x2): Likewise.
8984 (vst1q_s8_x2): Likewise.
8985 (vst1q_p8_x2): Likewise.
8986 (vst1q_s16_x2): Likewise.
8987 (vst1q_p16_x2): Likewise.
8988 (vst1q_s32_x2): Likewise.
8989 (vst1q_s64_x2): Likewise.
8990 (vst1q_u8_x2): Likewise.
8991 (vst1q_u16_x2): Likewise.
8992 (vst1q_u32_x2): Likewise.
8993 (vst1q_u64_x2): Likewise.
8994 (vst1q_f16_x2): Likewise.
8995 (vst1q_f32_x2): Likewise.
8996 (vst1q_f64_x2): Likewise.
8997 (vst1q_p64_x2): Likewise.
8998 (vst1_s64_x3): Likewise.
8999 (vst1_u64_x3): Likewise.
9000 (vst1_f64_x3): Likewise.
9001 (vst1_s8_x3): Likewise.
9002 (vst1_p8_x3): Likewise.
9003 (vst1_s16_x3): Likewise.
9004 (vst1_p16_x3): Likewise.
9005 (vst1_s32_x3): Likewise.
9006 (vst1_u8_x3): Likewise.
9007 (vst1_u16_x3): Likewise.
9008 (vst1_u32_x3): Likewise.
9009 (vst1_f16_x3): Likewise.
9010 (vst1_f32_x3): Likewise.
9011 (vst1_p64_x3): Likewise.
9012 (vst1q_s8_x3): Likewise.
9013 (vst1q_p8_x3): Likewise.
9014 (vst1q_s16_x3): Likewise.
9015 (vst1q_p16_x3): Likewise.
9016 (vst1q_s32_x3): Likewise.
9017 (vst1q_s64_x3): Likewise.
9018 (vst1q_u8_x3): Likewise.
9019 (vst1q_u16_x3): Likewise.
9020 (vst1q_u32_x3): Likewise.
9021 (vst1q_u64_x3): Likewise.
9022 (vst1q_f16_x3): Likewise.
9023 (vst1q_f32_x3): Likewise.
9024 (vst1q_f64_x3): Likewise.
9025 (vst1q_p64_x3): Likewise.
9027 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9029 * config/msp430/msp430.c (msp430_output_labelref): Prepend
9030 user_label_prefix to name.
9032 * tree-core.h: Update comment about the format of NAME string
9033 passed to handler in attribute_spec.
9035 * config/msp430/msp430.md: Remove erroneous subreg expression from
9036 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
9037 zero_extend{q,h}isi2.
9039 2018-05-30 Borislav Petkov <bp@suse.de>
9041 * doc/extend.texi: Document some architecture specific
9042 constraints and sort entries.
9044 2018-05-30 Martin Sebor <msebor@redhat.com>
9047 * builtins.c (expand_builtin_stpcpy_1): New function.
9048 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
9049 only if the former succeeds.
9051 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
9053 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
9056 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9058 * doc/invoke.texi (-flinker-output): Document
9060 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9062 * passes.c (ipa_write_summaries): Only modify statements if body
9064 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
9065 incrementally linking.
9066 (ipa_passes): Likewise.
9067 * lto-cgraph.c (lto_output_node): When incrementally linking do not
9068 pass down resolution info.
9069 * common.opt (flag_incremental_link): Update info.
9070 * gcc.c (plugin specs): Turn flinker-output=* to
9071 -plugin-opt=-linker-output-known
9072 * toplev.c (compile_file): Also cut compilation when doing incremental
9074 * flag-types. (enum lto_partition_model): Add
9075 LTO_LINKER_OUTPUT_NOLTOREL.
9076 (invoke.texi): Add -flinker-output docs.
9077 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
9078 link same way as WPA; do not stream in dead initializers.
9080 * dwarf2out.c (dwarf2out_die_ref_for_decl,
9081 darf2out_register_external_decl): Support incremental link.
9083 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9085 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
9087 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9089 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
9090 it down to simple_object_copy_lto_debug_sections.
9091 (run_gcc): Determine incremental LTO link time and configure
9092 lto1 into non-wpa mode, disable renaming of debug sections.
9094 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
9096 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
9097 descriptions of various incorrectly documented functions.
9099 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
9102 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9103 address check not strict.
9105 2018-05-30 Richard Biener <rguenther@suse.de>
9107 PR tree-optimization/85964
9108 * tracer.c (better_p): Drop initialized count check, we only
9109 call the function with initialized counts now.
9110 (find_best_successor): Do find a best edge if one
9111 has uninitialized count.
9112 (find_best_predecessor): Likewise. Do BB frequency check only
9113 if count is initialized.
9115 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
9117 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
9118 (aarch64_ldrstr_offset_compare): New.
9119 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
9120 load/store orderings.
9121 (aarch64_gen_adjusted_ldpstp): Likewise.
9123 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
9125 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9126 Check for subset of GENERAL_REGS and FP_REGS.
9127 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
9130 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
9132 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
9133 and wi::to_poly_offset. Add the current offset and then check
9134 whether the sum fits, rather than using an unchecked addition of
9135 a checked term. Check for a shwi rather than a uhwi.
9136 * expr.c (get_bit_range): Use tree_to_poly_uint64.
9137 (store_constructor): Use poly_int_tree_p.
9138 (expand_expr_real_1): Likewise.
9139 * function.c (assign_temp): Likewise.
9140 * fold-const.c (const_binop): Use poly_int_tree_p and
9142 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
9144 * ipa-icf-gimple.c (func_checker::compare_operand): Use
9145 to_poly_offset for MEM offsets.
9146 * ipa-icf.c (sem_variable::equals): Likewise.
9147 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
9148 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
9149 wi::to_poly_offset for BIT_FIELD_REF offsets.
9150 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
9152 * var-tracking.c (emit_note_insn_var_location): Use
9153 tree_to_poly_uint64.
9155 2018-05-29 Jim Wilson <jimw@sifive.com>
9157 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
9159 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
9162 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
9163 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
9165 (sse4_1_round<mode>2): Use nonimmediate_operand
9166 for operand 1 predicate.
9168 2018-05-29 Martin Sebor <msebor@redhat.com>
9169 Richard Biener <rguenther@suse.de>
9172 * calls.c (get_size_range): Call determine_value_range instead
9173 of get_value_range..
9174 * tree-vrp.h (determine_value_range): Declared new function.
9175 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
9177 2018-05-29 Richard Biener <rguenther@suse.de>
9179 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
9180 sure to use non-pattern stmts for get_earlier_stmt arguments.
9181 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
9182 called on pattern stmts.
9183 (get_later_stmt): Likewise.
9185 2018-05-29 Martin Liska <mliska@suse.cz>
9187 PR gcov-profile/85759
9188 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
9191 2018-05-29 Jakub Jelinek <jakub@redhat.com>
9193 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
9195 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
9199 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
9200 VEC_PACK_FLOAT_EXPR): New tree codes.
9201 * tree-pretty-print.c (op_code_prio): Handle
9202 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
9203 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
9204 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
9205 * tree-inline.c (estimate_operator_cost): Likewise.
9206 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
9207 * fold-const.c (const_binop): Likewise.
9208 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
9209 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
9210 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9211 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
9212 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
9213 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
9214 * expr.c (expand_expr_real_2): Likewise.
9215 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
9216 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
9217 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
9219 * optabs.c (expand_widen_pattern_expr): For
9220 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
9221 sign from result type rather than operand's type.
9222 (expand_binop_directly): For vec_packu_float_optab and
9223 vec_packs_float_optab allow result type to be different from operand's
9225 * optabs-tree.c (optab_for_tree_code): Handle
9226 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
9227 VEC_PACK_FLOAT_EXPR. Formatting fixes.
9228 * tree-vect-generic.c (expand_vector_operations_1): Handle
9229 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
9230 VEC_PACK_FLOAT_EXPR.
9231 * tree-vect-stmts.c (supportable_widening_operation): Handle
9233 (supportable_narrowing_operation): Handle FLOAT_EXPR.
9234 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
9235 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
9236 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
9237 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
9239 (vec_pack<floatprefix>_float_<mode>): New expander.
9240 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
9242 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
9243 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
9244 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
9245 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
9246 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
9248 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
9249 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
9250 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
9251 VEC_PACK_FLOAT_EXPR): Document.
9253 2018-05-29 Richard Biener <rguenther@suse.de>
9255 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
9257 (stmt_vec_info_vec): Make pointer.
9258 (init_stmt_vec_info_vec): Remove.
9259 (free_stmt_vec_info_vec): Likewise.
9260 (set_stmt_vec_info_vec): New function.
9261 (free_stmt_vec_infos): Likewise.
9262 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
9263 (set_vinfo_for_stmt): Likewise.
9264 (get_earlier_stmt): Likewise.
9265 (get_later_stmt): Likewise.
9266 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
9267 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
9268 (vec_info::~vec_info): Free stmt_vec_infos.
9269 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
9270 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
9271 (pass_slp_vectorize::execute): Likewise.
9272 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
9273 (free_stmt_vec_info_vec): Likewise.
9274 (set_stmt_vec_info_vec): New function.
9275 (free_stmt_vec_infos): Likewise.
9276 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
9277 the global stmt_vec_info_vec.
9278 * tree-parloops.c (gather_scalar_reductions): Use
9279 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
9282 2018-05-29 Richard Biener <rguenther@suse.de>
9284 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
9286 2018-05-29 Martin Liska <mliska@suse.cz>
9287 David Malcolm <dmalcolm@redhat.com>
9289 * vec.c (test_reverse): New.
9290 (vec_c_tests): Add new test.
9291 * vec.h (vl_ptr>::reverse): New function.
9293 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
9295 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
9297 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
9300 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
9302 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
9304 2018-05-28 Richard Biener <rguenther@suse.de>
9306 PR tree-optimization/85933
9307 * tree-vect-data-refs.c (vect_record_base_alignments): Only
9308 look at stmts marked as vectorizable.
9310 2018-05-28 Richard Biener <rguenther@suse.de>
9312 PR tree-optimization/85934
9313 * tree-vect-generic.c (expand_vector_operations_1): Hoist
9314 vector boolean check before scalar optimization.
9316 2018-05-28 Jakub Jelinek <jakub@redhat.com>
9318 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
9321 2018-05-28 Mark Wielaard <mark@klomp.org>
9323 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
9324 if it is an expression containing a minus sign.
9326 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
9328 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
9330 2018-05-27 Paul Koning <ni1d@arrl.net>
9332 * config/pdp11/pdp11.md (truncsihi2): Remove.
9334 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
9335 Chung-Ju Wu <jasonwucj@gmail.com>
9337 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
9339 (unaligned_store_dw): Ditto.
9340 * config/nds32/nds32-memory-manipulation.c
9341 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
9342 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
9343 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
9344 (emit_setmem_word_loop): Rename to ...
9345 (emit_setmem_doubleword_loop): ... this.
9346 (nds32_gen_dup_4_byte_to_word_value): New function.
9347 (nds32_gen_dup_8_byte_to_double_word_value): New function.
9348 (nds32_expand_setmem_loop): Refine implementation.
9349 (nds32_expand_setmem_loop_v3m): Ditto.
9350 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
9353 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
9355 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
9357 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
9359 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
9360 (nds32_init_machine_status): Initialize machine->attr_naked_p and
9361 machine->attr_no_prologue_p.
9362 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
9363 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
9364 (nds32_expand_epilogue): Consider attr_naked_p.
9365 (nds32_expand_epilogue_v3pop): Likewise.
9366 (nds32_can_use_return_insn): Likewise.
9367 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
9368 attr_no_prologue_p fields.
9369 * config/nds32/nds32.opt (mret-in-naked-func): New option.
9371 2018-05-27 Jakub Jelinek <jakub@redhat.com>
9374 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
9376 * config/i386/sse.md
9377 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
9379 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
9381 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
9383 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
9385 (*<floatsuffix>floatv2div2sf2): Rename to ...
9386 (*float<floatunssuffix>v2div2sf2): ... this.
9387 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
9388 (float<floatunssuffix>v2div2sf2_mask): ... this.
9389 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
9390 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
9391 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
9393 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
9395 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
9397 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
9399 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
9401 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
9403 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
9404 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
9405 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
9406 gen_ufix_truncv8dfv8si2.
9407 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
9408 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
9409 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
9410 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
9411 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
9412 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
9413 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
9414 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
9416 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
9420 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
9422 2018-05-25 Jim Wilson <jimw@sifive.com>
9424 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
9425 * config/riscv/riscv.c (struct machine_function): Add
9426 interrupt_handler_p and attribute_checked_p fields.
9427 (riscv_attribute_table): Add interrupt.
9428 (riscv_interrupt_type_p): New.
9429 (riscv_save_reg_p): Save extra regs for interrupt handler.
9430 (riscv_use_save_libcall): Return false for interrupt handler.
9431 (riscv_first_stack_step): Add forward declaration.
9432 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
9433 for interrupt handler with large frame. Use it for saved reg list.
9434 (riscv_expand_prologue): Move flag_stack_usage_info support to
9435 eliminate duplication.
9436 (riscv_expand_epilogue): Generate mret for interrupt handler.
9437 (riscv_epilogue_uses): New.
9438 (riscv_can_use_return_insn): Return false for interrupt handler.
9439 (riscv_function_ok_for_sibcall): Likewise.
9440 (riscv_set_current_function): Add interrupt handler support.
9441 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
9442 * config/riscv/riscv.md (UNSPECV_MRET): New.
9444 (riscv_frflags, riscv_fsflags): Use tab after opcode.
9446 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
9448 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
9450 PR tree-optimization/85712
9451 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
9452 this candidate has already been replaced in-situ by a copy.
9454 2018-05-25 Jason Merrill <jason@redhat.com>
9456 PR c++/80485 - inline function non-zero address.
9457 * symtab.c (nonzero_address): Check DECL_COMDAT.
9459 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
9462 * config/alpha/alpha.md (ashlsi3): New insn pattern.
9463 (*ashlsi_se): Rename from *ashldi_se. Define as sign
9464 extension of SImode operation. Use const123_operand predicate.
9465 (*saddsi_1): Remove.
9466 (*saddl_se_1): Ditto.
9468 (*ssubl_se_1): Ditto.
9469 * config/alpha/predicates.md (const123_operand): New predicate.
9470 * config/alpha/constraints.md (P): Use IN_RANGE.
9472 2018-05-25 Richard Biener <rguenther@suse.de>
9474 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
9476 (ref_maybe_used_by_stmt_p): Likewise.
9477 (stmt_may_clobber_ref_p): Likewise.
9478 (stmt_may_clobber_ref_p_1): Likewise.
9479 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
9481 (ref_maybe_used_by_stmt_p): Likewise.
9482 (stmt_may_clobber_ref_p): Likewise.
9483 (stmt_may_clobber_ref_p_1): Likewise.
9484 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
9485 the alias oracle to disambiguate DRs with stmts DR analysis
9487 (vect_analyze_data_refs): Do not give up on not analyzable
9488 DRs for BB vectorization. Remove code truncating the dataref
9491 2018-05-25 Jakub Jelinek <jakub@redhat.com>
9494 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
9495 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
9496 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
9498 2018-05-25 Richard Biener <rguenther@suse.de>
9500 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
9501 function, combining stmt data ref gathering and fatal analysis
9503 (vect_analyze_data_refs): Remove now redudnant code and simplify.
9504 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
9505 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
9506 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
9507 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
9509 2018-05-25 Bin Cheng <bin.cheng@arm.com>
9511 PR tree-optimization/85720
9512 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
9513 SCC if all partitions are builtins.
9514 (version_loop_by_alias_check): New parameter. Generate cancelable
9515 runtime alias check if all partitions are builtins.
9516 (distribute_loop): Update call to above function.
9518 2018-05-25 Bin Cheng <bin.cheng@arm.com>
9520 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
9521 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
9522 (parm_default_def_partition_arg): Ditto.
9523 (set_parm_default_def_partition): Ditto.
9524 (get_parm_default_def_partitions): Ditto and make it static.
9525 (get_undefined_value_partitions): Ditto and make it static.
9526 (remove_ssa_form): Refactor call to init_var_map here.
9527 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
9528 computation for loop region.
9529 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
9530 (register_default_def): Delete.
9531 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
9532 (parm_default_def_partition_arg): Ditto.
9533 (set_parm_default_def_partition): Ditto.
9534 (get_parm_default_def_partitions): Ditto and make it static.
9535 (get_undefined_value_partitions): Ditto and make it static.
9536 (coalesce_with_default, coalesce_with_default): Update comment.
9537 (create_coalesce_list_for_region): New func factored out from
9538 create_outofssa_var_map.
9539 (populate_coalesce_list_for_outofssa): New func factored out from
9540 create_outofssa_var_map and coalesce_ssa_name.
9541 (create_outofssa_var_map): Delete.
9542 (coalesce_ssa_name): Refactor to support live range computation.
9543 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
9544 (get_parm_default_def_partitions): Delete.
9545 (get_undefined_value_partitions): Ditto.
9546 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
9547 computation for loop region.
9548 (new_tree_live_info, loe_visit_block): Ditto.
9549 (live_worklist, set_var_live_on_entry): Ditto.
9550 (calculate_live_on_exit, verify_live_on_entry): Ditto.
9551 * tree-ssa-live.h (struct _var_map): New fields.
9552 (init_var_map): Change decl.
9553 (region_contains_p): New.
9555 2018-05-25 Bin Cheng <bin.cheng@arm.com>
9557 * tree-ssa-live.h (live_merge_and_clear): Delete.
9559 2018-05-25 Richard Biener <rguenther@suse.de>
9562 * tree-dump.c (dequeue_and_dump): Remove access to removed
9563 operand 2 of a SWITCH_EXPR.
9565 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9567 * doc/sourcebuild.texi (vect_double_cond_arith): Include
9568 multiplication and division.
9569 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
9570 (cond_udiv@var{m}, cond_umod@var{m}): Document.
9571 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
9572 (cond_udiv_optab, cond_umod_optab): New optabs.
9573 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
9574 (IFN_COND_RDIV): New internal functions.
9575 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
9576 TRUNC_MOD_EXPR and RDIV_EXPR.
9577 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
9578 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
9580 (SVE_INT_BINARY): Include mult.
9581 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
9582 (optab, sve_int_op): Handle mult.
9583 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
9585 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
9586 for SVE_INT_BINARY_SD.
9588 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9590 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
9591 (optab, sve_int_op): Handle div and udiv.
9592 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
9593 for SVE_INT_BINARY_SD.
9594 (*<optab><mode>3): New insn for the same.
9596 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9598 * tree-vect-patterns.c: Include predict.h.
9599 (vect_recog_divmod_pattern): Restrict check for division support
9600 to when optimizing for size.
9602 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9604 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
9605 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
9606 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
9607 (gimple_match_op::set_op): Likewise.
9608 (gimple_resimplify4): Declare.
9609 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
9610 (expr::gen_transform): Likewise.
9611 (decision_tree::gen): Generate a simplification routine for 4 operands.
9612 * gimple-match-head.c (gimple_simplify): Add an overload for
9613 4 operands. In the top-level function, handle up to 4 call
9614 arguments and call gimple_resimplify4.
9615 (gimple_resimplify4): New function.
9616 (build_call_internal): Pass a fourth operand.
9617 (maybe_push_to_seq): Likewise.
9618 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
9619 Fold VEC_COND_EXPRs of an operation and a default value into
9620 an IFN_COND_* function if possible.
9621 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
9623 (SVE_COND_FP_BINARY): Include them.
9624 (optab, sve_fp_op): Handle them.
9625 (SVE_INT_BINARY_REV): New code iterator.
9626 (SVE_COND_FP_BINARY_REV): New int iterator.
9627 (commutative): New int attribute.
9628 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
9630 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
9632 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
9633 (*cond_<optab><mode>): New patterns for reversed operands.
9635 2018-05-25 Richard Biener <rguenther@suse.de>
9637 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
9638 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
9639 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
9640 (STMT_VINFO_GROUPED_ACCESS): Adjust.
9641 * tree-vect-data-refs.c (everywhere): Adjust users.
9642 * tree-vect-loop.c (everywhere): Likewise.
9643 * tree-vect-slp.c (everywhere): Likewise.
9644 * tree-vect-stmts.c (everywhere): Likewise.
9645 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
9647 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9649 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
9651 (gcc_cv_as_section_exclude): ... this.
9652 Try Solaris as #exclude syntax.
9653 * configure: Regenerate.
9654 * config.in: Regenerate.
9655 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
9657 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
9658 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
9660 * varasm.c (default_elf_asm_named_section): Don't check if
9661 HAVE_GAS_SECTION_EXCLUDE is defined.
9663 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9665 * doc/md.texi: Update the documentation of the cond_* optabs
9666 to mention the new final operand. Fix GET_MODE_NUNITS call.
9667 Describe the scalar case too.
9668 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
9669 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
9671 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
9672 (get_conditional_internal_fn): Update comment.
9673 * tree-vect-loop.c (vectorizable_reduction): Pass the original
9674 accumulator value as a final argument to conditional functions.
9675 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
9676 a define_expand and add an "else" operand. Assert for now that
9677 the else operand is equal to operand 2. Use SVE_INT_BINARY and
9678 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
9679 (*cond_<optab><mode>): New patterns.
9680 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
9681 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
9682 (UNSPEC_COND_EOR): Delete.
9683 (optab): Remove associated mappings.
9684 (SVE_INT_BINARY): New code iterator.
9685 (sve_int_op): Remove int attribute and add "minus" to the code
9687 (SVE_COND_INT_OP): Delete.
9688 (SVE_COND_FP_OP): Rename to...
9689 (SVE_COND_FP_BINARY): ...this.
9691 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
9693 * optabs.c (can_reuse_operands_p): New function.
9694 (maybe_legitimize_operands): Try to reuse the results for
9697 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
9699 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
9700 Add {q} suffix to insn mnemonic.
9702 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9704 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
9705 (msp430_warn_func_return): New.
9707 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
9709 * fold-const.c (tree_nonzero_bits): New function.
9710 * fold-const.h (tree_nonzero_bits): Likewise.
9711 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
9712 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
9714 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
9718 * varasm.c (assemble_alias): Check ifunc_resolver only on
9721 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
9724 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
9725 when memory input operand is handled.
9727 2018-05-24 Luis Machado <luis.machado@linaro.org>
9729 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
9731 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
9733 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
9735 * match.pd: Delay FMA folds until after vectorization.
9737 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
9740 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9741 address check not strict.
9743 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
9745 * gimple-match.h (gimple_match_op): New class.
9746 (mprts_hook): Replace parameters with a gimple_match_op *.
9747 (maybe_build_generic_op): Likewise.
9748 (gimple_simplified_result_is_gimple_val): Replace parameters with
9749 a const gimple_match_op *.
9750 (gimple_simplify): Replace code_helper * and tree * parameters with
9751 a gimple_match_op * parameter.
9752 (gimple_resimplify1): Replace code_helper *, tree and tree *
9753 parameters with a gimple_match_op * parameter.
9754 (gimple_resimplify2): Likewise.
9755 (gimple_resimplify3): Likewise.
9756 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
9757 parameters with a gimple_match_op * parameter.
9758 * gimple-match-head.c (gimple_simplify): Change prototypes of
9759 auto-generated functions to take a gimple_match_op * instead of
9760 separate code_helper * and tree * parameters. Make the same
9761 change in the top-level overload and update calls to the
9762 gimple_resimplify routines. Update calls to the auto-generated
9763 functions and to maybe_push_res_to_seq in the publicly-facing
9764 operation-specific gimple_simplify overloads.
9765 (gimple_match_op::MAX_NUM_OPS): Define.
9766 (gimple_resimplify1): Replace rcode and ops with a single res_op
9767 parameter. Update call to gimple_simplify.
9768 (gimple_resimplify2): Likewise.
9769 (gimple_resimplify3): Likewise.
9770 (mprts_hook): Replace parameters with a gimple_match_op *.
9771 (maybe_build_generic_op): Likewise.
9772 (build_call_internal): Replace type, nargs and ops with
9773 a gimple_match_op *.
9774 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
9775 with a single gimple_match_op *. Update calls to mprts_hook,
9776 build_call_internal and gimple_simplified_result_is_gimple_val.
9777 Factor out code that is common to the tree_code and combined_fn cases.
9778 * genmatch.c (expr::gen_transform): Replace tem_code and
9779 tem_ops with a gimple_match_op called tem_op. Update calls
9780 to the gimple_resimplify functions and maybe_push_res_to_seq.
9781 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
9782 res_ops. Update call to the gimple_resimplify functions.
9783 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
9784 (decision_tree::gen): Make the functions take a gimple_match_op *
9785 called res_op instead of separate res_code and res_ops parameters.
9786 Update call accordingly.
9787 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
9788 and ops with a single res_op parameter. Update calls to
9789 maybe_build_generic_op and maybe_push_res_to_seq.
9790 (fold_stmt_1): Update calls to gimple_simplify and
9791 replace_stmt_with_simplification.
9792 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
9793 and gimple_simplified_result_is_gimple_val.
9794 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
9796 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
9797 with a gimple_match_op *.
9798 (vn_nary_build_or_lookup): Likewise. Update call to
9799 vn_nary_build_or_lookup_1.
9800 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
9801 gimple_match_op *. Update calls to the gimple_resimplify routines
9802 and to gimple_simplified_result_is_gimple_val.
9803 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
9804 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
9805 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
9806 (visit_nary_op): Likewise.
9807 (visit_reference_op_load): Likewise.
9809 2018-05-23 Luis Machado <luis.machado@linaro.org>
9811 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
9812 modifier for printing the step amount.
9814 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9817 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
9820 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
9822 * doc/sourcebuild.texi (Endianness): New subsubsection.
9824 2018-05-23 Luis Machado <luis.machado@linaro.org>
9826 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
9827 <prefetch_dynamic_strides>: New const bool field.
9828 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
9829 prefetch_dynamic_strides.
9830 (exynosm1_prefetch_tune): Likewise.
9831 (thunderxt88_prefetch_tune): Likewise.
9832 (thunderx_prefetch_tune): Likewise.
9833 (thunderx2t99_prefetch_tune): Likewise.
9834 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
9836 (aarch64_override_options_internal): Update to set
9837 PARAM_PREFETCH_DYNAMIC_STRIDES.
9838 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
9839 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
9840 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
9841 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
9842 prefetch-dynamic-strides setting.
9844 2018-05-23 Luis Machado <luis.machado@linaro.org>
9846 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
9847 <minimum_stride>: New const int field.
9848 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
9849 minimum_stride field defaulting to -1.
9850 (exynosm1_prefetch_tune): Likewise.
9851 (thunderxt88_prefetch_tune): Likewise.
9852 (thunderx_prefetch_tune): Likewise.
9853 (thunderx2t99_prefetch_tune): Likewise.
9854 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
9855 <default_opt_level>: Set to 3.
9856 (aarch64_override_options_internal): Update to set
9857 PARAM_PREFETCH_MINIMUM_STRIDE.
9858 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
9859 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
9860 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
9861 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
9862 stride is constant and is below the minimum stride threshold.
9864 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9866 * config/arm/arm-cpus.in (mode26): Delete.
9867 (armv4): Delete mode26 reference.
9868 * config/arm/arm.c (arm_configure_build_target): Delete use of
9871 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
9873 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
9875 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
9876 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
9878 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
9879 (floatunsdidf2): Ditto.
9881 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
9883 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
9884 (fixuns_trunc<mode>si2_avx512f): Ditto.
9885 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
9886 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
9887 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
9889 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
9891 PR rtl-optimization/79985
9892 * df-scan.c (df_insn_refs_collect): Remove special case for
9893 global registers and asm statements.
9895 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
9897 * extend.texi (Global Register Variables): Rewrite the bullet list.
9898 Note that the register is available for allocation. Note that access
9899 via inline asm must use constraints. Add note about async-signal
9900 handlers. Remove paragraph about automagic register selection.
9902 2018-05-23 Richard Biener <rguenther@suse.de>
9904 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
9905 of fixed offset from memset VN.
9907 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
9909 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
9911 (alloc_cand_and_find_basis): Initialize first_interp field.
9912 (slsr_process_mul): Modify first_interp field.
9913 (slsr_process_add): Likewise.
9914 (slsr_process_cast): Modify first_interp field for each new
9916 (slsr_process_copy): Likewise.
9917 (dump_candidate): Dump first_interp field.
9918 (replace_mult_candidate): Process all interpretations, not just
9920 (replace_rhs_if_not_dup): Likewise.
9921 (replace_one_candidate): Likewise.
9923 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
9925 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
9927 (aarch64_needs_frame_chain): New function.
9928 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
9930 2018-05-23 Sudakshina Das <sudi.das@arm.com>
9933 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
9934 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
9935 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
9936 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
9937 as true for strict-align.
9938 (aarch64_can_inline_p): Perform checks even when callee has no
9939 attributes to check for strict alignment.
9940 * doc/extend.texi (AArch64 Function Attributes): Document
9942 * doc/invoke.texi: (AArch64 Options): Likewise.
9944 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
9946 PR tree-optimization/85853
9947 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
9948 the handling of the root of the node to...
9949 (vect_slp_analyze_node_operations_1): ...this new function,
9950 and run the whole thing with the child nodes' def types
9951 set according to their SLP node's def type.
9953 2018-05-23 Richard Biener <rguenther@suse.de>
9956 * tree-data-ref.c (create_runtime_alias_checks): Defer
9957 and ignore overflow warnings.
9959 2018-05-23 Yury Gribov <tetra2005@gmail.com>
9961 PR tree-optimization/85822
9962 * tree-vrp.c (is_masked_range_test): Fix handling of negative
9965 2018-05-23 Richard Biener <rguenther@suse.de>
9967 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
9968 memset constants via native_interpret_expr.
9970 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
9973 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
9975 (cgraph_node::create_alias): Likewise.
9976 (cgraph_node::get_availability): Check ifunc_resolver instead
9977 of looking up ifunc attribute.
9978 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
9979 * varasm.c (do_assemble_alias): Likewise.
9980 (assemble_alias): Likewise.
9981 (default_binds_local_p_3): Likewise.
9982 * cgraph.h (cgraph_node): Add ifunc_resolver.
9983 (cgraph_node::only_called_directly_or_aliased_p): Return false
9985 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
9987 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
9988 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
9989 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
9990 instead of looking up ifunc attribute.
9992 2018-05-22 Luis Machado <luis.machado@linaro.org>
9994 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
9996 2018-05-22 Martin Sebor <msebor@redhat.com>
9999 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
10000 only when expasion succeeds.
10001 (expand_builtin_strcmp): Same.
10002 (expand_builtin_strncmp): Same.
10004 2018-05-22 Martin Sebor <msebor@redhat.com>
10006 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
10008 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
10009 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10011 * config/aarch64/aarch64-ldpstp.md: Replace uses of
10012 aarch64_mem_pair_operand with memory_operand and delete operand swapping
10014 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
10015 Add check for legitimate_address.
10016 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
10017 (aarch64_swap_ldrstr_operands): New.
10018 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
10021 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
10022 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10024 * config/aarch64/aarch64.md: New patterns to generate stp
10026 (store_pair_sw, store_pair_dw): New patterns to generate stp for
10027 single words and double words.
10028 (load_pair_sw, load_pair_dw): Likewise.
10029 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
10031 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
10033 * config/aarch64/aarch64-ldpstp.md: Modify peephole
10034 for different mode ldpstp and add peephole for merged zero stores.
10035 Likewise for loads.
10036 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
10038 (aarch64_gen_store_pair): Rename calls to match new patterns.
10039 (aarch64_gen_load_pair): Rename calls to match new patterns.
10040 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
10041 (load_pair<DREG:mode><DREG2:mode>): ... This.
10042 (store_pair<mode>): Rename to...
10043 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
10044 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
10045 New mode iterators.
10046 (V_INT_EQUIV): Handle SImode.
10047 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
10050 2018-05-22 Martin Sebor <msebor@redhat.com>
10053 * calls.c (maybe_warn_nonstring_arg): Use string length to set
10054 or ajust the presumed bound on an operation to avoid unnecessary
10057 2018-05-22 Martin Sebor <msebor@redhat.com>
10059 PR tree-optimization/85826
10060 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
10061 assuming that a DECL necesarily has a constant size.
10063 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
10065 PR middle-end/85862
10066 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
10068 2018-05-22 Richard Biener <rguenther@suse.de>
10070 PR tree-optimization/85834
10071 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
10072 non-constant and non-zero memset arguments.
10074 2018-05-22 Martin Liska <mliska@suse.cz>
10077 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
10079 2018-05-22 Richard Biener <rguenther@suse.de>
10081 PR tree-optimization/85863
10082 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
10083 comparisons when vectype is specified.
10084 (vectorizable_condition): Do not specify vectype for
10085 vect_is_simple_cond when SLP vectorizing.
10087 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
10090 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
10091 define __ibm128 as long double.
10092 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
10093 as a distinct type when IEEE 128-bit support is enabled.
10094 (init_float128_ieee): Fix up conversions between IFmode and IEEE
10095 128-bit types to use the correct functions.
10096 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
10097 convert between 128-bit floating point types that have different
10098 modes but the same representation, instead of using gen_lowpart to
10100 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
10102 (IFKF_reg): New attributes to give the register constraints for
10104 (extend<mode>tf2_internal): New insns to mark an explicit
10105 conversion between 128-bit floating point types that have a
10106 different mode but share the same representation.
10108 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
10110 PR tree-optimization/85814
10111 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
10112 a null return from get_strinfo when unsharing the next
10113 strinfo in the chain.
10115 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
10118 * varasm.c (weak_finish): Clean up weak_decls.
10120 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10122 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
10123 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
10124 UNSPEC_UADALP values.
10125 * config/aarch64/iterators.md (ABAL): New int iterator.
10128 (sur): Add mappings for the above.
10129 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
10131 (aarch64_<sur>abal<mode>_4): Likewise.
10132 (aarch64_<sur>adalp<mode>_3): Likewise.
10133 (<sur>sadv16qi): New define_expand.
10135 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
10137 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
10138 (*movdf_internal): Ditto.
10139 (*rcpsf2_sse): Ditto.
10140 (*rsqrtsf2_sse): Ditto.
10141 (*sqrt<mode>2_sse): Ditto.
10143 2018-05-21 Tamar Christina <tamar.christina@arm.com>
10145 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
10147 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
10148 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
10149 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
10150 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
10152 * config/aarch64/arm_neon.h: Likewise.
10153 * config/aarch64/iterators.md (VQ_I): New.
10155 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
10157 * config.gcc: Add arc/t-multilib-linux to tmake_file for
10159 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
10162 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
10164 * config/nds32/constraints.md (S): New constraint.
10165 * config/nds32/nds32.md (call_internal): Use constraint S.
10166 (call_value_internal): Likewise.
10167 (sibcall_internal): Likewise.
10168 (sibcall_value_internal): Likewise.
10170 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
10171 Chung-Ju Wu <jasonwucj@gmail.com>
10173 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
10174 into consideration.
10176 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
10177 Chung-Ju Wu <jasonwucj@gmail.com>
10179 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
10180 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
10181 (nds32_rtx_costs_impl): Simplify.
10182 (nds32_address_cost_impl): Simplify.
10183 (nds32_init_rtx_costs): New function.
10184 (nds32_rtx_costs_speed_prefer): Likewise.
10185 (nds32_rtx_costs_size_prefer): Likewise.
10186 (nds32_address_cost_speed_prefer): Likewise.
10187 (nds32_address_cost_speed_fwprop): Likewise.
10188 (nds32_address_cost_size_prefer): Likewise.
10189 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
10190 * config/nds32/nds32.c (nds32_option_override): Use
10191 nds32_init_rtx_costs function.
10193 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
10195 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
10196 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
10197 (TARGET_PIPELINE_N8): Likewise.
10198 (TARGET_PIPELINE_N10): Likewise.
10199 (TARGET_PIPELINE_N13): Likewise.
10200 (TARGET_PIPELINE_GRAYWOLF): Likewise.
10202 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
10204 * config/nds32/nds32-fpu.md: Update copyright year.
10206 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10208 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
10210 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10212 * config/nds32/nds32.c
10213 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
10214 * config/nds32/nds32.opt (minline-asm-r15): New option.
10216 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10218 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
10220 * config/nds32/nds32.md (abssi2): New pattern.
10222 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
10224 * config/i386/i386.md (rex64namesuffix): New mode attribute.
10225 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
10226 Merge insn pattern from sse_cvtsi2ss<round_name> and
10227 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
10228 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
10229 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
10230 using SWI48 mode iterator.
10231 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
10232 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
10233 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
10234 pattern from sse_cvttss2si<round_saeonly_name>
10235 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
10236 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
10237 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
10238 using SWI48 mode iterator.
10239 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
10240 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
10241 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
10242 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
10243 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
10244 using SWI48 mode iterator.
10245 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
10246 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
10247 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
10248 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
10249 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
10250 SWI48 mode iterator.
10251 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
10252 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
10253 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
10254 pattern from sse_cvttsd2si<round_saeonly_name>
10255 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
10257 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10259 * config/nds32/nds32-md-auxiliary.c
10260 (nds32_valid_smw_lwm_base_p): Refine.
10261 (nds32_output_smw_single_word): Refine.
10262 (nds32_output_smw_double_word): New.
10263 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
10265 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10267 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
10268 (nds32_output_stack_pop): Refine.
10269 (nds32_expand_unaligned_load): Refine.
10270 (nds32_expand_unaligned_store): Refine.
10272 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10273 Chung-Ju Wu <jasonwucj@gmail.com>
10275 * config/nds32/constants.md: Add TP_REGNUM constant.
10276 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
10277 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
10279 * config/nds32/nds32-doubleword.md: Consider flag_pic.
10280 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
10281 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
10282 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
10283 and PIC code generation.
10284 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
10286 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
10288 * config/nds32/nds32.md: Support TLS and PIC.
10289 * config/nds32/nds32.c: Support TLS and PIC.
10290 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
10291 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
10294 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10296 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
10297 mode with E_ prefix.
10299 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10300 Chung-Ju Wu <jasonwucj@gmail.com>
10302 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
10303 * config/nds32/nds32-md-auxiliary.c
10304 (symbolic_reference_mentioned_p): New.
10305 (nds32_legitimize_ict_address): New.
10306 (nds32_expand_ict_move): New.
10307 (nds32_indirect_call_referenced_p): New.
10308 (nds32_symbol_binds_local_p): Delete.
10309 (nds32_long_call_p): Modify.
10310 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
10311 * config/nds32/nds32-protos.h
10312 (symbolic_reference_mentioned_p): Declare.
10313 (nds32_legitimize_ict_address): Declare.
10314 (nds32_expand_ict_move): Declare.
10315 (nds32_indirect_call_referenced_p): Declare.
10316 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
10317 (nds32_relax_group): Use nds32_ict_const_p as condition.
10318 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
10319 (nds32_asm_file_start): Output ict_model directive in asm code.
10320 (nds32_legitimate_address_p): Consider indirect call.
10321 (nds32_print_operand): Consider indirect call.
10322 (nds32_print_operand_address): Consider indirect call.
10323 (nds32_insert_attributes): Handle "indirect_call" attribute.
10324 (TARGET_LEGITIMATE_ADDRESS_P): Define.
10325 (TARGET_LEGITIMATE_CONSTANT_P): Define.
10326 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
10327 (TARGET_DELEGITIMIZE_ADDRESS): Define.
10328 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
10329 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
10330 (TARGET_ICT_MODEL_SMALL): Define.
10331 (TARGET_ICT_MODEL_LARGE): Define.
10332 * config/nds32/nds32.md (movsi): Consider ict model.
10333 (call, call_value): Consider ict model.
10334 (sibcall, sibcall_value): Consider ict model.
10335 * config/nds32/nds32.opt (mict-model): New option.
10336 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
10339 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
10340 Monk Chiang <sh.chiang04@gmail.com>
10341 Jim Wilson <jimw@sifive.com>
10343 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
10344 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
10345 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
10346 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
10347 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
10348 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
10349 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
10350 compute save_libcall_adjustment properly.
10351 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
10352 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
10353 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
10354 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
10355 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
10356 (ABI_SPEC): Handle mabi=ilp32e.
10357 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
10358 (RVE): Add RVE mask.
10359 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
10360 <-march>: Add rv32e as an example.
10362 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
10365 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
10366 (intra_create_variable_infos): Handle C++ constructors.
10368 2018-05-18 Martin Liska <mliska@suse.cz>
10370 * passes.def: Remove a redundant pass.
10372 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
10375 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
10377 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10379 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
10381 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
10383 (armv2, armv2a, armv3, armv3m): Delete architectures.
10384 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
10385 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
10386 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
10388 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
10389 (*mulsidi3adddi): Likewise.
10390 (mulsidi3): Likewise.
10391 (*mulsidi3_nov6): Likewise.
10392 (umulsidi3): Likewise.
10393 (umulsidi3_nov6): Likewise.
10394 (umaddsidi4): Likewise.
10395 (*umulsidi3adddi): Likewise.
10396 (smulsi3_highpart): Likewise.
10397 (*smulsi3_highpart_nov6): Likewise.
10398 (umulsi3_highpart): Likewise.
10399 (*umulsi3_highpart_nov6): Likewise.
10400 * config/arm/arm.h (arm_arch3m): Delete.
10401 * config/arm/arm.c (arm_arch3m): Delete.
10402 (arm_option_override_internal): Update armv3-related comment.
10403 (arm_configure_build_target): Delete use of isa_bit_mode32.
10404 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
10405 (arm_rtx_costs_internal): Delete check of arm_arch3m.
10406 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
10407 (mulsa3): Likewise.
10408 (mulusa3): Likewise.
10409 * config/arm/arm-protos.h (arm_arch3m): Delete.
10410 * config/arm/arm-tables.opt: Regenerate.
10411 * config/arm/arm-tune.md: Likewise.
10412 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
10413 deleted architectures.
10415 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10417 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
10418 (armv5t, armv5te): New features.
10419 (ARMv5, ARMv5e): Delete fgroups.
10420 (ARMv5t, ARMv5te): Adjust for above changes.
10421 (ARMv6m): Likewise.
10422 (armv5, armv5e): Delete arches.
10423 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
10425 (*call_reg_arm): Likewise.
10426 (*call_value_reg_armv5): Likewise.
10427 (*call_value_reg_arm): Likewise.
10428 (*call_symbol): Likewise.
10429 (*call_value_symbol): Likewise.
10430 (*sibcall_insn): Likewise.
10431 (*sibcall_value_insn): Likewise.
10432 (clzsi2): Likewise.
10433 (prefetch): Likewise.
10434 (define_split and define_peephole2 dependent on arm_arch5):
10436 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
10438 (TARGET_ARM_QBIT): Likewise.
10439 (TARGET_DSP_MULTIPLY): Likewise.
10440 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
10441 (arm_arch5, arm_arch5e): Delete.
10442 (arm_arch5t, arm_arch5te): Declare.
10443 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
10444 (arm_arch5t): Declare.
10445 (arm_option_reconfigure_globals): Update for the above.
10446 (arm_options_perform_arch_sanity_checks): Update comment, replace
10447 use of arm_arch5 with arm_arch5t.
10448 (use_return_insn): Likewise.
10449 (arm_emit_call_insn): Likewise.
10450 (output_return_instruction): Likewise.
10451 (arm_final_prescan_insn): Likewise.
10452 (arm_coproc_builtin_available): Likewise.
10453 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
10454 arm_arch5e with arm_arch5t and arm_arch5te.
10455 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
10456 (arm_arch5t, arm_arch5te): Declare.
10457 * config/arm/arm-tables.opt: Regenerate.
10458 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
10459 * config/arm/t-multilib: Likewise.
10460 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
10461 instead of arm_arch5.
10462 (*call_reg_thumb1): Likewise.
10463 (*call_value_reg_thumb1_v5): Likewise.
10464 (*call_value_reg_thumb1): Likewise.
10465 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
10467 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
10469 2018-05-18 Martin Liska <mliska@suse.cz>
10471 PR gcov-profile/84846
10472 * doc/gcov.texi: Document -t option of gcov tool.
10474 2018-05-18 Martin Liska <mliska@suse.cz>
10476 PR gcov-profile/84846
10477 * gcov.c (print_usage): Add new -t option.
10478 (process_args): Handle the option.
10479 (generate_results): Use stdout as output when requested by
10482 2018-05-18 Martin Liska <mliska@suse.cz>
10484 PR gcov-profile/84846
10485 * coverage.c (coverage_init): Write PWD to .gcno file.
10486 * doc/gcov.texi: Document how working directory is printed.
10487 * gcov-dump.c (dump_gcov_file): Print PWD.
10488 * gcov.c (output_intermediate_file): Likewise.
10489 (read_graph_file): Read PWD string.
10490 (output_lines): Print PWD.
10492 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10494 PR middle-end/85817
10495 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
10496 for retval and return false if all args to phi are zero.
10498 2018-05-18 Richard Biener <rguenther@suse.de>
10500 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
10502 (evrp_dom_walker::before_dom_children): Call it.
10504 2018-05-18 Richard Biener <rguenther@suse.de>
10506 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
10507 results when processing array refs with variable index.
10509 2018-05-18 Toon Moene <toon@moene.org>
10511 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
10512 directly after that of -floop-interchange. Indicate that both
10513 options are enabled by default when specifying -O3.
10515 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10517 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
10518 iterator. Delete separate integer-mode vec_set<mode> expander.
10519 (aarch64_simd_vec_setv2di): Delete.
10520 (vec_setv2di): Delete.
10521 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
10522 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
10523 the "w, r" alternative.
10525 2018-05-18 Martin Liska <mliska@suse.cz>
10527 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
10528 * tree-pass.h (make_pass_lower_switch_O0): New function.
10529 * tree-switch-conversion.c (node_has_low_bound): Remove.
10530 (node_has_high_bound): Likewise.
10531 (node_is_bounded): Likewise.
10532 (class pass_lower_switch): Make it a template type and create
10534 (pass_lower_switch::execute): Add template argument.
10535 (make_pass_lower_switch): New function.
10536 (make_pass_lower_switch_O0): New function.
10537 (do_jump_if_equal): Remove.
10538 (emit_case_nodes): Simplify to just handle all 3 cases and leave
10539 all the hard work to tree optimization passes.
10541 2018-05-18 Martin Liska <mliska@suse.cz>
10543 * dbgcnt.c (limit_low): Renamed from limit.
10544 (limit_high): New variable.
10545 (dbg_cnt_is_enabled): Check for upper limit.
10546 (dbg_cnt): Adjust dumping.
10547 (dbg_cnt_set_limit_by_index): Add new argument for high
10549 (dbg_cnt_set_limit_by_name): Likewise.
10550 (dbg_cnt_process_single_pair): Parse new format.
10551 (dbg_cnt_process_opt): Use strtok.
10552 (dbg_cnt_list_all_counters): Remove 'value' and add
10554 * doc/invoke.texi: Document changes.
10556 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
10558 * doc/sourcebuild.texi (scalar_all_fma): Document.
10559 * tree.def (FMA_EXPR): Delete.
10560 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
10561 * internal-fn.c (ternary_direct): New macro.
10562 (expand_ternary_optab_fn): Likewise.
10563 (direct_ternary_optab_supported_p): Likewise.
10564 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
10565 * builtins.c (fold_builtin_fma): Delete.
10566 (fold_builtin_3): Don't call it.
10567 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
10568 * expr.c (expand_expr_real_2): Likewise.
10569 * fold-const.c (operand_equal_p): Likewise.
10570 (fold_ternary_loc): Likewise.
10571 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
10572 * gimple.c (DEFTREECODE): Likewise.
10573 * gimplify.c (gimplify_expr): Likewise.
10574 * optabs-tree.c (optab_for_tree_code): Likewise.
10575 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10576 * tree-eh.c (operation_could_trap_p): Likewise.
10577 (stmt_could_throw_1_p): Likewise.
10578 * tree-inline.c (estimate_operator_cost): Likewise.
10579 * tree-pretty-print.c (dump_generic_node): Likewise.
10580 (op_code_prio): Likewise.
10581 * tree-ssa-loop-im.c (stmt_cost): Likewise.
10582 * tree-ssa-operands.c (get_expr_operands): Likewise.
10583 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
10584 * fold-const-call.h (fold_fma): Delete.
10585 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
10586 CFN_FNMA and CFN_FNMS.
10587 (fold_fma): Delete.
10588 * genmatch.c (combined_fn): New enum.
10589 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
10590 (commutative_op): New function.
10591 (commutate): Use it. Handle more than 2 operands.
10592 (dt_operand::gen_gimple_expr): Use commutative_op.
10593 (parser::parse_expr): Allow :c to be used with non-binary
10594 operators if the commutative operand is known.
10595 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
10596 CFN_FMS, CFN_FNMA and CFN_FNMS.
10597 (backprop::process_assign_use): Remove FMA_EXPR handling.
10598 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
10599 (gen_hsa_fma): New function.
10600 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
10601 IFN_FNMA and IFN_FNMS.
10602 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
10603 * gimple-fold.h (follow_all_ssa_edges): Declare.
10604 * gimple-fold.c (follow_all_ssa_edges): New function.
10605 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
10606 gimple_build interface and use follow_all_ssa_edges to fold the result.
10607 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
10608 instead of checking for optabs directly.
10609 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
10610 rather than FMA_EXPRs.
10611 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
10612 call to IFN_FMA instead of an FMA_EXPR.
10614 2018-05-17 Jim Wilson <jimw@sifive.com>
10616 * expr.c (do_tablejump): When converting index to Pmode, if we have a
10617 sign extended promoted subreg, and the range does not have the sign bit
10618 set, then do a sign extend.
10620 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
10621 test, check for sign extended subreg and/or constant operands, and
10622 do a sign extend in that case.
10624 2018-05-17 Steve Ellcey <sellcey@cavium.com>
10626 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
10627 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
10629 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
10630 Change logics_shift_reg to logics_shift_imm.
10631 (thunderx2t99_fp_loadpair_basic): Delete.
10632 (thunderx2t99_fp_storepair_basic): Delete.
10633 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
10634 (thunderx2t99_asimd_polynomial): Delete.
10635 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
10636 and neon_fp_mul_d_scalar_q.
10637 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
10638 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
10639 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
10640 (thunderx2t99_asimd_lut): Add missing tbl types.
10641 (thunderx2t99_asimd_ext): Delete.
10642 (thunderx2t99_asimd_load1_1_mult): Delete.
10643 (thunderx2t99_asimd_load1_2_mult): Delete.
10644 (thunderx2t99_asimd_load1_ldp): New.
10645 (thunderx2t99_asimd_load1): New.
10646 (thunderx2t99_asimd_load2): Add missing *load2* types.
10647 (thunderx2t99_asimd_load3): New.
10648 (thunderx2t99_asimd_load4): New.
10649 (thunderx2t99_asimd_store1_1_mult): Delete.
10650 (thunderx2t99_asimd_store1_2_mult): Delete.
10651 (thunderx2t99_asimd_store2_mult): Delete.
10652 (thunderx2t99_asimd_store2_onelane): Delete.
10653 (thunderx2t99_asimd_store_stp): New.
10654 (thunderx2t99_asimd_store1): New.
10655 (thunderx2t99_asimd_store2): New.
10656 (thunderx2t99_asimd_store3): New.
10657 (thunderx2t99_asimd_store4): New.
10659 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
10661 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
10662 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
10664 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
10665 Segher Boessenkool <segher@kernel.crashing.org>
10668 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
10671 2018-05-17 Richard Biener <rguenther@suse.de>
10673 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
10674 for pruning loop and prune defs feeding only already visited PHIs.
10676 2018-05-17 Richard Biener <rguenther@suse.de>
10678 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
10680 2018-05-17 Bin Cheng <bin.cheng@arm.com>
10681 Richard Biener <rguenther@suse.de>
10683 PR tree-optimization/85793
10684 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
10685 for VMAT_ELEMENTWISE.
10687 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
10689 * internal-fn.h (lookup_internal_fn): Declare
10690 * internal-fn.c (lookup_internal_fn): New function.
10691 * gimple.c (gimple_build_call_from_tree): Handle calls to
10692 internal functions.
10693 * gimple-pretty-print.c (dump_gimple_call): Print "." before
10694 internal function names.
10695 * tree-pretty-print.c (dump_generic_node): Likewise.
10696 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
10698 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
10700 * gimple-fold.h (gimple_build): Make the function forms take
10701 combined_fn rather than built_in_function.
10702 (gimple_simplify): Likewise.
10703 * gimple-match-head.c (gimple_simplify): Likewise.
10704 * gimple-fold.c (gimple_build): Likewise.
10705 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
10706 rather than gimple_build_call_internal.
10707 (get_initial_defs_for_reduction): Likewise.
10708 (vect_create_epilog_for_reduction): Likewise.
10709 (vectorizable_live_operation): Likewise.
10711 2018-05-17 Martin Liska <mliska@suse.cz>
10713 * gimple-ssa-sprintf.c (format_directive): Do not use
10714 space in between 'G_' and '('.
10716 2018-05-17 Jakub Jelinek <jakub@redhat.com>
10719 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
10720 even if the mask is not all ones.
10723 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
10725 (ix86_gimple_fold_builtin): Likewise.
10728 * config/i386/i386.c: Include tree-vector-builder.h.
10729 (ix86_vector_shift_count): New function.
10730 (ix86_fold_builtin): Fold shift builtins by scalar count.
10731 (ix86_gimple_fold_builtin): Likewise.
10733 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
10734 _mm512_setzero): New intrinsics.
10736 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
10737 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10739 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
10740 code generation for cases where splatting a value is not useful.
10741 * simplify-rtx.c (simplify_ternary_operation): Simplify
10742 vec_merge across a vec_duplicate and a paradoxical subreg forming
10743 a vector mode to a vec_concat.
10745 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
10747 * config.gcc: Support "goldmont-plus".
10748 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
10750 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10751 PROCESSOR_GOLDMONT_PLUS.
10752 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
10753 (processor_target_table): Add "goldmont-plus".
10754 (PTA_GOLDMONT_PLUS): Define.
10755 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
10756 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
10757 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
10758 (fold_builtin_cpu): Add "goldmont-plus".
10759 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
10760 (ix86_option_override_internal): Add "goldmont-plus".
10761 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
10762 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
10763 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
10764 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
10766 2018-05-17 Richard Biener <rguenther@suse.de>
10768 PR tree-optimization/85757
10769 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
10770 remove defs that only feed that PHI from further processing.
10772 2018-05-16 Jim Wilson <jimw@sifive.com>
10774 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
10776 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
10778 2018-05-16 Mark Wielaard <mark@klomp.org>
10780 * dwarf2out.c (count_index_strings): New function.
10781 (output_indirect_strings): Call count_index_strings and generate
10782 header for dwarf_version >= 5.
10784 2018-05-16 Mark Wielaard <mark@klomp.org>
10786 * dwarf2out.c (dwarf_FORM): New function.
10787 (set_indirect_string): Use dwarf_FORM.
10788 (reset_indirect_string): Likewise.
10789 (size_of_die): Likewise.
10790 (value_format): Likewise.
10791 (output_die): Likewise.
10792 (add_skeleton_AT_string): Likewise.
10793 (output_macinfo_op): Likewise.
10794 (index_string): Likewise.
10795 (output_index_string_offset): Likewise.
10796 (output_index_string): Likewise.
10797 (count_index_strings): Likewise.
10799 2018-05-16 Carl Love <cel@us.ibm.com>
10801 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
10802 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
10804 2018-05-16 Martin Jambor <mjambor@suse.cz>
10806 * ipa-prop.c (ipa_free_all_edge_args): Remove.
10807 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
10809 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
10811 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
10812 (fnma<mode>4): Likewise.
10813 (fms<mode>4): Likewise.
10814 (fnms<mode>4): Likewise.
10815 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
10816 (aarch64_fnma<mode>4): Likewise.
10817 (aarch64_fms<mode>4): Likewise.
10818 (aarch64_fnms<mode>4): Likewise.
10819 (aarch64_fnmadd<mode>4): Likewise.
10821 2018-05-16 Jason Merrill <jason@redhat.com>
10823 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
10825 2018-05-16 Richard Biener <rguenther@suse.de>
10827 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
10828 (dump_stmt_cost): Declare.
10829 (add_stmt_cost): Dump cost we add.
10830 (add_stmt_costs): New function.
10831 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
10832 No longer exported.
10833 (vect_analyze_stmt): Adjust prototype.
10834 (vectorizable_condition): Likewise.
10835 (vectorizable_live_operation): Likewise.
10836 (vectorizable_reduction): Likewise.
10837 (vectorizable_induction): Likewise.
10838 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
10839 cost vector to pass to vectorizable_ and record afterwards.
10840 (vect_model_reduction_cost): Take cost vector argument and adjust.
10841 (vect_model_induction_cost): Likewise.
10842 (vectorizable_reduction): Likewise.
10843 (vectorizable_induction): Likewise.
10844 (vectorizable_live_operation): Likewise.
10845 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
10846 SLP_TREE_NUMBER_OF_VEC_STMTS.
10847 (vect_analyze_slp_cost_1): Remove.
10848 (vect_analyze_slp_cost): Likewise.
10849 (vect_slp_analyze_node_operations): Take visited args and
10850 a target cost vector. Avoid processing already visited stmt sets.
10851 (vect_slp_analyze_operations): Use a local cost vector to gather
10852 costs and register those of non-discarded instances.
10853 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
10854 (vect_schedule_slp_instance): Remove copying of
10855 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
10857 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
10858 adding cost. Record cost entry location.
10859 (vect_prologue_cost_for_slp_op): Function to compute cost of
10860 a constant or invariant generated for SLP vect in the prologue,
10861 split out from vect_analyze_slp_cost_1.
10862 (vect_model_simple_cost): Make static. Adjust for SLP costing.
10863 (vect_model_promotion_demotion_cost): Likewise.
10864 (vect_model_store_cost): Likewise, make static.
10865 (vect_model_load_cost): Likewise.
10866 (vectorizable_bswap): Add cost vector arg and adjust.
10867 (vectorizable_call): Likewise.
10868 (vectorizable_simd_clone_call): Likewise.
10869 (vectorizable_conversion): Likewise.
10870 (vectorizable_assignment): Likewise.
10871 (vectorizable_shift): Likewise.
10872 (vectorizable_operation): Likewise.
10873 (vectorizable_store): Likewise.
10874 (vectorizable_load): Likewise.
10875 (vectorizable_condition): Likewise.
10876 (vectorizable_comparison): Likewise.
10877 (can_vectorize_live_stmts): Likewise.
10878 (vect_analyze_stmt): Likewise.
10879 (vect_transform_stmt): Adjust calls to vectorizable_*.
10880 * tree-vectorizer.c: Include gimple-pretty-print.h.
10881 (dump_stmt_cost): New function.
10883 2018-05-16 Richard Biener <rguenther@suse.de>
10885 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
10886 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
10887 * tree-ssa-dse.c: Include tree-ssa-loop.h.
10888 (check_name): New callback.
10889 (dse_classify_store): Track cycles via a visited bitmap of PHI
10890 defs and simplify handling of in-loop and across loop dead stores
10891 and properly fail for loop-variant refs. Handle byte-tracking with
10892 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
10895 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
10897 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
10898 (vect_get_mask_type_for_stmt): Likewise.
10899 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
10901 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
10902 to determine the statement's vector type and the vector type that
10903 should be used for calculating nunits. Deal with cases in which
10904 the type has to be deferred.
10905 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
10906 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
10907 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
10908 (vect_determine_vf_for_stmt): New functions, split out from...
10909 (vect_determine_vectorization_factor): ...here.
10910 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
10911 (vect_get_mask_type_for_stmt): New functions, split out from
10912 vect_determine_vectorization_factor.
10914 2018-05-16 Richard Biener <rguenther@suse.de>
10916 * tree-cfg.c (verify_gimple_assign_ternary): Properly
10917 verify the [VEC_]COND_EXPR embedded comparison.
10919 2018-05-15 Martin Sebor <msebor@redhat.com>
10921 PR tree-optimization/85753
10922 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
10923 RECORD_TYPE in addition to ARRAY_TYPE.
10925 2018-05-15 Martin Sebor <msebor@redhat.com>
10927 PR middle-end/85643
10928 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
10930 2018-05-15 Richard Biener <rguenther@suse.de>
10932 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
10933 add by_clobber_p one. Change algorithm to collect all defs
10934 representing uses we need to walk and try reducing them to
10935 a single one before failing.
10936 (dse_dom_walker::dse_optimize_stmt): Adjust.
10938 2018-05-13 Mark Wielaard <mark@klomp.org>
10940 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
10941 (size_of_loc_descr): Likewise.
10942 (output_loc_operands): Likewise.
10943 (output_loc_operands_raw): Likewise.
10944 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
10945 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
10946 (hash_loc_operands): Likewise.
10947 (compare_loc_operands): Likewise.
10949 2018-05-14 Mark Wielaard <mark@klomp.org>
10951 * dwarf2out.c (count_index_addrs): New function.
10952 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
10954 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10956 PR tree-optimization/83648
10957 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
10958 return value as malloc candidate.
10960 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10963 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
10964 param as true in call to suggest_attribute.
10966 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
10968 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
10969 -mreadonly-in-sdata.
10971 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10973 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
10975 (aarch64_crypto_aesd_fused): Likewise.
10977 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
10979 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
10980 (movsi_aarch64): Likewise.
10981 (load_pairsi): Likewise.
10982 (load_pairdi): Likewise.
10983 (store_pairsi): Likewise.
10984 (store_pairdi): Likewise.
10985 (load_pairsf): Likewise.
10986 (load_pairdf): Likewise.
10987 (store_pairsf): Likewise.
10988 (store_pairdf): Likewise.
10989 (zero_extend): Likewise.
10990 (trunc): Swap alternatives.
10991 (fcvt_target): Add '?' to prefer w over r.
10993 2018-05-14 Jakub Jelinek <jakub@redhat.com>
10996 * config/i386/i386.md: Disallow non-commutative arithmetics in
10997 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
10998 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
10999 in the peephole2 before it.
11001 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
11003 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
11004 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
11005 (ix86_handle_option): Handle -mcldemote.
11006 * config.gcc: New header.
11007 * config/i386/cldemoteintrin.h: New file.
11008 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
11009 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11011 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11012 OPTION_MASK_ISA_CLDEMOTE.
11013 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
11014 (ix86_valid_target_attribute_inner_p): Ditto.
11015 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
11016 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
11017 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
11018 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
11019 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
11021 * config/i386/i386.opt: Add -mcldemote.
11022 * config/i386/x86intrin.h: New header.
11023 * doc/invoke.texi: Add -mcldemote.
11025 2018-05-14 Richard Biener <rguenther@suse.de>
11027 * doc/match-and-simplify.texi: Adjust :s documentation.
11029 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
11031 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
11032 intended memcpy size.
11033 (REORDER_45): Likewise.
11035 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
11037 * sort.cc: New file.
11038 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
11039 * vec.c (qsort_chk): Use gcc_qsort.
11040 * Makefile.in (OBJS-libcommon): Add sort.o.
11041 (build/sort.o): New target. Use it...
11042 (BUILD_RTL): ... here, and...
11043 (build/gencfn-macros): ... here, and...
11044 (build/genmatch): ... here.
11046 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11047 Chung-Ju Wu <jasonwucj@gmail.com>
11049 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
11050 * config/nds32/nds32-graywolf.md: New file.
11051 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
11052 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
11054 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
11055 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
11056 * config/nds32/nds32.md (pipeline_model): Add graywolf.
11057 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
11058 * config/nds32/pipelines.md: Include n15 settings.
11060 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11061 Chung-Ju Wu <jasonwucj@gmail.com>
11063 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
11064 * config/nds32/nds32-n13.md: New file.
11065 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
11066 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
11068 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
11069 * config/nds32/nds32.md (pipeline_model): Add n13.
11070 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
11071 * config/nds32/pipelines.md: Include n13 settings.
11073 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11074 Chung-Ju Wu <jasonwucj@gmail.com>
11076 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
11077 * config/nds32/nds32-n10.md: New file.
11078 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
11079 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
11081 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
11082 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
11083 * config/nds32/nds32.md (pipeline_model): Add n10.
11084 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
11085 * config/nds32/pipelines.md: Include n10 settings.
11087 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
11088 Kito Cheng <kito.cheng@gmail.com>
11089 Chung-Ju Wu <jasonwucj@gmail.com>
11091 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
11092 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
11093 Add enum values for DSP extension instructions.
11094 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
11096 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
11097 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
11098 New code iterators.
11099 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
11100 * config/nds32/nds32-dspext.md: New file for DSP implementation.
11101 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
11102 * config/nds32/nds32-intrinsic.md: Likewise.
11103 * config/nds32/nds32_intrinsic.h: Likewise.
11104 * config/nds32/nds32-md-auxiliary.c: Likewise.
11105 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
11106 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
11107 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
11108 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
11109 * config/nds32/nds32-protos.h: New declarations for DSP extension.
11110 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
11111 TYPE_DMAC in switch statement.
11112 * config/nds32/nds32.c: New checking and implementation for DSP
11113 extension instructions.
11114 * config/nds32/nds32.h: Likewise.
11115 * config/nds32/nds32.md: Likewise.
11116 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
11117 * config/nds32/predicates.md: Implement new predicates for DSP
11120 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
11122 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
11123 Reformat alternatives and attributes so it is easier to identify
11124 which constraints/attributes go with which instruction.
11125 (mov<mode>_hardfloat32, FMOVE64): Likewise.
11126 (mov<mode>_softfloat32, FMOVE64): Likewise.
11127 (mov<mode>_hardfloat64, FMOVE64): Likewise.
11128 (mov<mode>_softfloat64, FMOVE64): Likewise.
11130 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
11132 * doc/extend.texi (PowerPC Built-in Functions): Rename this
11134 (Basic PowerPC Built-in Functions): The new name of the
11135 subsection previously known as "PowerPC Built-in Functions".
11136 (Basic PowerPC Built-in Functions Available on all Configurations):
11138 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
11139 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
11140 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
11141 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
11143 2018-05-11 Martin Jambor <mjambor@suse.cz>
11146 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
11149 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
11152 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
11154 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
11156 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
11157 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
11158 (ix86_handle_option): Handle -mwaitpkg.
11159 * config.gcc: New header.
11160 * config/i386/cpuid.h (bit_WAITPKG): New bit.
11161 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
11162 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
11164 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11165 OPTION_MASK_ISA_WAITPKG.
11166 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
11167 (ix86_option_override_internal): Add PTA_WAITPKG.
11168 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
11169 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
11170 IX86_BUILTIN_TPAUSE.
11171 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
11172 __builtin_ia32_umwait and __builtin_ia32_tpause.
11173 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
11174 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
11175 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
11176 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
11177 UNSPECV_TPAUSE): New.
11178 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
11179 * config/i386/i386.opt: Add -mwaitpkg.
11180 * config/i386/waitpkgintrin.h: New file.
11181 * config/i386/x86intrin.h: New header.
11182 * doc/invoke.texi: Add -mwaitpkg.
11184 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
11187 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
11189 (cortex-m0): Use armv6s-m isa.
11190 (cortex-m0plus): Likewise.
11191 (cortex-m1): Likewise.
11192 (cortex-m0.small-multiply): Likewise.
11193 (cortex-m0plus.small-multiply): Likewise.
11194 (cortex-m1.small-multiply): Likewise.
11196 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
11197 Jakub Jelinek <jakub@redhat.com>
11199 PR tree-optimization/85692
11200 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
11201 source permute as well.
11203 2018-05-11 Martin Liska <mliska@suse.cz>
11206 * doc/extend.texi: Document LLVM style format for no_sanitize
11209 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
11211 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
11212 mode_supports_vsx_dform_quad to mode_supports_dq_form.
11213 (mode_supports_vsx_dform_quad): Likewise.
11214 (mode_supports_vmx_dform): Move these functions to be next to the
11215 other mode_supports functions.
11216 (mode_supports_dq_form): Likewise.
11217 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
11218 mode_supports_dq_form.
11219 (reg_offset_addressing_ok_p): Likewise.
11220 (offsettable_ok_by_alignment): Likewise.
11221 (rs6000_legitimate_offset_address_p): Likewise.
11222 (legitimate_lo_sum_address_p): Likewise.
11223 (rs6000_legitimize_address): Likewise.
11224 (rs6000_legitimize_reload_address): Likewise.
11225 (rs6000_secondary_reload_inner): Likewise.
11226 (rs6000_preferred_reload_class): Likewise.
11227 (rs6000_output_move_128bit): Likewise.
11229 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
11231 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
11232 Generate SImode target register for null target.
11233 <case IX86_BUILTIN_XGETBV>: Ditto.
11234 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
11235 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
11237 2018-05-10 Carl Love <cel@us.ibm.com>
11239 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
11240 dcbtt and dcbtstt if operands[2] is 0.
11242 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
11245 * config/i386/sse.md (usadv64qi): New expander.
11247 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
11249 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
11250 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
11251 -maltivec=be support.
11252 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
11253 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
11254 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11255 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
11256 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
11257 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
11258 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
11259 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
11260 altivec_vsumsws): Adjust.
11261 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
11262 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
11263 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
11265 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
11266 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
11267 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
11268 (altivec_lve<VI_char>x): Delete expand.
11269 (*altivec_lve<VI_char>x_internal): Rename to...
11270 (altivec_lve<VI_char>x): ... this.
11271 (altivec_lvxl_<mode>): Delete expand.
11272 (*altivec_lvxl_<mode>_internal): Rename to ...
11273 (altivec_lvxl_<mode>): ... this.
11274 (altivec_stvxl_<mode>): Delete expand.
11275 (*altivec_stvxl_<mode>_internal): Rename to ...
11276 (altivec_stvxl_<mode>): ... this.
11277 (altivec_stve<VI_char>x): Delete expand.
11278 (*altivec_stve<VI_char>x_internal): Rename to ...
11279 (altivec_stve<VI_char>x): ... this.
11280 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
11281 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
11282 reduc_plus_scal_<mode>): Adjust.
11283 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
11285 (rs6000_cpu_cpp_builtins): Adjust.
11286 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
11287 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
11288 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
11289 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
11290 -maltivec=be support.
11291 (rs6000_split_vec_extract_var): Adjust.
11292 (rs6000_split_v4si_init): Adjust.
11293 (swap_selector_for_mode): Delete.
11294 (altivec_expand_lvx_be, altivec_expand_stvx_be,
11295 altivec_expand_stvex_be): Delete.
11296 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
11297 -maltivec=be support.
11298 (rs6000_gimple_fold_builtin): Ditto.
11299 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
11301 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
11302 (TARGET_DIRECT_MOVE_64BIT): Adjust.
11303 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
11304 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
11305 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
11306 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
11307 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
11308 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
11309 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
11310 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
11311 anonymous split): Adjust.
11312 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
11313 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
11315 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
11317 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
11318 when --with-gxx-include-dir is also specified.
11319 * configure: Regenerate.
11321 2018-05-09 Jim Wilson <jimw@sifive.com>
11324 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
11325 * config/riscv/t-withmultilib: New.
11326 * config/riscv/withmultilib.h: New.
11327 * doc/install.texi: Document RISC-V --with-multilib-list support.
11329 2018-05-09 Richard Biener <rguenther@suse.de>
11331 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
11333 (vect_bb_vectorization_profitable_p): Adjust. Compute
11334 actual scalar cost using the cost vector and the add_stmt_cost
11337 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
11339 PR rtl-optimization/85645
11340 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
11341 in the REG_CFA_REGISTER note for LR, don't leave it empty.
11343 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
11345 PR rtl-optimization/85645
11346 * shrink-wrap.c (spread_components): Return a boolean saying if
11347 anything was changed.
11348 (try_shrink_wrapping_separate): Iterate spread_components until
11349 nothing changes anymore.
11351 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
11353 PR rtl-optimization/85645
11354 * regrename.c (build_def_use): Also kill the chains that include the
11355 destination of a REG_CFA_REGISTER note.
11357 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
11359 PR rtl-optimization/85645
11360 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
11361 insn that has a REG_CFA_REGISTER note.
11363 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
11365 * cfgexpand.c (expand_clobber): New function.
11366 (expand_gimple_stmt_1): Use it.
11367 * tree-vect-stmts.c (vect_clobber_variable): New function,
11369 (vectorizable_simd_clone_call): ...here.
11370 (vectorizable_store): Emit a clobber either side of an
11371 IFN_STORE_LANES sequence.
11372 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
11374 2018-05-09 Tom de Vries <tom@codesourcery.com>
11377 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
11378 (define_insn "trap_if_false"): Add exit after trap.
11380 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
11382 PR rtl-optimization/85638
11383 * bb-reorder.c: Include common/common-target.h.
11384 (create_forwarder_block): New function extracted from...
11385 (fix_up_crossing_landing_pad): ...here. Rename into...
11386 (dw2_fix_up_crossing_landing_pad): ...this.
11387 (sjlj_fix_up_crossing_landing_pad): New function.
11388 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
11389 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
11390 from both partitions and exit the loop after one iteration.
11392 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
11395 * doc/extend.texi (PowerPC Built-in Functions): Rename this
11397 (Basic PowerPC Built-in Functions): The new name of the
11398 subsection previously known as "PowerPC Built-in Functions".
11399 (Basic PowerPC Built-in Functions Available on all Configurations):
11401 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
11403 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
11405 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
11407 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
11410 2018-05-08 Jim Wilson <jimw@sifive.com>
11412 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
11413 (LD_EMUL_SUFFIX): New.
11414 (LINK_SPEC): Use it.
11416 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
11418 * doc/extend.texi (PowerPC Built-in Functions): Rename this
11420 (Basic PowerPC Built-in Functions): The new name of the
11421 subsection previously known as "PowerPC Built-in Functions".
11422 (Basic PowerPC Built-in Functions Available on all Configurations):
11424 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
11426 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
11428 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
11430 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
11433 2018-05-08 Jakub Jelinek <jakub@redhat.com>
11436 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
11437 after cmpelim optimization.
11439 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
11441 * config.gcc: Support "goldmont".
11442 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
11443 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11444 PROCESSOR_GOLDMONT.
11445 * config/i386/i386.c (m_GOLDMONT): Define.
11446 (processor_target_table): Add "goldmont".
11447 (PTA_GOLDMONT): Define.
11448 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
11449 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
11450 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
11451 (fold_builtin_cpu): Add "goldmont".
11452 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
11453 (ix86_option_override_internal): Add "goldmont".
11454 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
11455 (processor_type): Add PROCESSOR_GOLDMONT.
11456 * config/i386/i386.md: Add CPU "glm".
11457 * config/i386/glm.md: New file.
11458 * config/i386/x86-tune.def: Add m_GOLDMONT.
11459 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
11461 2018-05-08 Jakub Jelinek <jakub@redhat.com>
11464 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
11466 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
11467 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
11468 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
11471 * config/i386/i386.c (ix86_fold_builtin): Handle
11472 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
11475 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
11476 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
11478 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
11481 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
11482 (check_arch): Likewise.
11483 (check_fpu): Return the result rather than printing it.
11484 (end arch): Fix operator precedence.
11485 (end cpu): Likewise.
11486 (END): Print the result from check_fpu.
11488 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
11489 Alan Hayward <alan.hayward@arm.com>
11490 David Sherwood <david.sherwood@arm.com>
11492 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
11493 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
11494 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
11495 (*fcmuo<mode>_and): New patterns.
11497 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
11499 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
11500 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
11501 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
11502 (cmp_op, sve_imm_con): New code attributes.
11503 (SVE_COND_INT_CMP, imm_con): Delete.
11504 (cmp_op): Remove above unspecs from int attribute.
11505 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
11507 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
11508 comparison-specific unspecs.
11509 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
11510 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
11511 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
11512 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
11513 (*vec_fcm<cmp_op><mode>): Rename to...
11514 (*fcm<cmp_op><mode>): ...this and adjust likewise.
11515 (*vec_fcmuo<mode>): Rename to...
11516 (*fcmuo<mode>): ...this and adjust likewise.
11517 (*pred_fcm<cmp_op><mode>): New pattern.
11518 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
11519 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
11521 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
11523 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
11524 (aarch64_emit_sve_predicated_cond): New function.
11525 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
11526 (aarch64_emit_unspec_cond_or): Replace with...
11527 (aarch64_emit_sve_or_conds): ...this new function. Use
11528 aarch64_emit_sve_ptrue_op for the individual comparisons and
11529 aarch64_emit_binop to OR them together.
11530 (aarch64_emit_inverted_unspec_cond): Replace with...
11531 (aarch64_emit_sve_inverted_cond): ...this new function. Use
11532 aarch64_emit_sve_ptrue_op for the comparison and
11533 aarch64_emit_unop to invert the result.
11534 (aarch64_expand_sve_vec_cmp_float): Update after the above
11535 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
11537 2018-05-07 Nathan Sidwell <nathan@acm.org>
11539 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
11540 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
11541 (Backwards Compatibility): Likewise.
11543 2018-05-07 Luis Machado <luis.machado@linaro.org>
11547 2018-05-07 Luis Machado <luis.machado@linaro.org>
11549 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11550 <prefetch_dynamic_strides>: New const bool field.
11551 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11552 prefetch_dynamic_strides.
11553 (exynosm1_prefetch_tune): Likewise.
11554 (thunderxt88_prefetch_tune): Likewise.
11555 (thunderx_prefetch_tune): Likewise.
11556 (thunderx2t99_prefetch_tune): Likewise.
11557 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
11559 (aarch64_override_options_internal): Update to set
11560 PARAM_PREFETCH_DYNAMIC_STRIDES.
11561 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11562 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11563 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11564 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11565 prefetch-dynamic-strides setting.
11567 2018-05-07 Luis Machado <luis.machado@linaro.org>
11569 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11570 <minimum_stride>: New const int field.
11571 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11572 minimum_stride field.
11573 (exynosm1_prefetch_tune): Likewise.
11574 (thunderxt88_prefetch_tune): Likewise.
11575 (thunderx_prefetch_tune): Likewise.
11576 (thunderx2t99_prefetch_tune): Likewise.
11577 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
11578 (aarch64_override_options_internal): Update to set
11579 PARAM_PREFETCH_MINIMUM_STRIDE.
11580 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11581 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11582 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11583 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11584 stride is constant and is below the minimum stride threshold.
11586 2018-05-07 Luis Machado <luis.machado@linaro.org>
11588 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
11591 2018-05-07 Luis Machado <luis.machado@linaro.org>
11593 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11594 <prefetch_dynamic_strides>: New const bool field.
11595 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11596 prefetch_dynamic_strides.
11597 (exynosm1_prefetch_tune): Likewise.
11598 (thunderxt88_prefetch_tune): Likewise.
11599 (thunderx_prefetch_tune): Likewise.
11600 (thunderx2t99_prefetch_tune): Likewise.
11601 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
11603 (aarch64_override_options_internal): Update to set
11604 PARAM_PREFETCH_DYNAMIC_STRIDES.
11605 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11606 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11607 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11608 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11609 prefetch-dynamic-strides setting.
11611 2018-05-07 Luis Machado <luis.machado@linaro.org>
11613 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11614 <minimum_stride>: New const int field.
11615 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11616 minimum_stride field.
11617 (exynosm1_prefetch_tune): Likewise.
11618 (thunderxt88_prefetch_tune): Likewise.
11619 (thunderx_prefetch_tune): Likewise.
11620 (thunderx2t99_prefetch_tune): Likewise.
11621 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
11622 (aarch64_override_options_internal): Update to set
11623 PARAM_PREFETCH_MINIMUM_STRIDE.
11624 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11625 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11626 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11627 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11628 stride is constant and is below the minimum stride threshold.
11630 2018-05-06 Jakub Jelinek <jakub@redhat.com>
11633 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
11634 the type is addressable. Don't force op into register if it has
11637 2018-05-05 Roland McGrath <mcgrathr@google.com>
11640 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
11641 any section for which we don't know a specific type it should have,
11642 regardless of name. Previously this was done only for the exact
11643 names ".init_array", ".fini_array", and ".preinit_array".
11644 (default_elf_asm_named_section): Add comment about
11645 relationship with default_section_type_flags and SECTION_NOTYPE.
11646 (get_section): Don't consider it a type conflict if one side has
11647 SECTION_NOTYPE and the other doesn't, as long as neither has the
11648 SECTION_BSS et al used in the default_section_type_flags logic.
11650 2018-05-05 Tom de Vries <tom@codesourcery.com>
11653 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
11654 (workaround_barsyncs): New function.
11655 (nvptx_reorg): Use workaround_barsyncs.
11656 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
11657 (define_expand "nvptx_membar_cta"): New define_expand.
11658 (define_insn "*nvptx_membar_cta"): New insn.
11660 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
11662 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
11663 To improve optimization opportunities.
11664 * builtin-types.def: The new needed builtin types for the above.
11666 2018-05-04 Richard Biener <rguenther@suse.de>
11668 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
11669 * gimple-ssa-store-merging.c
11670 (imm_store_chain_info::output_merged_store): Remove redundant create,
11671 release split_store vector contents on failure.
11672 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
11673 scalar stmt vector on cache hit.
11675 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
11677 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11679 * config.gcc (powerpc-xilinx-eabi*): Remove.
11680 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
11682 (fusion_addis_mem_combo_load): Ditto.
11683 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
11685 (rs6000_cpu_cpp_builtins): Ditto.
11686 * config/rs6000/rs6000-linux.c
11687 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
11688 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
11689 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
11691 (rs6000_setup_reg_addr_masks): Ditto.
11692 (rs6000_init_hard_regno_mode_ok): Ditto.
11693 (rs6000_option_override_internal): Ditto.
11694 (legitimate_lo_sum_address_p): Ditto.
11695 (rs6000_legitimize_address): Ditto.
11696 (rs6000_legitimize_reload_address): Ditto.
11697 (rs6000_legitimate_address_p): Ditto.
11698 (abi_v4_pass_in_fpr): Ditto.
11699 (setup_incoming_varargs): Ditto.
11700 (rs6000_gimplify_va_arg): Ditto.
11701 (rs6000_split_multireg_move): Ditto.
11702 (rs6000_savres_strategy): Ditto.
11703 (rs6000_emit_prologue_components): Ditto.
11704 (rs6000_emit_epilogue_components): Ditto.
11705 (rs6000_emit_prologue): Ditto.
11706 (rs6000_emit_epilogue): Ditto.
11707 (rs6000_elf_file_end): Ditto.
11708 (rs6000_function_value): Ditto.
11709 (rs6000_libcall_value): Ditto.
11710 * config/rs6000/rs6000.h: Ditto.
11711 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
11712 (TARGET_MINMAX): ... this. New.
11713 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
11714 * config/rs6000/rs6000.md: Remove Xilinx FP support.
11715 (*movsi_internal1_single): Delete.
11716 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
11717 mfpu=, mxilinx-fpu): Delete.
11718 * config/rs6000/singlefp.h: Delete.
11719 * config/rs6000/sysv4.h: Remove Xilinx FP support.
11720 * config/rs6000/t-rs6000: Ditto.
11721 * config/rs6000/t-xilinx: Delete.
11722 * config/rs6000/titan.md: Adjust for fp_type removal.
11723 * config/rs6000/vsx.md: Remove Xilinx FP support.
11724 (VStype_simple): Delete.
11725 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
11726 * config/rs6000/xfpu.h: Delete.
11727 * config/rs6000/xfpu.md: Delete.
11728 * config/rs6000/xilinx.h: Delete.
11729 * config/rs6000/xilinx.opt: Delete.
11730 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
11731 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
11733 2018-05-04 Tom de Vries <tom@codesourcery.com>
11736 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
11739 2018-05-04 Richard Biener <rguenther@suse.de>
11741 PR middle-end/85627
11742 * tree-complex.c (update_complex_assignment): We are always in SSA form.
11743 (expand_complex_div_wide): Likewise.
11744 (expand_complex_operations_1): Likewise.
11745 (expand_complex_libcall): Preserve EH info of the original stmt.
11746 (tree_lower_complex): Handle removed blocks.
11747 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
11748 on complex multiplication and division libcall builtins.
11750 2018-05-04 Richard Biener <rguenther@suse.de>
11752 PR middle-end/85574
11753 * fold-const.c (negate_expr_p): Restrict negation of operand
11754 zero of a division to when we know that can happen without
11756 (fold_negate_expr_1): Likewise.
11758 2018-05-04 Jakub Jelinek <jakub@redhat.com>
11761 * real.h (real_nextafter): Declare.
11762 * real.c (real_nextafter): New function.
11763 * fold-const-call.c (fold_const_nextafter): New function.
11764 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
11765 CASE_CFN_NEXTTOWARD.
11766 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
11767 even when arg1_mode is different from arg0_mode.
11769 2018-05-03 Nathan Sidwell <nathan@acm.org>
11771 * doc/extend.texi (Deprecated Features): Remove
11772 -ffriend-injection.
11773 (Backwards Compatibility): Likewise.
11774 * doc/invoke.texi (C++ Language Options): Likewise.
11775 (C++ Dialect Options): Likewise.
11777 2018-05-03 Jakub Jelinek <jakub@redhat.com>
11780 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
11781 _mm512_mask_mullox_epi64): New intrinsics.
11783 2018-05-03 Tom de Vries <tom@codesourcery.com>
11786 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
11787 dump files): Add offload-tree.
11789 2018-05-03 Richard Biener <rguenther@suse.de>
11791 PR tree-optimization/85615
11792 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
11793 to loops not nested in BBs loop father to avoid creating multi-entry
11796 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11798 PR tree-optimization/70291
11799 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
11800 arguments. Change return type to tree. Emit libcall as a new
11801 statement rather than replacing existing one when inplace_p is true.
11802 (expand_complex_multiplication_components): New function.
11803 (expand_complex_multiplication): Expand floating-point complex
11804 multiplication using the above.
11805 (expand_complex_division): Rename inner_type parameter to type.
11806 Update expand_complex_libcall call-site.
11807 (expand_complex_operations_1): Update expand_complex_multiplication
11808 and expand_complex_division call-sites.
11810 2018-05-02 Jakub Jelinek <jakub@redhat.com>
11813 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
11814 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
11815 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
11816 the highest significant bit of the shift count mask is clear. In
11817 check whether and[sq]i3 is needed verify that all significant bits
11818 of the shift count other than the highest are set.
11820 2018-05-02 Tom de Vries <tom@codesourcery.com>
11823 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
11824 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
11825 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
11826 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
11827 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
11828 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
11829 __builtin_goacc_parlevel_size.
11831 2018-05-02 Richard Biener <rguenther@suse.de>
11833 PR tree-optimization/85597
11834 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
11835 do not use split vect_get_vec_defs call but call vect_get_slp_defs
11838 2018-05-02 Tom de Vries <tom@codesourcery.com>
11841 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
11842 dump files): Add ltrans-tree.
11844 2018-05-02 Tom de Vries <tom@codesourcery.com>
11847 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
11848 dump files): Add wpa-ipa.
11850 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
11852 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
11853 powerpc*-*-linux*paired* target.
11854 * config/rs6000/750cl.h: Delete.
11855 * config/rs6000/paired.h: Delete.
11856 * config/rs6000/paired.md: Delete.
11857 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
11859 * config/rs6000/rs6000-builtin.def: Remove paired float support.
11860 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11861 comment. Remove paired float support.
11862 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
11863 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
11865 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
11866 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
11868 * config/rs6000/rs6000.c: Remove paired float support.
11869 (paired_expand_vector_init, paired_expand_vector_move,
11870 paired_emit_vector_compare, paired_emit_vector_cond_expr,
11871 (paired_expand_lv_builtin, paired_expand_stv_builtin,
11872 paired_expand_builtin, paired_expand_predicate_builtin,
11873 paired_init_builtins): Delete.
11874 * config/rs6000/rs6000.h: Remove paired float support.
11875 * config/rs6000/rs6000.md: Remove paired float support.
11876 (move_from_CR_ov_bit): Delete.
11877 * config/rs6000/rs6000.opt (mpaired): Delete.
11878 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
11879 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
11881 2018-05-02 Richard Biener <rguenther@suse.de>
11883 PR middle-end/85567
11884 * gimplify.c (gimplify_save_expr): When in SSA form allow
11885 SAVE_EXPRs to compute to SSA vars.
11887 2018-05-02 Jakub Jelinek <jakub@redhat.com>
11890 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
11891 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
11892 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
11893 clobber operands[2], instead use a new pseudo. Formatting fixes.
11895 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
11897 PR tree-optimization/85586
11898 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
11899 exit early for statements in the same group if the accesses are
11902 2018-05-02 Tom de Vries <tom@codesourcery.com>
11905 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
11908 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
11910 PR tree-optimization/85143
11911 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
11913 2018-05-01 Tom de Vries <tom@codesourcery.com>
11916 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
11917 not found" error message.
11919 2018-05-01 Tom de Vries <tom@codesourcery.com>
11922 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
11923 * vec.c (test_ordered_remove_if): New function.
11924 (vec_c_tests): Call test_ordered_remove_if.
11925 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
11926 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
11927 * tree-vect-patterns.c (vect_pattern_recog_1): Use
11928 VEC_ORDERED_REMOVE_IF.
11930 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11932 PR tree-optimization/82665
11933 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
11934 pointer subtraction where arguments come from a memchr call.
11936 2018-05-01 Jakub Jelinek <jakub@redhat.com>
11938 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
11939 --push-state --as-needed and --pop-state instead of --as-needed and
11940 --no-as-needed if ld supports it.
11941 * configure: Regenerated.
11944 * doc/install.texi2html: Replace _002d with - and _002a with * in
11945 generated html files using sed.
11947 2018-04-30 David Malcolm <dmalcolm@redhat.com>
11950 * gcc-rich-location.c (blank_line_before_p): New function.
11951 (use_new_line): New function.
11952 (gcc_rich_location::add_fixit_insert_formatted): New function.
11953 * gcc-rich-location.h
11954 (gcc_rich_location::add_fixit_insert_formatted): New function.
11956 2018-04-30 David Malcolm <dmalcolm@redhat.com>
11958 * selftest.c (assert_streq): Rename "expected" and "actual" to
11959 "val1" and "val2". Extend NULL-handling to cover both inputs
11960 symmetrically, while still requiring both to be non-NULL for a pass.
11961 * selftest.h (assert_streq): Rename "expected" and "actual" to
11963 (ASSERT_EQ): Likewise.
11964 (ASSERT_EQ_AT): Likewise.
11965 (ASSERT_KNOWN_EQ): Likewise.
11966 (ASSERT_KNOWN_EQ_AT): Likewise.
11967 (ASSERT_NE): Likewise.
11968 (ASSERT_MAYBE_NE): Likewise.
11969 (ASSERT_MAYBE_NE_AT): Likewise.
11970 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
11971 the assertion to pass.
11972 (ASSERT_STREQ_AT): Likewise.
11974 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
11976 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
11977 interaction with -pie.
11979 2018-04-30 David Malcolm <dmalcolm@redhat.com>
11981 * selftest.h: Fix alphabetization of per-source-file selftest
11984 2018-04-30 Jason Merrill <jason@redhat.com>
11986 PR c++/61982 - dead stores to destroyed objects.
11987 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
11990 2018-04-30 Jason Merrill <jason@redhat.com>
11992 * tree.c (build_clobber): New.
11993 * tree.h: Declare it.
11994 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
11996 2018-04-30 David Malcolm <dmalcolm@redhat.com>
11998 * diagnostic-show-locus.c (layout::layout): Update for
11999 location_get_source_line returning a char_span.
12000 (struct char_span): Move to input.h.
12001 (struct correction): Update for fields in char_span becoming
12003 (struct source_line): Update for location_get_source_line
12004 returning a char_span.
12005 (layout::print_line): Likewise.
12006 * edit-context.c (edited_file::print_content): Likewise.
12007 (edited_file::print_diff_hunk): Likewise.
12008 (edited_file::print_run_of_changed_lines): Likewise.
12009 (edited_file::get_num_lines): Likewise.
12010 (edited_line::edited_line): Likewise.
12011 * final.c (asm_show_source): Likewise.
12012 * input.c (location_get_source_line): Convert return type
12013 from const char * to char_span, losing the final "line_len"
12015 (dump_location_info): Update for the above.
12016 (get_substring_ranges_for_loc): Likewise. Use a char_span
12017 when handling the literal within the line.
12018 (test_reading_source_line): Update for location_get_source_line
12019 returning a char_span.
12020 * input.h (class char_span): Move here from
12021 diagnostic-show-locus.c, converting from a struct to a class.
12022 Make data members private.
12023 (char_span::operator bool): New.
12024 (char_span::length): New.
12025 (char_span::get_buffer): New.
12026 (char_span::operator[]): New.
12027 (char_span::subspan): Make const.
12028 (char_span::xstrdup): New.
12029 (location_get_source_line): Convert return type from const char *
12030 to char_span, losing the final "line_size" param.
12032 2018-04-30 Jan Hubicka <jh@suse.cz>
12034 * lto-wrapper.c (ltrans_priorities): New static var.
12035 (cmp_priority): New.
12036 (run_gcc): Read priorities and if doing parallel build order
12037 the Makefile by them.
12039 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12041 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
12043 2018-04-30 Richard Biener <rguenther@suse.de>
12045 * tree-cfg.c (verify_address): Remove base argument, add
12046 flag whether to check TREE_ADDRESSABLE and do that.
12047 (verify_expr): Remove.
12048 (verify_types_in_gimple_reference): Add pieces from verify_expr.
12049 (verify_gimple_assign_single): Likewise.
12050 (verify_gimple_switch): Likewise.
12051 (verify_expr_location_1): Dereference tp once. Add (disabled)
12052 piece from verify_expr.
12053 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
12055 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12057 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
12059 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12061 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
12062 (small_data_pattern): Likewise.
12063 (arc_rewrite_small_data): Likewise.
12064 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
12065 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
12066 (get_symbol_alignment): New function.
12067 (legitimate_small_data_address_p): Likewise.
12068 (legitimate_scaled_address): Update, call
12069 legitimate_small_data_address_p.
12070 (output_sdata): New static variable.
12071 (arc_print_operand): Update how we handle small data operands.
12072 (arc_print_operand_address): Likewise.
12073 (arc_legitimate_address_p): Update, use
12074 legitimate_small_data_address_p.
12075 (arc_rewrite_small_data_p): Remove.
12076 (arc_rewrite_small_data_1): Likewise.
12077 (arc_rewrite_small_data): Likewise.
12078 (small_data_pattern): Likewise.
12079 (compact_sda_memory_operand): Update to use
12080 legitimate_small_data_address_p and get_symbol_alignment.
12081 (prepare_move_operands): Don't rewite sdata pattern.
12082 (prepare_extend_operands): Remove.
12083 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
12085 (zero_extendqisi2): Likewise.
12086 (zero_extendhisi2): Likewise.
12087 (extendqihi2): Likewise.
12088 (extendqisi2): Likewise.
12089 (extendhisi2): Likewise.
12090 (addsi3): Likewise.
12091 (subsi3): Likewise.
12092 (andsi3): Likewise.
12093 * config/arc/constraints.md (Usd): Change it to memory constraint.
12095 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12097 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
12098 as source of std instructions.
12099 * config/arc/arc.md (movsi_insn): Update pattern predicate to
12100 allow 6-bit constants as source for store instructions.
12101 (movdi_insn): Update instruction pattern to allow 6-bit constants
12102 as source for store instructions.
12104 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
12106 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
12108 2018-04-30 Nathan Sidwell <nathan@acm.org>
12109 Sandra Loosemore <sandra@codesourcery.com>
12111 * dumpfile.c (dump_open): Allow '-' for stdout.
12112 * doc/invoke.texi (Developer Options): Document dump filename
12113 determination early. Document stdin/stdout selection.
12115 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
12117 Microblaze Target: PIC data text relative
12119 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
12120 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
12122 * config/microblaze/microblaze.h (microblaze_constant_address_p):
12123 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
12124 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
12125 New addressing mode for data-text relative position indepenedent code.
12126 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
12127 'ADDRESS_SYMBOLIC_TXT_REL'.
12128 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
12129 (microblaze_legitimate_pic_operand): Exclude function calls from
12130 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
12131 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
12133 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
12134 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
12135 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
12136 for 'address + offset'.
12137 (microblaze_expand_prologue): Add new function prologue call for
12139 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
12140 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
12141 table in case of TARGET_PIC_DATA_TEXT_REL.
12142 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
12143 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
12144 Add new macros 'UNSPEC_TEXT',
12145 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
12146 + exclude function calls from 'UNSPEC_PLT' in case of data text
12148 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
12149 new target hook for generating address diff vector tables in case of
12151 * doc/tm.texi : Regenerate.
12152 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
12153 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
12154 of addr diff vector generation.
12155 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
12156 target hook definition.
12157 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
12158 Add default function for generate_pic_addr_diff_vec -> flag_pic.
12159 * doc/invoke.texi (Add new pic option): Add new microblaze pic
12160 option for data text relative.
12162 2018-04-30 Richard Biener <rguenther@suse.de>
12164 * tree-chrec.h (evolution_function_is_constant_p): Remove
12166 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
12168 2018-04-30 Richard Biener <rguenther@suse.de>
12171 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
12173 2018-04-30 Richard Biener <rguenther@suse.de>
12175 PR tree-optimization/28364
12176 PR tree-optimization/85275
12177 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
12178 copying first exit test.
12180 2018-04-28 Mark Wielaard <mark@klomp.org>
12182 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
12183 dwarf_version >= 5.
12184 (dwarf_AT): Handle DW_AT_addr_base.
12185 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
12187 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
12190 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
12191 (*ashl<dwi>3_doubleword_mask_1): Ditto.
12192 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
12193 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
12195 2018-04-28 Richard Biener <rguenther@suse.de>
12197 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
12198 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
12199 to reflect use. Only add interesting stmts.
12201 2018-04-27 Martin Jambor <mjambor@suse.cz>
12204 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
12205 the jump function allows for passing through aggregate values.
12207 2018-04-27 David Malcolm <dmalcolm@redhat.com>
12209 * input.h (in_system_header_at): Convert from macro to inline
12211 (from_macro_expansion_at): Likewise.
12212 (from_macro_definition_at): Likewise.
12214 2018-04-27 Jeff Law <law@redhat.com>
12216 * config.gcc: Mark tile* targets as deprecated/obsolete.
12218 2018-04-27 Richard Biener <rguenther@suse.de>
12220 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
12223 2018-04-27 Richard Biener <rguenther@suse.de>
12225 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
12227 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
12229 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
12230 with Yd constraint. Set "preferred_for_speed" attribute from
12231 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
12232 with Yd constraint.
12233 (*movdi_internal): Ditto.
12234 (movti_interunit splitters): Remove
12235 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
12236 (movdi_interunit splitters): Ditto.
12237 * config/i386/constraints.md (Ye): Remove.
12238 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
12240 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12243 * config/aarch64/constraints.md (Usg): Limit to 31.
12244 (Usj): Limit to 63.
12246 2018-04-27 Jakub Jelinek <jakub@redhat.com>
12248 PR tree-optimization/85529
12249 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
12250 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
12251 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
12252 zero extension or masking of the MSB bit.
12253 (optimize_range_tests): Add FIRST_BB argument, pass it through
12254 to optimize_range_tests_var_bound.
12255 (maybe_optimize_range_tests, reassociate_bb): Adjust
12256 optimize_range_tests callers.
12258 2018-04-26 Richard Biener <rguenther@suse.de>
12259 Jakub Jelinek <jakub@redhat.com>
12261 * cgraph.h (symbol_table): Just declare debug method here.
12262 * symtab.c (symbol_table::debug): Define.
12264 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
12266 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
12268 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
12270 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
12271 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
12272 (*movdi_internal): Substitute Yi and Yj constraint with x
12273 and Ym and Yn constraint with y constraint. Update "isa"
12274 attribute and set "preferred_for_speed" attribute from
12275 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
12276 (*movsi_internal): Ditto.
12277 (*movdf_internal): Ditto.
12278 (*movsf_internal): Ditto.
12279 (*zero_extendsidi2): Ditto.
12280 * config/i386/sse.md (vec_set<mode>_0): Ditto.
12281 (sse2_loadld): Ditto.
12282 (*vec_extract<ssevecmodelower>_0): Ditto.
12283 (*vec_extractv4si_0_zext_sse4): Ditto.
12284 (vec_concatv2di): Ditto.
12285 (*vec_dup<mode>): Ditto.
12286 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
12287 * config/i386/constraints.md (Yi): Remove.
12292 2018-04-26 Nathan Sidwell <nathan@acm.org>
12294 * dumpfile.c (dump_open): New.
12295 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
12296 (dump_finish): Detect stdio/stderr by value not name.
12298 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
12300 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
12302 2018-04-26 Tom de Vries <tom@codesourcery.com>
12305 * config/nvptx/nvptx.c (verify_neutering_jumps)
12306 (verify_neutering_labels): New function
12307 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
12309 2018-04-26 Tom de Vries <tom@codesourcery.com>
12312 * config/nvptx/nvptx.c (needs_neutering_p): New function.
12313 (nvptx_single): Use needs_neutering_p to skip over insns that do not
12316 2018-04-26 Richard Biener <rguenther@suse.de>
12317 Tom de Vries <tom@codesourcery.com>
12320 * lto-streamer-out.c (output_function): Fixup loops if required to match
12321 discovery done in the reader.
12323 2018-04-26 Richard Biener <rguenther@suse.de>
12325 PR tree-optimization/85116
12326 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
12327 have a loop exit from the single latch predecessor. Remove
12328 case of header with just condition.
12329 (ch_base::copy_headers): Exclude infinite loops from any
12331 (pass_ch::execute): Record exits.
12333 2018-04-26 Richard Biener <rguenther@suse.de>
12335 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
12336 prologue cost vector and pass it to vect_get_load_cost.
12337 (vect_get_peeling_costs_all_drs): Likewise.
12338 (vect_peeling_hash_get_lowest_cost): Likewise.
12339 (vect_enhance_data_refs_alignment): Likewise.
12341 2018-04-26 Richard Biener <rguenther@suse.de>
12343 PR middle-end/85450
12344 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
12345 checking of integer<->pointer conversions.
12346 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
12347 sign-/zero-extending pointer types.
12348 (expand_omp_for_static_chunk): Likewise.
12350 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
12351 Jean Lee <xiaoyur347@gmail.com>
12353 * config/mips/mips.c (mips_asan_shadow_offset): New function.
12354 (TARGET_ASAN_SHADOW_OFFSET): Define.
12355 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
12356 true for -fsanitize=address.
12358 2018-04-25 Mark Wielaard <mark@klomp.org>
12360 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
12363 2018-04-25 Jakub Jelinek <jakub@redhat.com>
12365 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
12366 than "alu", remove explicit "memory" and "imm_disp" attributes.
12367 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
12369 PR middle-end/85414
12370 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
12371 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
12372 gen_lowpart_no_emit.
12374 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
12377 * config/i386/i386.c (ix86_expand_builtin): Change memory
12378 operand to XI, extend p0 to Pmode.
12379 * config/i386/i386.md: Change unspec volatile and operand
12380 1 mode to XI, change operand 0 mode to P.
12382 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
12384 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
12385 GET_MODE_MASK before any checking.
12386 (nds32_can_use_bset_p): Likewise.
12387 (nds32_can_use_btgl_p): Likewise.
12389 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
12391 * config/nds32/nds32-doubleword.md: New define_split pattern for
12392 illegal register number.
12394 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
12396 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
12398 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
12400 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
12402 2018-04-25 Richard Biener <rguenther@suse.de>
12404 * lto-streamer.h (LTO_major_version): Bump to 8.
12406 2018-04-25 Jakub Jelinek <jakub@redhat.com>
12408 * BASE-VER: Set to 9.0.0.
12410 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
12412 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
12413 in __abskf2 and __powikf2.
12415 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12418 * config/aarch64/constraints.md (Usg, Usj): New constraints.
12419 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
12420 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
12421 Use the above on operand 2. Reindent.
12422 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
12424 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
12427 * common/config/i386/i386-common.c (ix86_handle_option): Don't
12429 * config/i386/i386.opt (mcet): Removed.
12430 * doc/install.texi: Remove -mcet documentation.
12431 * doc/invoke.texi: Likewise.
12433 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
12436 * doc/install.texi: Remove -mcet from bootstrap-cet.
12438 2018-04-24 Jakub Jelinek <jakub@redhat.com>
12441 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
12442 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
12446 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
12447 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
12448 containing a CONST_VECTOR.
12450 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
12452 * doc/install.texi: Update newlib dependency for nvptx.
12454 2018-04-24 Jakub Jelinek <jakub@redhat.com>
12457 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
12458 instead of INTVAL when shifting x left.
12460 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
12462 PR tree-optimization/85478
12463 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
12464 vect_grouped_store_supported for single element vectors.
12466 2018-04-24 Richard Biener <rguenther@suse.de>
12469 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
12470 load cost increase to the case of non-constant step.
12472 2018-04-24 Jakub Jelinek <jakub@redhat.com>
12475 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
12476 destination if any_malformed_asm.
12478 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
12480 PR middle-end/85496
12481 * expr.c (store_field): In the bitfield case, if the value comes from
12482 a function call and is returned in registers by means of a PARALLEL,
12483 do not change the mode of the temporary unless BLKmode and VOIDmode.
12485 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
12487 PR rtl-optimization/85423
12488 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
12489 dependencies to debug insns when the previous insn is non-debug.
12491 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
12493 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
12494 enums into a single definition.
12495 (fls): Fix predicates and printing.
12498 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
12500 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
12501 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
12502 and short u6 immediate.
12503 (check_if_valid_sleep_operand): Remove.
12504 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
12506 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
12508 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
12509 flag_always_save_lp condition.
12510 * config/nds32/nds32.opt (malways-save-lp): New option.
12512 2018-04-22 Shiva Chen <shiva0217@gmail.com>
12514 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
12515 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
12516 * config/nds32/nds32.h
12517 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
12518 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
12520 2018-04-22 Shiva Chen <shiva0217@gmail.com>
12522 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
12523 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
12525 2018-04-22 Shiva Chen <shiva0217@gmail.com>
12526 Chung-Ju Wu <jasonwucj@gmail.com>
12528 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
12530 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
12531 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
12533 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
12535 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
12537 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
12539 * config/nds32/nds32-protos.h (nds32_data_alignment,
12540 nds32_local_alignment): Declare.
12541 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
12542 nds32_local_alignment): New functions.
12543 (TARGET_CONSTANT_ALIGNMENT): Define.
12544 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
12546 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
12548 * config/nds32/nds32.c
12549 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
12550 (TARGET_MODES_TIEABLE_P): Likewise.
12552 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
12554 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
12555 level Ofast and Og.
12557 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
12558 Chung-Ju Wu <jasonwucj@gmail.com>
12560 * config/nds32/constants.md (unspec_volatile_element): Add enum values
12561 for unaligned access.
12562 * config/nds32/nds32-intrinsic.c: Implementation of expanding
12564 * config/nds32/nds32-intrinsic.md: Likewise.
12565 * config/nds32/nds32_intrinsic.h: Likewise.
12566 * config/nds32/nds32.h (nds32_builtins): Likewise.
12567 * config/nds32/nds32.opt (munaligned-access): New option.
12568 * config/nds32/nds32.c (nds32_asm_file_start): Display
12569 flag_unaligned_access status.
12571 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
12573 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
12574 -mno-relax is present.
12575 * config/riscv/linux.h (LINK_SPEC): Ditto.
12577 2018-04-20 Martin Sebor <msebor@redhat.com>
12580 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
12582 (gimple_fold_builtin_stxcpy_chk): Same.
12583 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
12585 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
12588 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
12589 __powikf2 when long double is IEEE 128-bit.
12591 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
12593 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
12594 step to make sure stack always aligned.
12596 2018-04-20 Carl Love <cel@us.ibm.com>
12599 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
12600 size check for arg0.
12602 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
12603 Tom de Vries <tom@codesourcery.com>
12606 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
12607 Emit insns for calls too.
12608 (nvptx_find_par): Always look for worker-level predecessor insn.
12609 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
12611 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
12612 (nvptx_process_pars): Propagate frames for calls.
12614 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
12617 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
12619 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
12620 (ix86_handle_option): Don't handle OPT_mibt.
12621 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
12623 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
12625 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
12626 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
12627 (ix86_target_macros): Define __CET__ with flag_cf_protection
12628 for -fcf-protection.
12629 * config/i386/i386.c (isa2_opts): Remove -mibt.
12630 * config/i386/i386.h (TARGET_IBT): Removed.
12631 (TARGET_IBT_P): Likewise.
12632 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
12633 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
12634 * config/i386/i386.opt (mcet): Update help message.
12635 (mshstk): Likewise.
12637 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
12638 -mcet as an alias for -mshstk.
12640 2018-04-20 Richard Biener <rguenther@suse.de>
12642 PR middle-end/85475
12643 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
12644 complexity by forcing a single use of the multiply operand.
12646 2018-04-20 Martin Jambor <mjambor@suse.cz>
12649 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
12650 recursion dependency to only apply to non-clones.
12652 2018-04-20 Martin Jambor <mjambor@suse.cz>
12655 * ipa-cp.c (create_specialized_node): Check that clones of
12656 self-recursive edges exist during IPA-CP.
12658 2018-04-19 Toon Moene <toon@moene.org>
12660 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
12663 2018-04-19 Jakub Jelinek <jakub@redhat.com>
12665 PR tree-optimization/85467
12666 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
12667 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
12668 VECTOR_CST element to type.
12670 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
12673 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
12674 * config/i386/i386.md (builtin_setjmp_setup): Removed.
12675 (builtin_longjmp): Likewise.
12676 (save_stack_nonlocal): New pattern.
12677 (restore_stack_nonlocal): Likewise.
12679 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
12682 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
12683 Replace ASM_OUTPUT_LABEL with fprintf.
12685 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
12688 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
12689 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
12690 * config/i386/i386-c.c (ix86_target_macros_internal): Also
12691 define __IBT__ and __SHSTK__ for -fcf-protection.
12692 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
12694 (ix86_trampoline_init): Likewise.
12695 (x86_output_mi_thunk): Likewise.
12696 (ix86_notrack_prefixed_insn_p): Likewise.
12697 (ix86_option_override_internal): Don't disallow -fcf-protection.
12698 * config/i386/i386.md (rdssp<mode>): Also enable for
12700 (incssp<mode>): Likewise.
12701 (nop_endbr): Likewise.
12702 * config/i386/i386.opt (mcet): Change help message to built-in
12705 (mshstk): Likewise.
12706 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
12707 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
12708 enable CET built-in functions.
12710 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
12712 * common/config/i386/i386-common.c
12713 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
12714 OPTION_MASK_ISA_MOVDIRI_UNSET,
12715 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
12716 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
12717 * config.gcc (movdirintrin.h): New header.
12718 * config/i386/cpuid.h (bit_MOVDIRI,
12719 bit_MOVDIR64B): New bits.
12720 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
12722 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
12723 (VOID, PVOID, PCVOID)): New function types.
12724 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
12725 __builtin_ia32_directstoreu_u64,
12726 __builtin_ia32_movdir64b): New builtins.
12727 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
12728 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
12730 (ix86_valid_target_attribute_inner_p): Ditto.
12731 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
12732 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
12733 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
12734 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
12735 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
12736 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
12737 (movdiri<mode>, movdir64b_<mode>): New.
12738 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
12739 * config/i386/immintrin.h: Include movdirintrin.h.
12740 * config/i386/movdirintrin.h: New file.
12741 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
12743 2018-04-19 Richard Biener <rguenther@suse.de>
12745 PR middle-end/85455
12746 * cfg.c (clear_bb_flags): When loop state says we have
12747 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
12749 2018-04-19 Richard Biener <rguenther@suse.de>
12751 PR tree-optimization/84737
12752 * tree-vect-data-refs.c (vect_copy_ref_info): New function
12753 copying restrict info.
12754 (vect_setup_realignment): Use it.
12755 * tree-vectorizer.h (vect_copy_ref_info): Declare.
12756 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
12757 the first DR to all generated stores.
12758 (vectorizable_load): Likewise for loads.
12760 2018-04-19 Jakub Jelinek <jakub@redhat.com>
12762 PR tree-optimization/85446
12763 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
12764 the integral and pointer types to have the same precision.
12766 * doc/install.texi: Document --disable-cet being the default and
12769 2018-04-18 Martin Liska <mliska@suse.cz>
12771 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
12774 2018-04-18 Martin Liska <mliska@suse.cz>
12777 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
12780 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
12781 arguments if they are comparable.
12783 2018-04-18 Martin Liska <mliska@suse.cz>
12786 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
12789 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
12792 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
12795 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
12796 ENDBR after calling __morestack.
12798 2018-04-18 David Malcolm <dmalcolm@redhat.com>
12801 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
12802 by using gcc_base_ver to generate a gcc_driver_version, and use
12803 it when generating GCC_DRIVER_NAME.
12804 * configure: Regenerate.
12806 2018-04-18 Jakub Jelinek <jakub@redhat.com>
12809 * config.gcc: Obsolete powerpc*-*-*spe*.
12811 2018-04-17 Jakub Jelinek <jakub@redhat.com>
12814 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
12815 (stabstr_D): Change type of unum from unsigned int to
12816 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
12819 2018-04-17 Jim Wilson <jimw@sifive.com>
12822 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
12823 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
12824 Set arg_pointer_offset after using pretend_args_size.
12826 2018-04-17 Jakub Jelinek <jakub@redhat.com>
12828 PR rtl-optimization/85431
12829 * dse.c (record_store): Ignore zero width stores.
12832 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
12833 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
12834 __builtin_stack_restore rather than after it.
12835 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
12836 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
12837 argument instead of virtual_dynamic_stack_rtx.
12839 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
12841 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
12843 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12844 Add note to error message to explain internal mapping of overloaded
12845 built-in function name to non-overloaded built-in function name.
12846 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
12849 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
12852 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
12853 where the inputs overlap with the output.
12855 2018-04-17 Jakub Jelinek <jakub@redhat.com>
12857 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
12858 (=v, v) alternative and explicit "memory" attribute.
12859 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
12860 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
12862 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
12863 "sselog1" type instead of "sselog".
12864 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
12865 "sselog". Remove explicit "memory" attribute.
12866 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
12867 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
12869 (vec_extract_hi_v32hi): Merge all alternatives into one, use
12870 "sselog1" type instead of "sselog". Remove explicit "memory"
12872 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
12873 use "sselog1" type instead of "sselog". Remove explicit "memory"
12875 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
12876 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
12878 (vec_extract_hi_v64qi): Merge all alternatives into one, use
12879 "sselog1" type instead of "sselog". Remove explicit "memory"
12881 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
12882 use "sselog1" type instead of "sselog". Remove explicit "memory"
12886 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
12888 PR middle-end/85414
12889 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
12892 2018-04-17 Martin Jambor <mjambor@suse.cz>
12895 * ipa-cp.c (create_specialized_node): Call
12896 expand_all_artificial_thunks if necessary.
12898 2018-04-17 Martin Liska <mliska@suse.cz>
12901 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
12902 in message, remote space in between '_G' and '('.
12904 2018-04-17 Jakub Jelinek <jakub@redhat.com>
12907 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
12908 avx512f_vmcmp<mode>3<round_saeonly_name>,
12909 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
12910 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
12911 avx512f_rndscale<mode><round_saeonly_name>,
12912 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
12913 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
12914 Use %<iptr>2 instead of %2 for -masm=intel.
12915 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
12916 avx512f_vcvttss2usi<round_saeonly_name>,
12917 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
12919 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
12920 avx512f_vcvttsd2usi<round_saeonly_name>,
12921 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
12922 Use %q1 instead of %1 for -masm=intel.
12923 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
12924 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
12925 of %3 for -masm=intel.
12926 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
12928 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
12930 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
12932 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
12933 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
12935 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
12937 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
12938 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
12939 %g1 and one with %0 and %1.
12940 (avx512er_vmrcp28<mode><round_saeonly_name>,
12941 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
12942 %1 for -masm=intel.
12943 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
12944 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
12945 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
12946 of %0 and %{%4%} for -masm=intel.
12947 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
12948 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
12949 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
12950 order of %0 and %{%5%}%{z%} for -masm=intel.
12952 2018-04-17 Jan Hubicka <jh@suse.cz>
12955 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
12957 2018-04-17 Martin Liska <mliska@suse.cz>
12960 * multiple_target.c (create_dispatcher_calls): Set apostrophes
12961 for target_clone error message. Make default implementation
12962 clone to be a local declaration.
12963 (separate_attrs): Add new argument and check for an empty
12965 (expand_target_clones): Handle it.
12966 (ipa_target_clone): Make redirection just for target_clones
12969 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
12970 Tom de Vries <tom@codesourcery.com>
12972 PR middle-end/84955
12973 * omp-expand.c (expand_oacc_for): Add dummy false branch for
12974 tiled basic blocks without omp continue statements.
12976 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
12979 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
12980 vec_extract expression as having side effects to make sure it gets
12983 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
12986 * config/i386/i386.c (get_builtin_code_for_version): Check
12989 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
12992 * config.gcc: Support "skylake".
12993 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12995 * config/i386/i386.c (m_SKYLAKE): Define.
12996 (processor_target_table): Add "skylake".
12997 (ix86_option_override_internal): Add "skylake".
12998 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
12999 PROCESSOR_CANNONLAKE.
13000 (get_builtin_code_for_version): Fix priority for
13001 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
13002 PROCESSOR_SKYLAKE-AVX512.
13003 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
13004 (processor_type): Add PROCESSOR_SKYLAKE.
13006 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
13007 Jason Merrill <jason@redhat.com>
13010 * convert.c (convert_to_integer_1): Use direct recursion for
13011 enumeral types and types with a precision less than the number
13012 of bits in their mode.
13014 2018-04-16 Julia Koval <julia.koval@intel.com>
13017 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
13018 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
13020 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
13023 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
13024 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
13025 and -mno-direct-move.
13027 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
13030 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
13031 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
13032 Ensure negative shifts result in {0}.
13034 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
13036 PR rtl-optimization/79916
13037 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
13038 regs (if any) to define how to gnerate SD moves when LRA is in
13041 2018-04-13 Jakub Jelinek <jakub@redhat.com>
13043 PR rtl-optimization/85393
13044 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
13045 * except.c (expand_dw2_landing_pad_for_region): Make static.
13046 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
13047 a label and unconditional jump to old_bb, rather than
13048 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
13051 PR rtl-optimization/85376
13052 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
13053 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
13054 instead of a specific value.
13056 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
13057 Bin Cheng <bin.cheng@arm.com>
13059 PR tree-optimization/82965
13060 PR tree-optimization/83991
13061 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
13062 by_profile_only parameter.
13063 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
13064 information if the loop was predicted to iterate too many times.
13065 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
13067 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
13070 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
13073 2018-04-13 Martin Liska <mliska@suse.cz>
13074 Jakub Jelinek <jakub@redhat.com>
13076 PR middle-end/81657
13077 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
13078 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
13079 * builtins.c (expand_builtin_memory_copy_args): Use
13080 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
13081 handle dest_addr == pc_rtx.
13083 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
13086 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
13087 asked to not generate direct moves.
13088 (fix_trunc<mode>si2_stfiwx): Similar.
13089 (fix_trunc<mode>si2_internal): Similar.
13091 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13094 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
13095 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
13096 lookup if dest in some wider mode is known to be const0_rtx and
13097 if so, record permanent equivalence for it to be ZERO_EXTEND of
13098 the narrower mode destination.
13100 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
13102 * lto-streamer-out.c (output_function): Revert 259346.
13103 * omp-expand.c (expand_oacc_for): Likewise.
13105 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
13107 PR rtl-optimization/85354
13108 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
13109 * sel-sched.c (sel_global_init): ... here.
13111 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
13114 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
13115 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
13116 mode for PE-COFF targets.
13117 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
13118 (i386_pe_asm_lto_end): Likewise.
13119 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
13120 (TARGET_ASM_LTO_END): Likewise.
13121 * config/i386/winnt.c (saved_debug_info_level): New static variable.
13122 (i386_pe_asm_lto_start): New function.
13123 (i386_pe_asm_lto_end): Likewise.
13125 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
13126 Richard Biener <rguenther@suse.de>
13128 PR middle-end/84955
13129 * lto-streamer-out.c (output_function): Fix CFG loop state before
13131 * omp-expand.c (expand_oacc_for): Handle calls to internal
13132 functions like regular functions.
13134 2018-04-12 Richard Biener <rguenther@suse.de>
13137 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
13138 for the early LTO debug to properly generate references to it
13139 during DIE emission. Do not re-use that for the skeleton for
13141 (dwarf2out_early_finish): Likewise.
13143 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13146 * config/i386/sse.md
13147 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
13148 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
13149 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
13150 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
13151 and output is a reg, avoid creating invalid lowpart subreg, but
13152 instead split into a 512-bit move. Don't split if not AVX512VL,
13153 input is xmm16+ reg and output is a mem.
13154 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
13155 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
13156 xmm16+ reg and output is a mem.
13158 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13160 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
13161 also for flag_dwarf2_cfi_asm.
13163 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13165 PR rtl-optimization/85342
13166 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
13167 a bool scalar var inside of the loop instead. Don't try to update
13168 recog_data.operand after failed apply_change_group.
13170 2018-04-12 Tom de Vries <tom@codesourcery.com>
13173 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
13174 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
13175 array with flexible array member as array without given dimension.
13176 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
13177 argument for undefined param to true.
13179 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
13182 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
13183 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
13184 from PowerPC section.
13185 * config/rs6000/sysv4.opt (mcall-): Improve help text.
13186 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
13187 help text that is too long.
13188 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
13189 help text that is too long.
13190 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
13191 help text that is too long.
13193 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
13195 * config/alpha/alpha.md (stack_probe_internal): Rename
13196 from "probe_stack". Update all callers.
13198 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13200 PR rtl-optimization/84566
13201 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
13202 sched_macro_fuse_insns.
13204 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13207 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
13208 (compute_block_dependences): ... from here.
13210 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13212 PR tree-optimization/85331
13213 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
13214 from int to HOST_WIDE_INT.
13216 2018-04-11 Martin Jambor <mjambor@suse.cz>
13219 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
13220 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
13221 not the same as the source val.
13222 (cgraph_edge_brings_value_p): New parameter.
13223 (gather_edges_for_value): Pass destination value to
13224 cgraph_edge_brings_value_p.
13225 (perhaps_add_new_callers): Likewise.
13226 (get_info_about_necessary_edges): Likewise and exclude values brought
13227 only by self-recursive edges.
13228 (create_specialized_node): Redirect only clones of self-calling edges.
13229 (+self_recursive_pass_through_p): New function.
13230 (find_more_scalar_values_for_callers_subset): Use it.
13231 (find_aggregate_values_for_callers_subset): Likewise.
13232 (known_aggs_to_agg_replacement_list): Removed.
13233 (decide_whether_version_node): Re-calculate known constants for all
13234 remaining context clones.
13236 2018-04-11 Richard Biener <rguenther@suse.de>
13239 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
13240 from early DWARF output.
13241 (dwarf2out_early_finish): Output line info unconditionally into
13242 early DWARF and add reference to it.
13244 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13247 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
13248 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
13249 other than V2DFmode using iptr mode attribute.
13250 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
13252 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13254 PR rtl-optimization/84659
13255 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
13257 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13260 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
13262 (output_loc_list): Pass address of a dummy size variable even in the
13263 locview handling loop.
13264 (index_location_lists): Add comment on why skip_loc_list_entry can't
13267 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
13270 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
13273 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
13276 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
13277 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
13278 and -mstring-compare-inline-limit.
13280 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13283 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
13284 for stack clash protection in a register whenever we need it to be in
13287 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13289 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
13290 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
13292 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13295 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
13297 (mlong-double-): Ditto.
13298 * config/rs6000/sysv4.opt (msdata=): Ditto.
13299 (mtls-size=): Ditto.
13301 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13303 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
13304 erroneous entries for
13305 "vector int vec_ldl (int, long int *)", and
13306 "vector unsigned int vec_ldl (int, unsigned long int *)".
13307 Add comments and entries for
13308 "vector bool char vec_ldl (int, bool char *)",
13309 "vector bool short vec_ldl (int, bool short *)",
13310 "vector bool int vec_ldl (int, bool int *)",
13311 "vector bool long long vec_ldl (int, bool long long *)",
13312 "vector pixel vec_ldl (int, pixel *)",
13313 "vector long long vec_ldl (int, long long *)",
13314 "vector unsigned long long vec_ldl (int, unsigned long long *)".
13315 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
13316 type tree bool_long_long_type_node and correct definition of
13317 bool_V2DI_type_node to make reference to this new type tree.
13318 (rs6000_mangle_type): Replace erroneous reference to
13319 bool_long_type_node with bool_long_long_type_node.
13320 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
13321 comments to emphasize sign distinctions for char and int types and
13322 replace RS6000_BTI_bool_long constant with
13323 RS6000_BTI_bool_long_long constant. Also add comment to restrict
13324 use of RS6000_BTI_pixel.
13325 (bool_long_type_node): Remove this macro definition.
13326 (bool_long_long_type_node): New macro definition
13328 2018-04-10 Jakub Jelinek <jakub@redhat.com>
13330 PR rtl-optimization/85300
13331 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
13332 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
13333 simplify_unary_operation fails.
13335 2018-04-10 Martin Liska <mliska@suse.cz>
13337 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
13338 cgraph_edge and ipa_ref.
13340 2018-04-10 Jakub Jelinek <jakub@redhat.com>
13344 * config/i386/sse.md
13345 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
13346 computation of the VEC_MERGE selector from mask.
13347 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
13348 Fix decoding of the VEC_MERGE selector into mask.
13350 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
13352 PR tree-optimization/85286
13353 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
13355 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
13357 * final.c (final_1): Set insn_last_address as well as
13358 insn_current_address.
13360 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13363 * explow.c (emit_stack_probe): Call validize_mem on memory location
13364 before passing it to gen_probe_stack. Create address operand and
13365 legitimize it for the probe_stack_address case.
13367 2018-04-09 Jan Hubicka <jh@suse.cz>
13370 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
13371 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
13372 * tree.c (free_lang_data_in_type): Fix handling of binfos;
13374 (free_lang_data): Rebuild type inheritance graph.
13376 2018-04-09 Martin Sebor <msebor@redhat.com>
13378 * invoke.texi (-finline-small-functions): Mention other optimization
13380 (-findirect-inlining, -fpartial-inlining): Same.
13381 (-finline-functions-called-once): Same.
13382 (-freorder-blocks-and-partition): Same.
13384 2018-04-09 Jan Hubicka <jh@suse.cz>
13387 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
13388 jumps; choose last target that matches the criteria (i.e.
13389 no partition changes for non-crossing jumps).
13390 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
13391 support for redirecting crossing jumps to non-crossing.
13393 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
13395 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
13396 also for naked functions.
13398 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
13400 * config/arc/arc.md (add_shift): New pattern.
13401 (add_shift2): Likewise.
13402 (sub_shift): Likewise.
13403 (sub_shift_cmp0_noout): Likewise.
13404 (compare_si_ashiftsi): Likewise.
13405 (xbfu_cmp0_noout): New combine pattern.
13406 (xbfu_cmp0"): Likewise.
13407 (movsi_set_cc_insn): Place the predicable variant first.
13408 (commutative_binary_cmp0_noout): Remove clobber.
13409 (commutative_binary_cmp0): New pattern.
13410 (noncommutative_binary_cmp0): Likewise.
13411 (noncommutative_binary_cmp0_noout): Likewise.
13412 (noncommutative_binary_comparison_result_used): Removed.
13413 (rsub_cmp0): New pattern.
13414 (rsub_cmp0_noout): Likewise.
13415 (extzvsi): Changed, keep only meaningful variants.
13416 (SQH, SEZ): New iterators.
13417 (SQH_postfix): New mode attribute.
13418 (SEZ_prefix): New code attribute.
13419 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
13420 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
13421 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
13422 of numerical value.
13423 (noncommutative_operator): Check the availability of barrel
13426 2018-04-09 Richard Biener <rguenther@suse.de>
13428 PR tree-optimization/85284
13429 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
13430 Only use the niter constraining form of simple_iv when the exit
13431 is always executed.
13433 2018-04-09 Tom de Vries <tom@codesourcery.com>
13436 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13437 (define_expand "*memory_barrier"): New define_expand.
13438 (define_insn "memory_barrier"): New insn.
13440 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
13442 PR rtl-optimization/80463
13443 PR rtl-optimization/83972
13444 PR rtl-optimization/83480
13446 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
13447 correct producer for the insn.
13448 (tidy_control_flow): Fixup seqnos in case of debug insns.
13450 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
13452 PR rtl-optimization/83913
13454 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
13455 different sched-times when merging exprs.
13457 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
13459 PR rtl-optimization/83962
13461 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
13462 tidy_fallthru_edge and tidy_control_flow.
13464 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
13466 PR rtl-optimization/83530
13468 * sel-sched.c (force_next_insn): New global variable.
13469 (remove_insn_for_debug): When force_next_insn is true, also leave only
13470 next insn in the ready list.
13471 (sel_sched_region): When the region wasn't scheduled, make another pass
13472 over it with force_next_insn set to 1.
13474 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
13476 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
13478 * config/nds32/constants.md (unspec_volatile_element): Add enum values
13479 for interrupt control.
13480 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
13481 functions for interrupt control.
13482 * config/nds32/nds32-intrinsic.md: Likewise.
13483 * config/nds32/nds32_intrinsic.h: Likewise.
13484 * config/nds32/nds32.h (nds32_builtins): Likewise.
13486 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
13488 * config/nds32/nds32.c (nds32_init_machine_status,
13489 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
13490 strict_aligned_p field.
13491 (nds32_expand_to_rtl_hook): New function.
13492 (TARGET_EXPAND_TO_RTL_HOOK): Define.
13493 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
13495 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
13496 Chung-Ju Wu <jasonwucj@gmail.com>
13498 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
13499 * config/nds32/nds32-n7.md: New file.
13500 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
13501 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
13503 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
13504 * config/nds32/nds32.md (pipeline_model): Add n7.
13505 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
13506 * config/nds32/pipelines.md: Include n7 settings.
13508 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
13509 Chung-Ju Wu <jasonwucj@gmail.com>
13511 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
13512 * config/nds32/nds32-e8.md: New file.
13513 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
13514 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
13516 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
13517 * config/nds32/nds32.md (pipeline_model): Add e8.
13518 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
13519 * config/nds32/pipelines.md: Include e8 settings.
13521 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
13522 Chung-Ju Wu <jasonwucj@gmail.com>
13524 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
13525 * config/nds32/nds32-n8.md: New file.
13526 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
13527 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
13529 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
13530 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
13531 * config/nds32/nds32.md (pipeline_model): Add n8.
13532 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
13533 * config/nds32/pipelines.md: Include n8 settings.
13535 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
13536 Chung-Ju Wu <jasonwucj@gmail.com>
13538 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
13539 * config/nds32/nds32-n9-2r1w.md: New file.
13540 * config/nds32/nds32-n9-3r2w.md: New file.
13541 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
13542 nds32_register_ports): New or modify for cpu n9.
13543 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
13545 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
13546 * config/nds32/nds32-utils.c: New file.
13547 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
13548 TARGET_MUL_SLOW): Define.
13549 * config/nds32/nds32.md (pipeline_model): New attribute.
13550 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
13551 New options that support cpu n9.
13552 * config/nds32/pipelines.md: Include n9 settings.
13553 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
13555 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
13557 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
13558 information if necessary.
13559 (output_cond_branch_compare_zero): Likewise.
13560 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
13561 (nds32_target_alignment): Refine for alignment.
13562 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
13563 (FUNCTION_BOUNDARY): Modify.
13564 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
13566 * config/nds32/nds32.opt (malways-align, malign-functions): New.
13568 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
13570 * config/nds32/constants.md (unspec_volatile_element): Add values for
13571 TLB operation and data prefetch.
13572 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
13573 functions for TLB operation and data prefetch.
13574 * config/nds32/nds32-intrinsic.md: Likewise.
13575 * config/nds32/nds32_intrinsic.h: Likewise.
13576 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
13577 (nds32_print_operand): Likewise.
13578 * config/nds32/nds32.h (nds32_builtins): Likewise.
13580 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13581 Andrew Pinski <pinsika@gcc.gnu.org>
13583 PR middle-end/82976
13584 * match.pd: Use constant_boolean_node of correct type instead of
13585 boolean_true_node or boolean_false_node for simplifying
13586 pointer comparisons to zero.
13588 2018-04-07 Jakub Jelinek <jakub@redhat.com>
13590 PR tree-optimization/80021
13591 * tree.c (verify_type_variant): Make error call in verify_variant_match
13592 translatable and remove final full stop.
13594 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
13596 * config/nds32/constants.md (unspec_volatile_element): Add
13597 UNSPEC_VOLATILE_EH_RETURN.
13598 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
13599 nds32_output_stack_pop): Support dwarf exception handling process.
13600 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
13601 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
13602 exception handling process.
13603 (nds32_compute_stack_frame): Likewise.
13604 (nds32_return_addr_rtx): Likewise.
13605 (nds32_initial_elimination_offset): Likewise.
13606 (nds32_expand_prologue): Likewise.
13607 (nds32_expand_epilogue): Likewise.
13608 (nds32_dynamic_chain_address): New function.
13609 * config/nds32/nds32.h (machine_function): Add fields for dwarf
13610 exception handling.
13611 (DYNAMIC_CHAIN_ADDRESS): Define.
13612 (EH_RETURN_DATA_REGNO): Define.
13613 (EH_RETURN_STACKADJ_RTX): Define.
13614 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
13615 patterns for dwarf exception handling.
13617 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
13619 * config/nds32/nds32.h: Clean up obsolete macros.
13621 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
13623 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13624 Add enum values for particular instructions.
13625 * config/nds32/nds32-intrinsic.c: Implementation of expanding
13626 particular intrinsic functions.
13627 * config/nds32/nds32-intrinsic.md: Likewise.
13628 * config/nds32/nds32_intrinsic.h: Likewise.
13629 * config/nds32/nds32.h (nds32_builtins): Likewise.
13630 * config/nds32/nds32.md (type): Add pbsad and pbsada.
13631 (btst, ave): New patterns for particular instructions.
13633 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
13635 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13636 Add enum values for atomic load/store and memory sync.
13637 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
13639 * config/nds32/nds32-intrinsic.md: Likewise.
13640 * config/nds32/nds32_intrinsic.h: Likewise.
13641 * config/nds32/nds32.h (nds32_builtins): Likewise.
13643 2018-04-07 Jakub Jelinek <jakub@redhat.com>
13645 PR tree-optimization/85257
13646 * fold-const.c (native_encode_vector): If not all elts could fit
13647 and off is -1, return 0 rather than offset.
13648 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
13649 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
13650 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
13651 adjust buffer in native_interpret_expr call.
13653 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
13655 * config/nds32/constants.md (unspec_volatile_element): Add cache
13656 control enum values.
13657 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
13658 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
13659 * config/nds32/nds32.c (nds32_cctl_names): New.
13660 (nds32_print_operand): Handle cache control register names.
13661 * config/nds32/nds32.h (nds32_builtins): New enum values.
13662 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
13664 * config/nds32/nds32.md (type): Add mmu.
13665 * config/nds32/pipelines.md (simple_insn): Add mmu.
13667 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
13669 * config/nds32/nds32.md (type): Remove call.
13670 * config/nds32/pipelines.md (simple_insn): Likewise.
13672 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
13674 * config/nds32/constants.md (unspec_volatile_element): Add
13675 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
13676 UNSPEC_VOLATILE_FMFCFG.
13677 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
13678 description for fmfcfg and fmfcsr.
13679 (bdesc_1arg): Add fmtcsr.
13680 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
13681 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
13682 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
13683 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
13684 unspec_fmfcfg): New patterns.
13685 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
13686 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
13687 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
13688 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
13689 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
13690 __nds32__fmfcfg): Define.
13692 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
13694 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
13695 intrinsic register names.
13696 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
13697 intrinsic register enum values and macros.
13699 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
13701 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
13702 for load/store addressing form.
13703 (nds32_print_operand_address): Likewise.
13705 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
13708 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
13709 based on LABEL_REF. Remove useless assertion.
13710 (pic_address_needs_scratch): Fix formatting.
13711 (sparc_legitimize_pic_address): Minor tweaks.
13712 (sparc_delegitimize_address): Adjust assertion accordingly.
13713 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
13714 into symbolic_operand.
13715 (movsi_high_pic_label_ref): Likewise.
13716 (movsi_lo_sum_pic_label_ref): Likewise.
13717 (movdi_pic_label_ref): Likewise.
13718 (movdi_high_pic_label_ref): Likewise.
13719 (movdi_lo_sum_pic_label_ref): Likewise.
13721 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
13723 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
13724 custom LIB_SPEC setup.
13726 2018-04-06 Ruslan Bukin <br@bsdpad.com>
13727 Kito Cheng <kito.cheng@gmail.com>
13729 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
13730 * config/riscv/freebsd.h: New.
13732 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
13734 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
13735 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
13738 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
13739 Kito Cheng <kito.cheng@gmail.com>
13741 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
13742 nds32_output_call, nds32_symbol_binds_local_p): New functions.
13743 * config/nds32/nds32-protos.h (nds32_output_call,
13744 nds32_output_return): Declare.
13745 * config/nds32/nds32.md: Refine all the call and return patterns.
13747 2018-04-06 Jakub Jelinek <jakub@redhat.com>
13750 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
13751 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
13753 PR rtl-optimization/84872
13754 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
13755 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
13756 EDGE_CROSSING edge.
13758 2018-04-06 Tamar Christina <tamar.christina@arm.com>
13760 * expr.c (copy_blkmode_to_reg): Revert 254862.
13761 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
13763 2018-04-06 Richard Biener <rguenther@suse.de>
13765 PR middle-end/85244
13766 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
13767 after seeing a component reference with an adjacent field. Treat
13768 refs to arrays at struct end of external decls similar to
13769 refs to unconstrained commons.
13771 2018-04-06 Jakub Jelinek <jakub@redhat.com>
13774 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
13775 look through SAVE_EXPRs with non-side-effects argument. Adjust
13777 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
13780 2018-04-06 Richard Biener <rguenther@suse.de>
13782 PR middle-end/85180
13783 * alias.c (find_base_term): New wrapper around find_base_term
13784 unwinding CSELIB_VAL_PTR changes.
13785 (find_base_term): Do not restore CSELIB_VAL_PTR during the
13788 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13790 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
13792 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
13793 constant definitions.
13794 ("nop"): lr 0,0 -> nopr r0
13795 ("nop_lr0", "nop_lr1"): New insn definitions.
13797 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
13799 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
13800 NDS32_V3PUSH_AVAILABLE_P macro.
13802 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
13803 Chung-Ju Wu <jasonwucj@gmail.com>
13805 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
13806 (nds32*-*-*): Add float and fpu_config into supported_defaults.
13807 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
13808 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
13809 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
13810 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
13811 * config/nds32/constraints.md: New constraints and checking for hard
13812 float configuration.
13813 * config/nds32/iterators.md: New mode iterator and attribute for hard
13814 float configuration.
13815 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
13817 * config/nds32/nds32-fpu.md: New file.
13818 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
13819 deal with hard float code generation.
13820 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
13822 (abi_type, float_reg_number): New enum type.
13823 * config/nds32/nds32-predicates.c: New predicates for hard float.
13824 * config/nds32/nds32-protos.h: Declare functions for hard float.
13825 * config/nds32/nds32.c: Implementation for hard float configuration.
13826 * config/nds32/nds32.h: Definitions for hard float configuration.
13827 * config/nds32/nds32.md: Include hard float machine description and
13828 modify patterns for hard float configuration.
13829 * config/nds32/nds32.opt: New options for hard float configuration.
13830 * config/nds32/predicates.md: New predicates for hard float
13833 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
13835 * common/config/nds32/nds32-common.c
13836 (nds32_option_optimization_table): Enable -mreleax-hint by default.
13838 2018-04-05 Jakub Jelinek <jakub@redhat.com>
13840 PR middle-end/85195
13841 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
13842 CONSTRUCTOR_ELT (ctor, ...)->value.
13844 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
13847 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
13849 2018-04-05 Tom de Vries <tom@codesourcery.com>
13852 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
13855 2018-04-05 Shiva Chen <shiva0217@gmail.com>
13856 Kito Cheng <kito.cheng@gmail.com>
13858 * config/nds32/constraints.md (U33): Fine-tune checking condition.
13859 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
13860 * config/nds32/nds32.h (nds32_16bit_address_type): Add
13861 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
13863 2018-04-05 Shiva Chen <shiva0217@gmail.com>
13864 Kito Cheng <kito.cheng@gmail.com>
13866 * config/nds32/constraints.md (Ufe): New memory constraint.
13867 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
13868 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
13869 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
13871 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
13872 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
13874 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13876 * config/nds32/nds32.md: Use optimize_size in the condition for
13877 alu-shift instructions.
13879 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13881 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
13883 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13885 * config/nds32/nds32.md (negsi2): Refine pattern.
13887 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
13888 Chung-Ju Wu <jasonwucj@gmail.com>
13890 * config/nds32/iterators.md (shift_rotate): New code iterator.
13891 (shift): New code attribute.
13892 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
13893 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
13894 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
13895 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
13896 bit-wise operations.
13897 (andsi3, *andsi3): Ditto.
13898 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
13899 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
13900 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
13901 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
13902 nds32_ior_operand, nds32_xor_operand): New predicates.
13904 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13906 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
13907 (addsi3, subsi3): ... this.
13909 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13911 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
13913 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13915 * config/nds32/nds32.md: Adjust indention.
13917 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
13919 * config/nds32/nds32.md (feature): New attribute.
13921 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
13923 * config/nds32/nds32.md (subtype): New attribute.
13925 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13928 * config/arm/arm-builtins.c (arm_expand_builtin): Change
13929 expansion to perform a bitwise AND of the argument followed by a
13930 boolean negation of the result.
13932 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
13934 PR rtl-optimization/84878
13935 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
13936 the basic block. Assert the use reference is not artificial and that
13937 it has an associated insn.
13939 2018-04-04 Michael Matz <matz@suse.de>
13941 * builtins.c (compute_objsize): Pass correct operand
13942 to array_at_struct_end_p.
13944 2018-04-04 Richard Biener <rguenther@suse.de>
13947 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
13948 from contexts for DINFO_LEVEL_TERSE and below.
13950 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
13952 * config/nds32/nds32-doubleword.md (move_<mode>): Require
13953 resiter_operand condition.
13954 * config/nds32/nds32.md (*move<mode>): Ditto.
13956 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
13957 Monk Chiang <sh.chiang04@gmail.com>
13959 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
13961 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
13963 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
13965 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
13966 Kito Cheng <kito.cheng@gmail.com>
13968 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
13969 nds32_cond_code_str, output_cond_branch,
13970 output_cond_branch_compare_zero, nds32_expand_cbranch,
13971 nds32_expand_cstore, nds32_expand_movcc,
13972 nds32_output_cbranchsi4_equality_zero,
13973 nds32_output_cbranchsi4_equality_reg,
13974 nds32_output_cbranchsi4_equality_reg_or_const_int,
13975 nds32_output_cbranchsi4_greater_less_zero: New functions.
13976 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
13977 nds32_expand_cstore, nds32_expand_movcc,
13978 nds32_output_cbranchsi4_equality_zero,
13979 nds32_output_cbranchsi4_equality_reg,
13980 nds32_output_cbranchsi4_equality_reg_or_const_int,
13981 nds32_output_cbranchsi4_greater_less_zero): Declare.
13982 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
13983 nds32_rimm11s_operand): New predicates.
13984 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
13985 * config/nds32/nds32.md: Rewrite all the branch and conditional move
13988 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
13990 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
13991 * config/nds32/nds32.md: Ditto.
13992 * config/nds32/pipelines.md: Ditto.
13994 2018-04-04 Richard Biener <rguenther@suse.de>
13996 PR tree-optimization/85168
13997 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
13998 propagating abnormals.
14000 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14002 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
14004 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14005 Kito Cheng <kito.cheng@gmail.com>
14007 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
14008 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
14009 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
14010 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
14011 * config/nds32/nds32.md (sibcall_internal): New.
14012 (sibcall_register): Remove.
14013 (sibcall_immediate): Remove.
14014 (sibcall_value_internal): New.
14015 (sibcall_value_register): Remove.
14016 (sibcall_value_immediate): Remove.
14017 * config/nds32/predicates.md (nds32_general_register_operand): New.
14018 (nds32_call_address_operand): New.
14020 2018-04-03 Jakub Jelinek <jakub@redhat.com>
14022 PR rtl-optimization/85167
14023 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
14024 bb_defs if *split_p, instead preinitialize it to NULL.
14026 PR tree-optimization/85156
14027 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
14028 evaluating the argument multiple times.
14030 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
14032 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
14034 (_mm_cvtpd_ps): Likewise.
14035 (_mm_cvttpd_epi32): Likewise.
14036 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
14037 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
14038 vector, pixel, and bool following altivec.h include.
14040 2018-04-03 Martin Sebor <msebor@redhat.com>
14042 * doc/extend.texi (Common Function Attributes): Clarify.
14043 (const attribute): Likewise.
14044 (pure attribute): Likewise.
14046 2018-04-03 Jakub Jelinek <jakub@redhat.com>
14049 * config/i386/i386.c (ix86_expand_vector_set): Use
14050 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
14052 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
14054 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
14055 instructions when changing rounding bits to preserve precision bits
14056 in the x87 control word.
14058 2018-04-03 Martin Liska <mliska@suse.cz>
14060 PR tree-optimization/82491
14061 * rtl.h (strip_offset_and_add): Replace += suboffset with
14062 poly_uint64 () + suboffset.
14064 2018-03-29 Martin Liska <mliska@suse.cz>
14065 Martin Jambor <mjambor@suse.cz>
14068 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
14069 param_type is not an integral or pointer type.
14071 2018-04-03 Richard Biener <rguenther@suse.de>
14073 * sese.h (recompute_all_dominators): Remove.
14075 2018-04-02 Martin Sebor <msebor@redhat.com>
14077 * doc/invoke.texi (-Wrestrict): Fix typos.
14079 2018-04-02 Jim Wilson <jimw@sifive.com>
14081 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
14082 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
14083 (<optab>di3, <optab>si3_extend): Likewise.
14084 (<optab>si3_mask, <optab>si3_mask_1): New.
14085 (<optab>di3_mask, <optab>di3_mask_1): New.
14086 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
14087 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
14088 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
14090 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
14092 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
14095 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
14097 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
14098 (nds32_canonicalize_comparison): New function.
14100 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14101 Kito Cheng <kito.cheng@gmail.com>
14102 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14104 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
14105 * config/nds32/constants.md (unspec_volatile_element): Add
14106 UNSPEC_VOLATILE_RELAX_GROUP.
14107 * config/nds32/nds32-relax-opt.c: New file.
14108 * config/nds32/nds32-predicates.c
14109 (nds32_symbol_load_store_p): New function.
14110 * config/nds32/nds32-protos.h
14111 (nds32_symbol_load_store_p): Declare function.
14112 (make_pass_nds32_relax_opt): Declare new rtl pass function.
14113 * config/nds32/nds32.c
14114 (nds32_register_pass): New function to register pass.
14115 (nds32_register_passes): New function to register passes.
14116 * config/nds32/nds32.md (relax_group): New pattern.
14117 * config/nds32/nds32.opt (mrelax-hint): New option.
14118 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
14120 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
14122 * config/nds32/t-nds32: Modify files dependency.
14124 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14126 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
14127 (PROFILE_HOOK): Define its implementation.
14129 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14131 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
14132 type and 32-bit size.
14134 2018-04-01 Jakub Jelinek <jakub@redhat.com>
14136 PR middle-end/85090
14137 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
14138 (V_128_256): New mode iterator.
14139 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
14140 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
14141 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
14143 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
14144 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
14146 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
14149 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
14150 NaN inputs correctly.
14152 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
14155 * config/rs6000/vsx.md (??r): New mode attribute.
14156 (*vsx_mov<mode>_64bit): Use it.
14157 (*vsx_mov<mode>_32bit): Likewise.
14159 2018-03-30 Martin Sebor <msebor@redhat.com>
14161 PR tree-optimization/84818
14162 * builtins.c (check_access): Use warning_n.
14164 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14167 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
14169 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
14172 2018-03-30 Julia Koval <julia.koval@intel.com>
14175 * x86-tune.def (movx, partial_reg_dependency): Enable for
14178 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
14180 PR inline-asm/84985
14181 * lra-constraints.c (process_alt_operands): Move setting
14182 this_alternative_matches below.
14184 2018-03-29 Martin Liska <mliska@suse.cz>
14187 * doc/invoke.texi: Document how LTO works with debug info.
14188 Describe auto-load support of binutils. Mention 'x86-64'
14189 as valid option value of -march option.
14191 2018-03-29 Jakub Jelinek <jakub@redhat.com>
14193 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
14196 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
14197 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
14198 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
14201 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
14204 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
14205 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
14206 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
14207 for RS6000_BTM_POWERPC64.
14208 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
14209 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
14210 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
14213 (DIVDEU): Likewise.
14215 2018-03-28 Carl Love <cel@us.ibm.com>
14218 2017-09-27 Carl Love <cel@us.ibm.com>
14220 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14221 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14222 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14225 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14227 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
14228 instead of __vector bool.
14229 (_mm_max_pu8): Likewise.
14230 (_mm_min_pi16): Likewise.
14232 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
14235 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
14236 (DIVWEUO): Likewise.
14237 (DIVDEO): Likewise.
14238 (DIVDEUO): Likewise.
14239 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
14240 DIVWEUO and DIVDEUO.
14241 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
14242 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
14243 (div_extend): Likewise.
14244 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
14246 (__builtin_divweuo): Likewise.
14247 (__builtin_divdeo): Likewise.
14248 (__builtin_divdeuo): Likewise.
14250 2018-03-28 Jakub Jelinek <jakub@redhat.com>
14253 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
14254 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
14256 PR tree-optimization/82004
14257 * gimple-match-head.c (optimize_pow_to_exp): New function.
14258 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
14259 Don't fold to exp if optimize_pow_to_exp is false.
14261 2018-03-28 Martin Liska <mliska@suse.cz>
14264 * calls.c (initialize_argument_information): Fix trailing space.
14265 * common.opt: Fix typo and provide better explanation for
14266 -fsanitize-coverage option.
14267 * config/i386/i386.opt: Fix typo.
14269 2018-03-28 Jakub Jelinek <jakub@redhat.com>
14270 Martin Liska <mliska@suse.cz>
14273 * gimplify.c (asan_poison_variable): Don't do the check for
14274 gimplify_omp_ctxp here.
14275 (gimplify_decl_expr): Do it here.
14276 (gimplify_target_expr): Likewise.
14278 2018-03-28 Martin Liska <mliska@suse.cz>
14281 * config/i386/i386.c (ix86_function_arg_advance): Do not call
14282 chkp_type_bounds_count if MPX is not enabled.
14284 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
14286 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
14288 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
14291 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
14292 function to create the function decl for complex long double
14293 multiply and divide for -mabi=ieeelongdouble.
14294 (init_float128_ieee): Call it.
14296 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
14299 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
14300 -fcf-protection=branch -mibt.
14301 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
14303 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14306 * config/arm/arm.c (arm_valid_symbolic_address): Handle
14307 arm_word_relocations.
14309 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
14312 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
14313 extern array declarations.
14315 2018-03-27 Richard Biener <rguenther@suse.de>
14317 PR middle-end/84067
14318 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
14319 explicit single_use checks.
14321 2018-03-27 Richard Biener <rguenther@suse.de>
14323 PR tree-optimization/85082
14324 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
14327 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14329 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
14330 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
14331 Turn on fasynchronous-unwind-tables and funwind-tables.
14333 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
14336 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
14337 (*bmi_blsr_<mode>_ccz): Ditto.
14339 2018-03-26 Tom de Vries <tom@codesourcery.com>
14341 PR tree-optimization/85063
14342 * omp-general.c (offloading_function_p): New function. Factor out
14344 * omp-offload.c (pass_omp_target_link::gate): ... here.
14345 * omp-general.h (offloading_function_p): Declare.
14346 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
14347 with attribute omp declare target for offloading functions.
14349 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
14351 PR tree-optimization/84005
14352 * tree-data-ref.h (get_base_for_alignment): Declare.
14353 * tree-data-ref.c (get_base_for_alignment_1): New function.
14354 (get_base_for_alignment): Likewise.
14355 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14356 get_base_for_alignment to find a suitable base object, instead
14357 of always using drb->base_address.
14359 2018-03-23 Jakub Jelinek <jakub@redhat.com>
14361 PR inline-asm/85022
14362 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
14363 known size by default.
14365 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
14367 PR inline-asm/85030
14368 * lra-constraints.c (process_alt_operands): Don't match BLKmode
14369 and non BLKmode operands.
14371 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14374 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
14375 Clean up attributes.
14377 2018-03-23 Richard Biener <rguenther@suse.de>
14380 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
14381 we are going to emit early debug for LTO.
14383 2018-03-23 Jakub Jelinek <jakub@redhat.com>
14385 PR inline-asm/85034
14386 * function.c (match_asm_constraints_1): Don't optimize if input
14387 doesn't satisfy general_operand predicate for output's mode.
14389 PR inline-asm/85022
14390 * alias.c (write_dependence_p): Don't require for x_canonicalized
14391 non-VOIDmode if x has VOIDmode.
14394 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
14395 just don't try to optimize it rather than assert it never happens.
14397 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
14399 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
14400 macro expansions for definition of ST_INTERNAL_<mode> and
14401 LD_INTERNAL_<mode> builtins.
14402 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
14404 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
14406 (altivec_expand_st_builtin): Likewise.
14407 (altivec_expand_builtin): Remove calls to deleted functions.
14408 (rs6000_address_for_altivec): Delete this function.
14409 * config/rs6000/vector.md: Remove expands for
14410 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
14412 2018-03-22 Sudakshina Das <sudi.das@arm.com>
14415 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
14416 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
14417 re-computing once computed.
14418 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
14419 (arm_init_machine_status): Initialize
14420 machine->static_chain_stack_bytes.
14422 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
14425 * doc/extend.texi: Add four new prototypes for vec_ld.
14426 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
14427 definitions for more logical presentation.
14428 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
14429 entries for V1TI variants of __builtin_altivec_ld builtin.
14430 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
14431 handling of V1TI variant of LVX icode pattern.
14432 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
14433 (rs6000_gimple_fold_builtin): Likewise.
14434 (altivec_init_builtins): Add code to define
14435 __builtin_altivec_lvx_v1ti function.
14437 2018-03-22 Jakub Jelinek <jakub@redhat.com>
14439 PR inline-asm/84941
14440 * function.c (match_asm_constraints_1): Don't do the optimization
14441 if input isn't a REG, SUBREG, MEM or constant.
14443 2018-03-22 Tom de Vries <tom@codesourcery.com>
14445 PR tree-optimization/84956
14446 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
14447 bb_has_abnormal_pred.
14449 2018-03-22 Jakub Jelinek <jakub@redhat.com>
14452 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
14453 DECL_INITIAL (decl) to decl at the end.
14454 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
14455 adjust the comment.
14457 2018-03-21 Joseph Myers <joseph@codesourcery.com>
14459 * doc/extend.texi (__builtin_tgmath): Document when complex
14460 integer types are treated as _Complex _Float64.
14462 2018-03-21 Tom de Vries <tom@codesourcery.com>
14464 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
14466 2018-03-21 Jakub Jelinek <jakub@redhat.com>
14468 PR tree-optimization/84960
14469 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
14470 if it is ENTRY block, move them into single succ of ENTRY in that case.
14472 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
14474 PR tree-optimization/84811
14475 * poly-int.h (poly_span_traits): Remove the T3 parameter and
14476 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
14477 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
14478 (known_subrange_p): Update accordingly. Cast each value involved
14479 in the size comparison, rather than casting the result of the
14482 2018-03-21 Jakub Jelinek <jakub@redhat.com>
14484 PR tree-optimization/84982
14485 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
14486 by flipping the least significant bit rather than all bits from
14487 bitpos to bitpos + bitsize - 1.
14489 2018-03-21 Nathan Sidwell <nathan@acm.org>
14491 * doc/extend.texi (Deprecated Features): Remove mention of
14492 long-deleted deprecations.
14494 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14497 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
14498 * configure: Regenerate.
14500 2018-03-21 Tom de Vries <tom@codesourcery.com>
14502 PR tree-optimization/83126
14503 * tree-parloops.c (num_phis): New function.
14504 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
14506 2018-03-21 Nathan Sidwell <nathan@acm.org>
14508 * doc/extend.texi (Deprecated Features): Update deprecated flags,
14509 mention anon-struct/union members and trailing attributes.
14511 2018-03-21 Bin Cheng <bin.cheng@arm.com>
14513 PR tree-optimization/84969
14514 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
14515 builtin memset partitions if they set different rhs values.
14517 2018-03-21 Jakub Jelinek <jakub@redhat.com>
14519 PR rtl-optimization/84989
14520 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
14521 VEC_DUPLICATE with scalar result mode.
14523 2018-03-21 Martin Liska <mliska@suse.cz>
14526 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
14527 not intended return statement.
14529 2018-03-21 Martin Liska <mliska@suse.cz>
14532 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
14533 (chkp_find_bound_slots_1): Limit number of iterations.
14535 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
14538 * Minor grammar fixes for x86 options.
14540 2018-03-20 Jakub Jelinek <jakub@redhat.com>
14543 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
14544 holding REG_CFA_RESTORE notes, instead turn them into a USE.
14546 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
14549 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
14550 (altivec_lvx_<mode>_1op): Likewise.
14551 (altivec_stvx_<mode>_2op): Likewise.
14552 (altivec_stvx_<mode>_1op): Likewise.
14553 (altivec_lvx_<VM2:mode>): New define_expand.
14554 (altivec_stvx_<VM2:mode>): Likewise.
14555 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
14556 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
14557 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
14558 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
14559 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
14560 (rs6000_gen_lvx): Likewise.
14561 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
14562 (altivec_expand_stv_builtin): Likewise.
14563 (altivec_expand_builtin): Likewise.
14564 * config/rs6000/vector.md: Likewise.
14566 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14569 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
14572 2018-03-20 Richard Biener <rguenther@suse.de>
14575 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
14576 sign-conversions as zero, fall back to standard scalar_stmt
14579 2018-03-20 Martin Liska <mliska@suse.cz>
14582 * predict.c (rebuild_frequencies): Handle case when we have
14583 PROFILE_ABSENT, but flag_guess_branch_prob is false.
14585 2018-03-20 Jakub Jelinek <jakub@redhat.com>
14588 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
14589 flag_section_anchors.
14590 * varasm.c (use_blocks_for_decl_p): Remove hack for
14591 dw2_force_const_mem.
14594 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
14596 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
14597 be created, use lowpart_subreg of operands[0] rather than operands[0]
14599 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
14600 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
14601 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
14602 and n constraint instead of aarch64_shift_imm_di and Usd.
14603 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
14604 (*aarch64_<optab>_reg_minus<mode>3): ... this.
14606 2018-03-20 Sudakshina Das <sudi.das@arm.com>
14609 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
14610 to favor GPR over NEON registers.
14611 (<shift>di3_neon): Likewise.
14613 2018-03-20 Tom de Vries <tom@codesourcery.com>
14616 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
14617 (nvptx_process_pars): Emit bar.sync asap and alap.
14619 2018-03-20 Tom de Vries <tom@codesourcery.com>
14622 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
14623 seen_label if seen_label is already set.
14625 2018-03-20 Jakub Jelinek <jakub@redhat.com>
14628 * config/i386/i386.c (fold_builtin_cpu): For features above 31
14629 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
14630 Use 1U instead of 1. Formatting fixes.
14633 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
14634 instead of TREE_TYPE (s1) for the return value.
14636 2018-03-19 Jakub Jelinek <jakub@redhat.com>
14638 PR tree-optimization/84946
14639 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
14640 bitsize + bitsize in poly_uint64 rather than poly_int64.
14643 * dwarf2asm.c: Include fold-const.c.
14644 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
14645 of decl rather than decl itself.
14647 PR rtl-optimization/84643
14648 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
14650 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
14653 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
14654 calling assemble_variable.
14656 2018-03-19 Sudakshina Das <sudi.das@arm.com>
14659 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
14660 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
14662 2018-03-19 Jim Wilson <jimw@sifive.com>
14665 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
14666 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
14667 (riscv_first_stack_step): Likewise.
14668 (riscv_option_override): Use STACK_BOUNDARY instead of
14669 MIN_STACK_BOUNDARY.
14670 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
14671 MIN_STACK_BOUNDARY.
14672 (BIGGEST_ALIGNMENT): Set to 128.
14673 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
14674 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
14677 2018-03-19 Richard Biener <rguenther@suse.de>
14679 PR tree-optimization/84933
14680 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
14681 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
14683 2018-03-19 Richard Biener <rguenther@suse.de>
14685 PR tree-optimization/84859
14686 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
14687 (cond_if_else_store_replacement): Perform sinking operation on
14688 single-store BBs regardless of MAX_STORES_TO_SINK setting.
14689 Generalize what a BB with a single eligible store is.
14691 2018-03-19 Richard Biener <rguenther@suse.de>
14693 PR tree-optimization/84929
14694 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
14695 chrec_is_positive against non-chrec arg.
14697 2018-03-19 Tamar Christina <tamar.christina@arm.com>
14700 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
14702 2018-03-18 Martin Liska <mliska@suse.cz>
14704 PR rtl-optimization/84635
14705 * regrename.c (build_def_use): Use matches_mode only when
14708 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
14710 PR tree-optimization/84913
14711 * tree-vect-loop.c (vectorizable_reduction): Don't try to
14712 vectorize chains of COND_EXPRs.
14714 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
14716 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
14718 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
14720 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
14722 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
14724 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
14726 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
14727 Kito Cheng <kito.cheng@gmail.com>
14729 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
14730 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
14731 (nds32_adjust_reg_alloc_order): New function.
14732 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
14734 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
14736 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
14737 nds32_print_operand, nds32_print_operand_address): Use
14738 HOST_WIDE_INT_PRINT_DEC instead.
14740 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
14742 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
14744 2018-03-17 Jakub Jelinek <jakub@redhat.com>
14747 * config/i386/i386.c (initial_ix86_tune_features,
14748 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
14749 unsigned long long.
14750 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
14751 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
14752 rather than 1u << ix86_tune. Formatting fix.
14753 (ix86_option_override_internal): Change ix86_arch_mask from
14754 unsigned int to unsigned HOST_WIDE_INT, initialize to
14755 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
14756 (ix86_function_specific_restore): Likewise.
14758 2018-03-16 Jakub Jelinek <jakub@redhat.com>
14761 * postreload.c (reload_combine_recognize_pattern): Perform
14762 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
14763 truncate_int_for_mode the result for the destination's mode.
14766 * hsa-gen.c (mem_type_for_type): Fix comment typo.
14767 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
14769 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
14772 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
14775 * lra-assigns.c (lra_split_hard_reg_for): Don't use
14776 regno_allocno_class_array and sorted_pseudos.
14777 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
14778 insns where regno is used.
14780 2018-03-16 Martin Liska <mliska@suse.cz>
14783 * multiple_target.c (create_dispatcher_calls): Redirect
14784 reference in the symbol table.
14786 2018-03-16 Martin Liska <mliska@suse.cz>
14789 * multiple_target.c (create_dispatcher_calls): Redirect also
14792 2018-03-16 Jakub Jelinek <jakub@redhat.com>
14796 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
14797 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
14798 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
14800 2018-03-16 Julia Koval <julia.koval@intel.com>
14802 * doc/invoke.texi (Skylake Server): Add CLWB.
14803 Cannonlake): Remove CLWB.
14805 2018-03-16 Jakub Jelinek <jakub@redhat.com>
14807 PR tree-optimization/84841
14808 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
14810 (FLOAT_ONE_CONST_TYPE): Define.
14811 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
14812 (sort_by_operand_rank): Put entries with higher constant_type last
14813 rather than first to match comments.
14815 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
14817 * config/nios2/nios2.md (movsi_internal): Fix thinko in
14820 2018-03-15 Jakub Jelinek <jakub@redhat.com>
14823 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
14824 check and use address of target always.
14826 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
14829 * config/i386/i386.c (indirect_thunk_needed): Update comments.
14830 (indirect_thunk_bnd_needed): Likewise.
14831 (indirect_thunks_used): Likewise.
14832 (indirect_thunks_bnd_used): Likewise.
14833 (indirect_return_needed): New.
14834 (indirect_return_bnd_needed): Likewise.
14835 (output_indirect_thunk_function): Add a bool argument for
14837 (output_indirect_thunk_function): Don't generate alias for
14838 function return thunk.
14839 (ix86_code_end): Call output_indirect_thunk_function to generate
14840 function return thunks.
14841 (ix86_output_function_return): Set indirect_return_bnd_needed
14842 and indirect_return_needed instead of indirect_thunk_bnd_needed
14843 and indirect_thunk_needed.
14845 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
14847 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
14848 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
14849 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
14851 2018-03-15 David Malcolm <dmalcolm@redhat.com>
14852 Paul Hua <paul.hua.gm@gmail.com>
14855 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
14857 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
14859 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
14860 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
14861 resp. SFmode cases.
14863 2018-03-15 Tamar Christina <tamar.christina@arm.com>
14866 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
14867 instead of GET_MODE_SIZE when comparing Units.
14869 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
14872 * varasm.c (hash_section): Return an unchangeble hash value
14873 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
14874 Return !aarch64_can_use_per_function_literal_pools_p ().
14876 2018-03-15 Jakub Jelinek <jakub@redhat.com>
14879 * optabs.c (emit_conditional_move): Pass address of cmode's copy
14880 rather than address of cmode as last argument to prepare_cmp_insn.
14882 2018-03-15 Julia Koval <julia.koval@intel.com>
14884 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
14885 F_AVX512VNNI, F_AVX512BITALG): New.
14887 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
14890 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
14891 insn for floating-point loads and stores.
14893 2018-03-14 Carl Love <cel@us.ibm.com>
14895 * config/rs6000/rs6000-c.c: Add macro definitions for
14896 ALTIVEC_BUILTIN_VEC_PERMXOR.
14897 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
14898 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
14899 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
14900 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
14902 * config/doc/extend.texi: Add prototypes for vec_permxor.
14904 2018-03-14 David Malcolm <dmalcolm@redhat.com>
14907 * diagnostic-show-locus.c (class layout_point): Convert m_line
14908 from int to linenum_type.
14909 (line_span::comparator): Use linenum "compare" function when
14910 comparing line numbers.
14911 (test_line_span): New function.
14912 (layout_range::contains_point): Convert param "row" from int to
14914 (layout_range::intersects_line_p): Likewise.
14915 (layout::will_show_line_p): Likewise.
14916 (layout::print_source_line): Likewise.
14917 (layout::should_print_annotation_line_p): Likewise.
14918 (layout::print_annotation_line): Likewise.
14919 (layout::print_leading_fixits): Likewise.
14920 (layout::annotation_line_showed_range_p): Likewise.
14921 (struct line_corrections): Likewise for field m_row.
14922 (line_corrections::line_corrections): Likewise for param "row".
14923 (layout::print_trailing_fixits): Likewise.
14924 (layout::get_state_at_point): Likewise.
14925 (layout::get_x_bound_for_row): Likewise.
14926 (layout::print_line): Likewise.
14927 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
14928 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
14929 * input.c (selftest::test_linenum_comparisons): New function.
14930 (selftest::input_c_tests): Call it.
14931 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
14932 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
14933 * selftest.h (ASSERT_GT): New macro.
14934 (ASSERT_GT_AT): New macro.
14935 (ASSERT_LT): New macro.
14936 (ASSERT_LT_AT): New macro.
14938 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
14940 PR rtl-optimization/84780
14941 * combine.c (distribute_links): Don't make a link based on pc_rtx.
14943 2018-03-14 Martin Liska <mliska@suse.cz>
14945 * tree.c (record_node_allocation_statistics): Use
14946 get_stats_node_kind.
14947 (get_stats_node_kind): New function extracted from
14948 record_node_allocation_statistics.
14949 (free_node): Use get_stats_node_kind.
14951 2018-03-14 Richard Biener <rguenther@suse.de>
14953 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
14954 that the value-set of ANTIC_IN doesn't grow.
14957 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
14959 (BB_VISITED_WITH_VISITED_SUCCS): New define.
14960 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
14962 2018-03-14 Julia Koval <julia.koval@intel.com>
14964 * config.gcc (icelake-client, icelake-server): New.
14966 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
14967 (initial_ix86_arch_features): Ditto.
14968 (PTA_SKYLAKE): Add SGX.
14969 (PTA_ICELAKE): Remove.
14970 (PTA_ICELAKE_CLIENT): New.
14971 (PTA_ICELAKE_SERVER): New.
14972 (ix86_option_override_internal): Split up icelake on icelake client and
14974 (get_builtin_code_for_version): Ditto.
14975 (fold_builtin_cpu): Ditto.
14976 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
14977 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
14978 * config/i386/i386.h (processor_type): Ditto.
14979 * doc/invoke.texi: Ditto.
14981 2018-03-14 Jakub Jelinek <jakub@redhat.com>
14984 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
14985 INTEGER_CST offset, add it together with bitpos / 8 and
14986 sign extend based on POINTER_SIZE.
14990 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
14993 * config/i386/constraints.md (Yc): New register constraint.
14994 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
14995 Use Yc constraint for alternative 2 of operand 0. Remove
14996 preferred_for_speed attribute.
14998 2018-03-14 Richard Biener <rguenther@suse.de>
15000 PR tree-optimization/84830
15001 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
15002 with the old one to avoid oscillations.
15004 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
15007 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
15009 (assign_by_spills): Return a flag of reload assignment failure.
15010 Do not process the reload assignment failures. Do not spill other
15011 reload pseudos if they has the same reg class. Update n if
15013 (lra_assign): Add a return arg. Set up from the result of
15014 assign_by_spills call.
15015 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
15016 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
15017 usage_insns if it is not NULL.
15018 (spill_hard_reg_in_range): New function.
15019 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
15020 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
15021 function prototypes.
15022 (lra_assign): Change prototype.
15023 * lra.c (lra): Add code to deal with fails by splitting hard reg
15026 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
15028 * config/riscv/riscv.opt (mrelax): New option.
15029 * config/riscv/riscv.c (riscv_file_start): Emit ".option
15030 "norelax" when riscv_mrelax is disabled.
15031 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
15033 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15036 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
15037 reassociation for int modes.
15039 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15041 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
15042 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
15044 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
15045 * config/aarch64/aarch64-sve.md
15046 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
15047 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
15048 (*extend<mode><Vwide>2): Rename to...
15049 (aarch64_sve_extend<mode><Vwide>2): ...this.
15050 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
15051 renaming the old pattern to...
15052 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
15054 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
15055 define_expand, renaming the old pattern to...
15056 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
15057 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
15058 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
15059 account when deciding which SVE instruction the optab should use.
15060 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
15062 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15064 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
15065 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
15066 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
15067 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
15068 (tlsdesc_small_<mode>): Turn a define_expand and use
15069 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
15070 (tlsdesc_small_advsimd_<mode>): ...this.
15071 (tlsdesc_small_sve_<mode>): New pattern.
15073 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15075 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
15076 (UNSPEC_UMUL_HIGHPART): New constants.
15077 (MUL_HIGHPART): New int iteraor.
15078 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
15079 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
15081 (*<su>mul<mode>3_highpart): New define_insn.
15083 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15086 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15089 2018-03-13 Martin Liska <mliska@suse.cz>
15092 * (sem_item_optimizer::sem_item_optimizer): Initialize new
15094 (sem_item_optimizer::~sem_item_optimizer): Release it.
15095 (sem_item_optimizer::merge_classes): Register variable aliases.
15096 (sem_item_optimizer::fixup_pt_set): New function.
15097 (sem_item_optimizer::fixup_points_to_sets): Likewise.
15098 * ipa-icf.h: Declare new variables and functions.
15100 2018-03-13 Jakub Jelinek <jakub@redhat.com>
15102 PR middle-end/84834
15103 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
15104 integer_pow2p@2 and test integer_pow2p in condition.
15105 (A < 0 ? C : 0): Similarly for @1.
15107 PR middle-end/84831
15108 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
15109 characters starting at p contain '\0' character, don't look beyond
15113 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
15114 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
15117 * reg-stack.c (change_stack): Change update_end var from int to
15118 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
15119 also call set_block_for_insn on the newly added insns and rescan.
15122 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
15123 on the last operand.
15126 * tree.c (stabilize_reference_1): Return save_expr (e) for
15127 STATEMENT_LIST even if it doesn't have side-effects.
15129 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
15131 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
15133 2018-03-12 Renlin Li <renlin.li@arm.com>
15135 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
15136 aarch64_output_scalar_simd_mov_immediate.
15138 2018-03-12 Martin Sebor <msebor@redhat.com>
15140 PR tree-optimization/83456
15141 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
15142 for perfectly overlapping calls to memcpy.
15143 (gimple_fold_builtin_memory_chk): Same.
15144 (gimple_fold_builtin_strcpy): Handle no-warning.
15145 (gimple_fold_builtin_stxcpy_chk): Same.
15146 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
15148 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
15150 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
15151 parameter. Use it for SFmode.
15152 (rs6000_function_arg_advance_1): Adjust.
15153 (rs6000_function_arg): Adjust.
15154 (rs6000_gimplify_va_arg): Pass false for that new parameter.
15156 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
15158 PR rtl-optimization/84169
15159 PR rtl-optimization/84780
15160 * combine.c (can_combine_p): Check for a 2-insn combination whether
15161 the destination register is used between the two insns, too.
15163 2018-03-12 Richard Biener <rguenther@suse.de>
15165 PR tree-optimization/84803
15166 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
15167 for refs DR analysis didn't process.
15169 2018-03-12 Richard Biener <rguenther@suse.de>
15171 PR tree-optimization/84777
15172 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
15173 force-vectorize loops ignore whether we are optimizing for size.
15175 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
15177 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
15178 (TARGET_MD_ASM_ADJUST): Define.
15180 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
15181 Kito Cheng <kito.cheng@gmail.com>
15182 Chung-Ju Wu <jasonwucj@gmail.com>
15184 * config/nds32/nds32.c (nds32_compute_stack_frame,
15185 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
15186 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
15187 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
15188 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
15189 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
15190 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
15191 * config/nds32/nds32.md (prologue, epilogue): Use macro
15192 NDS32_V3PUSH_AVAILABLE_P to do checking.
15194 2018-03-11 Jakub Jelinek <jakub@redhat.com>
15197 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
15198 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
15199 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
15200 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
15201 addition of most attributes on !orig_type_die or the attribute not
15202 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
15204 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15205 Chung-Ju Wu <jasonwucj@gmail.com>
15207 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
15208 __NDS32_VH__ macro.
15209 * config/nds32/nds32.opt (mvh): New option.
15211 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15212 Chung-Ju Wu <jasonwucj@gmail.com>
15214 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
15216 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
15217 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
15220 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15221 Chung-Ju Wu <jasonwucj@gmail.com>
15223 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
15225 * config/nds32/nds32-multiple.md (strlensi): New pattern.
15226 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
15228 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
15229 Kito Cheng <kito.cheng@gmail.com>
15230 Chung-Ju Wu <jasonwucj@gmail.com>
15232 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
15233 UNSPEC_FFMISM and UNSPEC_FLMISM.
15234 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
15235 for ffb, ffmism and flmism.
15236 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
15237 (unspec_ffmism): Ditto.
15238 (unspec_flmism): Ditto.
15239 (nds32_expand_builtin_impl): Check if string extension is available.
15240 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
15241 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
15243 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
15246 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
15249 * lra-assigns.c (assign_by_spills): Return a flag of reload
15250 assignment failure. Do not process the reload assignment
15251 failures. Do not spill other reload pseudos if they has the same
15253 (lra_assign): Add a return arg. Set up from the result of
15254 assign_by_spills call.
15255 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
15256 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
15257 usage_insns if it is not NULL.
15258 (spill_hard_reg_in_range): New function.
15259 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
15260 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
15261 function prototypes.
15262 (lra_assign): Change prototype.
15263 * lra.c (lra): Add code to deal with fails by splitting hard reg
15266 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
15269 * config/i386/i386.opt: Replace Enforcment with Enforcement.
15271 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
15274 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
15275 (dw_val_node): Add val_symbolic_view.
15276 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
15277 (symview_upper_bound): New.
15278 (new_line_info_table): Initialize symviews_since_reset.
15279 (dwarf2out_source_line): Count symviews_since_reset and set
15280 symview_upper_bound.
15281 (dw_val_equal_p): Handle symview.
15282 (add_AT_symview): New.
15283 (print_dw_val): Handle symview.
15284 (attr_checksum, attr_checksum_ordered): Likewise.
15285 (same_dw_val_p, size_of_die): Likewise.
15286 (value_format, output_die): Likewise.
15287 (add_high_low_attributes): Use add_AT_symview for entry_view.
15288 (dwarf2out_finish): Reset symview_upper_bound, clear
15291 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
15294 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
15295 Add strict argument and use it.
15296 (rs6000_split_multireg_move): Update for new strict argument.
15297 (mem_operand_gpr): Disallow all non-offsettable addresses.
15298 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
15300 2018-03-09 Jakub Jelinek <jakub@redhat.com>
15303 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
15304 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
15305 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
15308 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
15309 decl, use remap_type if we want to use the type.
15311 2018-03-09 Martin Sebor <msebor@redhat.com>
15313 PR tree-optimization/84526
15314 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
15316 (builtin_access::generic_overlap): Be prepared to handle non-array
15319 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
15321 PR rtl-optimization/84682
15322 * lra-constraints.c (process_address_1): Check is_address flag
15323 for address constraints.
15324 (process_alt_operands): Likewise.
15325 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
15326 preprocess_constraints.
15327 * recog.h (preprocess_constraints): Add oploc parameter.
15329 * recog.c (preprocess_constraints): Test address_operand for
15330 CT_ADDRESS constraints.
15332 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
15335 * lra-assigns.c (assign_by_spills): Return a flag of reload
15336 assignment failure. Do not process the reload assignment
15337 failures. Do not spill other reload pseudos if they has the same
15339 (lra_assign): Add a return arg. Set up from the result of
15340 assign_by_spills call.
15341 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
15342 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
15343 usage_insns if it is not NULL.
15344 (spill_hard_reg_in_range): New function.
15345 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
15346 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
15347 function prototypes.
15348 (lra_assign): Change prototype.
15349 * lra.c (lra): Add code to deal with fails by splitting hard reg
15352 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15355 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
15356 Accept complain bool parameter. Only emit errors if it is true.
15357 (arm_parse_cpu_option_name): Likewise.
15358 (arm_target_thumb_only): Adjust callers of the above.
15359 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
15360 prototype to take a default true bool parameter.
15361 (arm_parse_arch_option_name): Likewise.
15363 2018-03-09 David Malcolm <dmalcolm@redhat.com>
15364 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
15368 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
15369 * configure: Regenerate.
15370 * configure.ac ("linker --version-script option"): New.
15371 ("linker soname option"): New.
15373 2018-03-09 Richard Biener <rguenther@suse.de>
15375 PR tree-optimization/84775
15376 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
15377 immediate uses of predicate stmts and mark them modified.
15380 PR tree-optimization/84178
15381 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
15383 (version_loop_for_if_conversion): Delay update_ssa call.
15384 (tree_if_conversion): Delay update_ssa until after predicate
15387 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
15390 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
15391 when the function accesses prior frames.
15393 2018-03-08 Jakub Jelinek <jakub@redhat.com>
15396 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
15397 gen_llsym, otherwise call maybe_gen_llsym.
15399 PR inline-asm/84742
15400 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
15401 has ',' character inside of it.
15403 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15406 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
15407 as clobbering CC_REGNUM.
15409 2018-03-08 Richard Biener <rguenther@suse.de>
15411 PR middle-end/84552
15412 * tree-scalar-evolution.c: Include tree-into-ssa.h.
15413 (follow_copies_to_constant): Do not follow SSA names registered
15416 2018-03-08 Richard Biener <rguenther@suse.de>
15418 PR tree-optimization/84178
15419 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
15421 (version_loop_for_if_conversion): Delay update_ssa call.
15422 (tree_if_conversion): Delay update_ssa until after predicate
15425 2018-03-08 David Malcolm <dmalcolm@redhat.com>
15427 PR tree-optimization/84178
15428 * tree-if-conv.c (release_bb_predicate): Remove the
15429 the assertion that the stmts have NULL use_ops.
15430 Discard the statements, asserting that they haven't
15431 yet been added to a BB.
15433 2018-03-08 Richard Biener <rguenther@suse.de>
15435 PR tree-optimization/84746
15436 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
15437 (phi_translate): Pass in destination ANTIC_OUT set.
15438 (phi_translate_1): Likewise. For a simplified result lookup
15439 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
15440 (phi_translate_set): Adjust.
15441 (do_pre_regular_insertion): Likewise.
15442 (do_pre_partial_partial_insertion): Likewise.
15444 2018-03-08 Martin Liska <mliska@suse.cz>
15446 PR gcov-profile/84735
15447 * doc/gcov.texi: Document usage of profile files.
15448 * gcov-io.h: Document changes in the format.
15450 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
15454 * dwarf2out.c (struct dw_line_info_table): Update comments for
15456 (FORCE_RESET_NEXT_VIEW): New.
15457 (FORCE_RESETTING_VIEW_P): New.
15458 (RESETTING_VIEW_P): Check for -1 too.
15459 (ZERO_VIEW_P): Likewise.
15460 (new_line_info_table): Force-reset next view.
15461 (dwarf2out_begin_function): Likewise.
15462 (dwarf2out_source_line): Simplify zero_view_p initialization.
15463 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
15464 view directly. Omit view when omitting .loc at line 0.
15466 2018-03-08 Jakub Jelinek <jakub@redhat.com>
15468 PR tree-optimization/84740
15469 * tree-switch-conversion.c (process_switch): Call build_constructors
15470 only if info.phi_count is non-zero.
15472 PR tree-optimization/84739
15473 * tree-tailcall.c (find_tail_calls): Check call arguments against
15474 DECL_ARGUMENTS (current_function_decl) rather than
15475 DECL_ARGUMENTS (func) when checking for tail recursion.
15477 2018-03-07 Jakub Jelinek <jakub@redhat.com>
15479 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
15480 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
15481 Volker Reichelt's entry and add entries for people that perform
15482 GCC fuzzy testing and report numerous bugs.
15484 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
15487 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
15488 readonly data in sdata, if that is disabled.
15489 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
15490 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
15491 -mreadonly-in-sdata option.
15493 2018-03-07 Martin Sebor <msebor@redhat.com>
15495 PR tree-optimization/84468
15496 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
15497 basic block when looking for nul assignment.
15499 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
15502 * except.h (output_function_exception_table): Adjust prototype.
15503 * except.c (output_function_exception_table): Remove FNNAME parameter
15504 and add SECTION parameter. Ouput one part of the table at a time.
15505 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
15506 the first part of the exception table and emit unwind directives.
15507 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
15508 (i386_pe_seh_cold_init): Likewise.
15509 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
15510 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
15511 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
15512 (ix86_output_call_insn): Emit a nop in one more case for SEH.
15513 * config/i386/winnt.c: Include except.h.
15514 (struct seh_frame_state): Add reg_offset, after_prologue and
15515 in_cold_section fields.
15516 (i386_pe_seh_end_prologue): Set seh->after_prologue.
15517 (i386_pe_seh_cold_init): New function.
15518 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
15519 to seh->in_cold_section.
15520 (seh_emit_push): Record the offset of the push.
15521 (seh_emit_save): Record the offet of the save.
15522 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
15523 Test seh->after_prologue to disregard the epilogue.
15524 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
15525 (i386_pe_end_cold_function): New function.
15527 2018-03-07 Jakub Jelinek <jakub@redhat.com>
15530 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
15531 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
15534 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
15536 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
15537 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
15539 PR middle-end/84723
15540 * multiple_target.c: Include tree-inline.h and intl.h.
15541 (expand_target_clones): Diagnose and fail if node->definition and
15542 !tree_versionable_function_p (node->decl).
15544 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
15546 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
15548 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
15549 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15550 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
15552 2018-03-06 Jakub Jelinek <jakub@redhat.com>
15555 * combine.c (try_combine): Use reg_or_subregno instead of handling
15556 just paradoxical SUBREGs and REGs.
15558 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
15560 * config/arc/arc.c (arc_finalize_pic): Remove function.
15561 (arc_must_save_register): We use single base PIC register, remove
15562 checks to save/restore the PIC register.
15563 (arc_expand_prologue): Likewise.
15564 * config/arc/arc-protos.h (arc_set_default_type_attributes):
15566 (arc_verify_short): Likewise.
15567 (arc_attr_type): Likewise.
15568 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
15569 (walk_stores): Likewise.
15570 (arc_address_cost): Make it static.
15571 (arc_verify_short): Likewise.
15572 (branch_dest): Likewise.
15573 (arc_attr_type): Likewise.
15574 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
15575 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
15576 (arc_final_prescan_insn): Remove inserting the nops due to
15577 hardware hazards. It is done in reorg step.
15578 (insn_length_variant_t): Remove.
15579 (insn_length_parameters_t): Likewise.
15580 (arc_insn_length_parameters): Likewise.
15581 (arc_get_insn_variants): Likewise.
15582 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
15584 2018-03-06 Jakub Jelinek <jakub@redhat.com>
15586 PR inline-asm/84683
15587 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
15590 PR tree-optimization/84687
15591 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
15593 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
15595 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15597 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
15598 Rename to ppc_speculation_barrier.
15599 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
15600 __builtin_ppc_speculation_barrier.
15602 2018-03-05 Jakub Jelinek <jakub@redhat.com>
15605 * combine.c (combine_simplify_rtx): Don't try to simplify if
15606 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
15609 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
15611 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
15612 to 32 bytes when compiling for POWER9.
15614 2018-03-05 Jakub Jelinek <jakub@redhat.com>
15617 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
15618 regparm >= 3 with no arg reg available also for calls with
15619 flag_force_indirect_call. Pass decl to ix86_function_regparm.
15622 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
15624 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
15626 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
15629 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
15631 2018-03-05 Richard Biener <rguenther@suse.de>
15633 PR tree-optimization/84486
15634 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
15635 When inserting a __builtin_assume_aligned call set the LHS
15636 SSA name alignment info accordingly.
15638 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
15640 PR tree-optimization/84114
15641 * config/aarch64/aarch64.c (aarch64_reassociation_width)
15642 Avoid reassociation of FLOAT_MODE addition.
15644 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
15646 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
15647 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
15648 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
15649 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
15650 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
15651 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
15652 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
15654 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
15655 __builtin_ia32_wbinvd): New builtins.
15656 (SPECIAL_ARGS2): New.
15657 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
15658 (SPECIAL_ARGS2): New.
15659 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
15660 (ix86_valid_target_attribute_inner_p): Ditto.
15661 (ix86_init_mmx_sse_builtins): Add special_args2.
15662 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
15663 TARGET_WBNOINVD_P): New.
15664 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
15665 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
15666 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
15667 * config/i386/immintrin.h (_wbinvd): New intrinsic.
15668 * config/i386/pconfigintrin.h: New file.
15669 * config/i386/wbnoinvdintrin.h: Ditto.
15670 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
15672 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
15674 2018-03-05 Richard Biener <rguenther@suse.de>
15676 PR tree-optimization/84670
15677 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
15679 (BB_VISITED_WITH_VISITED_SUCCS): New define.
15680 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
15681 (compute_antic_aux): Only assert the number of values in ANTIC_IN
15682 doesn't grow if all successors (recursively) were visited at least
15685 2018-03-05 Richard Biener <rguenther@suse.de>
15687 PR tree-optimization/84650
15688 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
15689 if executed in the loop pipeline.
15691 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
15693 * doc/configfiles.texi (Configuration Files): Move info about
15694 conditionalizing $target-protos.h to...
15695 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
15696 differs from $target-protos.h.
15698 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
15699 Chung-Ju Wu <jasonwucj@gmail.com>
15701 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
15702 * config/nds32/nds32-multiple.md (setmemsi): Define.
15703 * config/nds32/nds32-memory-manipulation.c
15704 (nds32_gen_dup_4_byte_to_word_value): New.
15705 (emit_setmem_word_loop): New.
15706 (emit_setmem_byte_loop): New.
15707 (nds32_expand_setmem_loop): New.
15708 (nds32_expand_setmem_loop_v3m): New.
15709 (nds32_expand_setmem_unroll): New.
15710 (nds32_expand_setmem): New.
15712 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
15713 Chung-Ju Wu <jasonwucj@gmail.com>
15715 * config/nds32/nds32-memory-manipulation.c
15716 (nds32_emit_load_store): New.
15717 (nds32_emit_post_inc_load_store): New.
15718 (nds32_emit_mem_move): New.
15719 (nds32_emit_mem_move_block): New.
15720 (nds32_expand_movmemsi_loop_unknown_size): New.
15721 (nds32_expand_movmemsi_loop_known_size): New.
15722 (nds32_expand_movmemsi_loop): New.
15723 (nds32_expand_movmemsi_unroll): New.
15724 (nds32_expand_movmemqi): Rename ...
15725 (nds32_expand_movmemsi): ... to this.
15726 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
15727 (movmemsi): ... to this.
15728 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
15729 (nds32_expand_movmemsi): ... to this.
15731 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
15732 Monk Chiang <sh.chiang04@gmail.com>
15733 Chung-Ju Wu <jasonwucj@gmail.com>
15735 * config/nds32/nds32-protos.h
15736 (nds32_expand_load_multiple): New arguments.
15737 (nds32_expand_store_multiple): Ditto.
15738 (nds32_valid_multiple_load_store): Rename ...
15739 (nds32_valid_multiple_load_store_p): ... to this.
15740 * config/nds32/nds32-memory-manipulation.c
15741 (nds32_expand_load_multiple): Refine implementation.
15742 (nds32_expand_store_multiple): Ditto.
15743 * config/nds32/nds32-multiple.md
15744 (load_multiple): Update nds32_expand_load_multiple interface.
15745 (store_multiple): Update nds32_expand_store_multiple interface.
15746 * config/nds32/nds32-predicates.c
15747 (nds32_valid_multiple_load_store): Rename ...
15748 (nds32_valid_multiple_load_store_p): ... to this and refine
15750 * config/nds32/predicates.md
15751 (nds32_load_multiple_and_update_address_operation): New predicate.
15752 (nds32_store_multiple_and_update_address_operation): New predicate.
15754 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
15755 Chung-Ju Wu <jasonwucj@gmail.com>
15757 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
15758 (combo): New attribute.
15759 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
15761 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
15763 * config/nds32/nds32.opt: Change -mcmodel= default value.
15765 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
15766 Monk Chiang <sh.chiang04@gmail.com>
15767 Chung-Ju Wu <jasonwucj@gmail.com>
15769 * config/nds32/constants.md (unspec_element): New enum.
15770 * config/nds32/constraints.md (Umw): New constraint.
15771 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
15772 * config/nds32/nds32-intrinsic.md: Likewise.
15773 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
15774 (nds32_valid_smw_lwm_base_p): New.
15775 (nds32_output_smw_single_word): New.
15776 (nds32_output_lmw_single_word): New.
15777 (nds32_expand_unaligned_load): New.
15778 (nds32_expand_unaligned_store): New.
15779 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
15780 (nds32_output_smw_single_word): Declare.
15781 (nds32_output_lmw_single_word): Declare.
15782 (nds32_expand_unaligned_load): Declare.
15783 (nds32_expand_unaligned_store): Declare.
15784 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
15785 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
15786 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
15787 NDS32_BUILTIN_UASTORE_DW.
15788 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
15791 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
15792 Kito Cheng <kito.cheng@gmail.com>
15793 Chung-Ju Wu <jasonwucj@gmail.com>
15795 * config/nds32/nds32-intrinsic.c
15796 (nds32_expand_builtin_null_ftype_reg): Delete.
15797 (nds32_expand_builtin_reg_ftype_imm): Ditto.
15798 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
15799 (nds32_read_argument): New.
15800 (nds32_legitimize_target): Ditto.
15801 (nds32_legitimize_argument): Ditto.
15802 (nds32_check_constant_argument): Ditto.
15803 (nds32_expand_unop_builtin): Ditto.
15804 (nds32_expand_unopimm_builtin): Ditto.
15805 (nds32_expand_binop_builtin): Ditto.
15806 (nds32_builtin_decl_impl): Ditto.
15807 (builtin_description): Ditto.
15808 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
15809 (nds32_init_builtins_impl): Ditto.
15810 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
15811 (nds32_builtin_decl): New.
15812 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
15813 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
15815 2018-03-02 Jeff Law <law@redhat.com>
15817 * reorg.c (stop_search_p): Handle DEBUG_INSN.
15818 (redundant_insn, fill_simple_delay_slots): Likewise.
15819 (fill_slots_from_thread): Likewise.
15820 * resource.c (mark_referenced_resources): Likewise.
15821 (mark_set_resources, find_dead_or_set_registers): Likewise.
15823 2018-03-02 Jakub Jelinek <jakub@redhat.com>
15825 * substring-locations.h (format_warning_va): Formatting fix for
15826 ATTRIBUTE_GCC_DIAG.
15827 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
15829 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
15830 * substring-locations.c: Include intl.h.
15831 (format_warning_va): Turned into small wrapper around
15832 format_warning_n_va, renamed to ...
15833 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
15834 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
15836 (format_warning_at_substring_n): New function.
15837 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
15838 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
15839 format_warning_at_substring with just a shorter name instead of
15840 const function pointer.
15841 (fmtwarn_n): New function.
15842 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
15843 appropriate, get rid of all the fmtstr temporaries, move conditionals
15844 with G_() wrapped string literals directly into fmtwarn arguments,
15845 cast dir.len to (int), formatting fixes.
15847 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
15849 * doc/invoke.texi: Remove "Cilk Plus" references.
15851 2018-03-02 Jakub Jelinek <jakub@redhat.com>
15852 Richard Biener <rguenther@suse.de>
15855 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
15856 for error or warning attributes if CALL_FROM_THUNK_P is set.
15859 2018-03-02 Jakub Jelinek <jakub@redhat.com>
15862 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
15863 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
15866 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
15867 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
15869 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
15870 instead of -1U in last predictors element's probability member.
15872 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
15875 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
15876 arguments if they are comparable.
15878 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
15880 PR tree-optimization/84634
15881 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
15882 masks and masked_loop_p with a single loop_masks, making sure it's
15883 null for bb vectorization.
15885 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
15887 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
15888 (vect_analyze_data_ref_access): Use loop->safe_len rather than
15889 loop->force_vectorize to check whether there is no alias.
15891 2018-03-02 Jakub Jelinek <jakub@redhat.com>
15894 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
15896 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
15898 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
15899 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
15900 instead of a loop around prev_real_insn.
15901 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
15904 PR inline-asm/84625
15905 * config/i386/i386.c (ix86_print_operand): Use conditional
15906 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
15909 2018-03-02 Richard Biener <rguenther@suse.de>
15911 PR tree-optimization/84427
15912 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
15913 (bitmap_set_subtract_values): Rewrite to handle multiple
15916 (prune_clobbered_mems): Likewise.
15917 (phi_translate): Take edge instead of pred/phiblock.
15918 (phi_translate_1): Likewise.
15919 (phi_translate_set): Likewise. Insert all translated
15920 exprs for a value into the set, keeping possibly multiple
15921 expressions per value.
15922 (compute_antic_aux): Adjust for phi_translate changes.
15923 When intersecting union the expressions and prune those
15924 not in the final value set, keeping possibly multiple
15925 expressions per value. Do not use value-insertion
15926 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
15927 all expressions. Add verification that the value-sets
15928 only shrink during iteration.
15929 (compute_partial_antic_aux): Adjust for the phi_translate changes.
15930 (do_pre_regular_insertion): Likewise.
15931 (do_pre_partial_partial_insertion): Likewise.
15933 2018-03-02 Richard Biener <rguenther@suse.de>
15936 * config/darwin.c (saved_debug_info_level): New static global.
15937 (darwin_asm_lto_start): Disable debug info generation for LTO out.
15938 (darwin_asm_lto_end): Restore debug info generation settings.
15940 2018-03-01 Martin Liska <mliska@suse.cz>
15943 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
15944 volatile arguments.
15946 2018-03-01 Richard Biener <rguenther@suse.de>
15949 * dwarf2out.c (gen_variable_die): Properly handle late VLA
15950 type annotation with LTO when debug was disabled at compile-time.
15952 2018-03-01 Matthew Fortune <mfortune@gmail.com>
15954 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
15956 (mips_final_postscan_insn): Likewise.
15958 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
15960 PR rtl-optimization/84528
15961 * alias.c (init_alias_target): Add commentary.
15962 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
15963 a unique base value if the frame pointer is not eliminated
15964 to the stack pointer.
15966 2018-03-01 Tom de Vries <tom@codesourcery.com>
15968 PR rtl-optimization/83327
15969 * lra-int.h (hard_regs_spilled_into): Declare.
15970 * lra.c (hard_regs_spilled_into): Define.
15971 (init_reg_info): Init hard_regs_spilled_into.
15972 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
15973 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
15974 (process_bb_lives): Handle hard_regs_spilled_into.
15975 (lra_create_live_ranges_1): Before doing liveness propagation, clear
15976 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
15978 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
15980 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
15981 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
15982 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
15983 * config/rs6000/aix72.h: New file.
15985 2018-02-28 Jakub Jelinek <jakub@redhat.com>
15987 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
15988 instead of warning_at with conditional singular and plural messages
15992 * stor-layout.c (update_alignment_for_field): For
15993 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
15994 && !DECL_PACKED (field), do the alignment update, just use
15995 only desired_align instead of MAX (type_align, desired_align)
15997 (place_field): Don't do known_align < desired_align handling
15998 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
15999 is non-NULL, instead do it after rli->prev_field handling and
16000 only if not within a bitfield word. For DECL_PACKED (field)
16001 use type_align of BITS_PER_UNIT.
16003 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
16005 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
16006 superfluous parentheses and trailing spaces.
16008 2018-02-28 Richard Biener <rguenther@suse.de>
16010 PR tree-optimization/84584
16011 * graphite-scop-detection.c (scop_detection::add_scop): Discard
16012 SCoPs with fake exit edge.
16014 2018-02-28 Martin Liska <mliska@suse.cz>
16017 * timevar.c (timer::print): Fix format to properly print 100%
16020 2018-02-28 Richard Biener <rguenther@suse.de>
16022 PR middle-end/84607
16023 * genmatch.c (capture_info::walk_match): Do not mark
16024 captured expressions without operands as expr_p given
16025 they act more like predicates and should be subject to
16026 "lost tail" side-effect preserving.
16028 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
16030 PR rtl-optimization/81611
16031 * auto-inc-dec.c (attempt_change): Move dead note from
16032 mem_insn if it's the next use of regno
16033 (find_address): Take address use of reg holding
16034 non-incremented value. Add parm to limit search to the named
16036 (merge_in_block): Attempt to use a mem insn that is the next
16037 use of the original regno.
16039 2018-02-27 Martin Sebor <msebor@redhat.com>
16042 * doc/invoke.texi (-Wmissing-attributes): New option.
16043 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
16045 2018-02-27 Martin Sebor <msebor@redhat.com>
16047 PR translation/84207
16048 * diagnostic-core.h (warning_n, error_n, inform_n): Change
16049 n argument to unsigned HOST_WIDE_INT.
16050 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
16051 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
16052 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
16053 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
16055 2018-02-27 Richard Biener <rguenther@suse.de>
16057 PR tree-optimization/84512
16058 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
16059 Do not use the estimate returned from record_stmt_cost for
16060 the scalar iteration cost but sum properly using add_stmt_cost.
16062 2018-02-27 Richard Biener <rguenther@suse.de>
16064 PR tree-optimization/84466
16065 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
16066 Adjust last change to less strictly validate use operands.
16068 2018-02-27 Martin Liska <mliska@suse.cz>
16070 PR gcov-profile/84548
16071 * gcov.c (process_file): Allow partial overlap and consider it
16072 also as group functions.
16073 (output_lines): Properly calculate range of lines for a group.
16075 2018-02-27 Martin Liska <mliska@suse.cz>
16077 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
16078 'ggc' suffixes. Change first column width.
16079 (timer::print): Fix formatting of the column.
16081 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
16083 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
16084 preserve inline entry blocks for the sake of debug inline
16085 entry point markers alone.
16086 (remove_unused_locals): Suggest in comments a better place to
16087 force the preservation of inline entry blocks that are
16088 otherwise unused, but do not preserve them.
16090 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16092 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
16094 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16097 * config/i386/constraints.md (Bs): Replace
16098 ix86_indirect_branch_register with
16099 TARGET_INDIRECT_BRANCH_REGISTER.
16101 * config/i386/i386.md (indirect_jump): Likewise.
16102 (tablejump): Likewise.
16103 (*sibcall_memory): Likewise.
16104 (*sibcall_value_memory): Likewise.
16105 Peepholes of indirect call and jump via memory: Likewise.
16106 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
16107 (*sibcall_value_GOT_32): Likewise.
16108 * config/i386/predicates.md (indirect_branch_operand): Likewise.
16109 (GOT_memory_operand): Likewise.
16110 (call_insn_operand): Likewise.
16111 (sibcall_insn_operand): Likewise.
16112 (GOT32_symbol_operand): Likewise.
16113 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
16115 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
16117 PR rtl-optimization/83496
16118 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
16119 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
16120 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
16121 redundant insn, if any.
16122 (relax_delay_slots): Likewise.
16123 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
16125 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
16127 PR tree-optimization/83965
16128 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
16129 that grouped statements are part of a reduction chain. Return
16130 true if the statement is not marked as a reduction itself but
16131 is part of a group.
16132 (vect_recog_dot_prod_pattern): Don't check whether the statement
16133 is part of a group here.
16134 (vect_recog_sad_pattern): Likewise.
16135 (vect_recog_widen_sum_pattern): Likewise.
16137 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
16140 * final.c (rest_of_clean_state): Also look for calls inside sequences.
16142 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16145 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
16147 (ix86_output_indirect_function_return): New prototype.
16148 (ix86_split_simple_return_pop_internal): Likewise.
16149 * config/i386/i386.c (indirect_return_via_cx): New.
16150 (indirect_return_via_cx_bnd): Likewise.
16151 (indirect_thunk_name): Handle return va CX_REG.
16152 (output_indirect_thunk_function): Create alias for
16153 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
16154 (ix86_output_indirect_jmp): Remove the bool argument.
16155 (ix86_output_indirect_function_return): New function.
16156 (ix86_split_simple_return_pop_internal): Likewise.
16157 * config/i386/i386.md (*indirect_jump): Don't pass false
16158 to ix86_output_indirect_jmp.
16159 (*tablejump_1): Likewise.
16160 (simple_return_pop_internal): Change it to define_insn_and_split.
16161 Call ix86_split_simple_return_pop_internal to split it for
16162 -mfunction-return=.
16163 (simple_return_indirect_internal): Call
16164 ix86_output_indirect_function_return instead of
16165 ix86_output_indirect_jmp.
16167 2018-02-26 Jakub Jelinek <jakub@redhat.com>
16170 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
16171 memset and value initialization afterwards.
16173 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
16175 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
16177 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16180 * common/config/aarch64/aarch64-common.c
16181 (aarch_option_optimization_table[]): Switch
16182 off fomit-frame-pointer
16184 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
16185 Chung-Ju Wu <jasonwucj@gmail.com>
16187 * config/nds32/nds32-multiple.md (load_multiple): Disallow
16189 (store_multiple): Ditto.
16191 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
16193 * config.gcc: Add --with-cpu support for nds32 target.
16194 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
16195 * config/nds32/nds32.opt: Add -mcpu= option.
16197 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
16199 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
16200 isel=yes): Warn for these deprecated options.
16202 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
16204 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
16205 ISA_2_5_MASKS_EMBEDDED.
16207 2018-02-23 Jakub Jelinek <jakub@redhat.com>
16209 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
16210 p->max as pointers rather than using iterative_hash_expr.
16212 2018-02-23 Carl Love <cel@us.ibm.com>
16214 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
16215 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
16217 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
16218 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
16219 P8V_BUILTIN_VEC_VUNSIGNED2.
16221 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
16224 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
16225 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
16227 * lra-constraints.c (curr_insn_transform): Set up
16228 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
16230 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
16231 * lra-eliminations.c (spill_pseudos): Ditto.
16232 (process_insn_for_elimination): Ditto.
16233 * lra-lives.c (reg_early_clobber_p): Use the new macros.
16234 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
16235 LRA_NON_CLOBBERED_ALT.
16237 2018-02-22 Martin Sebor <msebor@redhat.com>
16239 PR tree-optimization/84480
16240 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
16241 to maybe_diag_stxncpy_trunc. Call it.
16242 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
16243 from gimple_fold_builtin_strcpy. Print inlining stack.
16244 (handle_builtin_stxncpy): Print inlining stack.
16245 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
16247 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
16250 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
16251 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
16252 and -fcheck-pointer-bounds are used together.
16253 (indirect_thunk_prefix): New enum.
16254 (indirect_thunk_need_prefix): New function.
16255 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
16256 "_nt" instead of "_bnd" for NOTRACK prefix.
16257 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
16258 (output_indirect_thunk_function): Likewise.
16260 (ix86_code_end): Update output_indirect_thunk_function calls.
16261 (ix86_output_indirect_branch_via_reg): Replace
16262 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
16263 (ix86_output_indirect_branch_via_push): Likewise.
16264 (ix86_output_function_return): Likewise.
16265 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
16266 incompatible with -fcf-protection=branch and
16267 -fcheck-pointer-bounds.
16269 2018-02-22 Steve Ellcey <sellcey@cavium.com>
16272 * config/aarch64/aarch64.c (aarch64_print_address_internal):
16273 Change gcc_assert call to output_operand_lossage.
16275 2018-02-22 Steve Ellcey <sellcey@cavium.com>
16277 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
16279 2018-02-22 DJ Delorie <dj@redhat.com>
16280 Sebastian Perta <sebastian.perta@renesas.com>
16281 Oleg Endo <olegendo@gcc.gnu.org>
16283 * config/rx/rx.c (rx_rtx_costs): New function.
16284 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
16286 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16288 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
16290 2018-02-22 Martin Liska <mliska@suse.cz>
16293 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
16294 Add "native" as a possible value.
16296 2018-02-22 Martin Liska <mliska@suse.cz>
16299 * config/i386/i386.c (ix86_option_override_internal):
16300 Add "native" as a possible value for -march and -mtune.
16302 2018-02-22 Jakub Jelinek <jakub@redhat.com>
16305 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
16306 to all type variants.
16308 PR tree-optimization/84503
16309 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
16310 width as info->bitpos + info->bitsize - start.
16311 (merged_store_group::merge_overlapping): Simplify width computation.
16312 (check_no_overlap): New function.
16313 (imm_store_chain_info::try_coalesce_bswap): Compute expected
16314 start + width and last_order of the group, fail if check_no_overlap
16316 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
16317 to group if check_no_overlap fails.
16319 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
16321 * config/rs6000/altivec.md: Delete contraint arguments to
16322 define_expand, define_split, and define_peephole2, and in
16323 define_insn_and_split if always unused.
16324 * config/rs6000/darwin.md: Ditto.
16325 * config/rs6000/dfp.md: Ditto.
16326 * config/rs6000/rs6000.md: Ditto.
16327 * config/rs6000/sync.md: Ditto.
16328 * config/rs6000/vector.md: Ditto.
16329 * config/rs6000/vsx.md: Ditto.
16331 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
16333 * config/rs6000/altivec.md: Write output control strings as braced
16334 blocks instead of double-quoted strings.
16335 * config/rs6000/darwin.md: Ditto.
16336 * config/rs6000/rs6000.md: Ditto.
16337 * config/rs6000/vector.md: Ditto.
16338 * config/rs6000/vsx.md: Ditto.
16340 2018-02-21 Jason Merrill <jason@redhat.com>
16342 PR c++/84314 - ICE with templates and fastcall attribute.
16343 * attribs.c (build_type_attribute_qual_variant): Remove assert.
16345 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
16347 * ipa-cp.c (determine_versionability): Fix comment typos.
16349 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
16352 * ipa-cp.c (determine_versionability): Do not version functions caling
16355 2018-02-21 Martin Liska <mliska@suse.cz>
16358 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
16359 Add "native" as a possible value.
16360 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
16361 the macro when native cpu detection is available.
16363 2018-02-21 Martin Liska <mliska@suse.cz>
16366 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
16367 Add "native" as a possible value.
16368 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
16369 when native cpu detection is available.
16371 2018-02-21 Jakub Jelinek <jakub@redhat.com>
16372 Martin Sebor <msebor@redhat.com>
16374 PR tree-optimization/84478
16375 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
16377 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
16378 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
16379 support which is conservatively correct, for 2 only stay conservative
16380 for maxlen. Formatting and comment capitalization fixes. Add STRICT
16381 argument to the 2 argument get_range_strlen, adjust 6 arg
16382 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
16384 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
16385 (gimple_fold_builtin_strlen): Pass true as last argument to
16388 2018-02-20 Martin Sebor <msebor@redhat.com>
16390 PR middle-end/84095
16391 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
16392 (builtin_memref::set_base_and_offset): Same. Handle inner references.
16393 (builtin_memref::builtin_memref): Factor out parts into
16394 set_base_and_offset and call it.
16396 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
16398 PR middle-end/84406
16399 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
16400 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
16401 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
16402 search at the associated MODE_INT.
16404 2018-02-20 Jeff Law <law@redhat.com>
16406 PR middle-end/82123
16407 PR tree-optimization/81592
16408 PR middle-end/79257
16409 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
16410 for range data rather than using global data.
16411 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
16412 range data rather than using global data.
16413 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
16414 pass it to children as needed.
16415 (struct directive::fmtresult): Similarly.
16416 (struct directive::set_width): Similarly.
16417 (struct directive::set_precision): Similarly.
16418 (format_integer, format_directive, parse_directive): Similarly.
16419 (format_none): Accept unnamed vr_values parameter.
16420 (format_percent, format_floating, format_character): Similarly.
16421 (format_string, format_plain): Similarly.
16422 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
16423 the EVRP range analyzer for range data rather than using global data.
16424 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
16425 gimple-ssa-evrp-analyze.h
16426 (class sprintf_dom_walker): Add after_dom_children member function.
16427 Add evrp_range_analyzer member.
16428 (sprintf_dom_walker::before_dom_children): Call into the EVRP
16429 range analyzer as needed.
16430 (sprintf_dom_walker::after_dom_children): New member function.
16431 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
16433 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
16434 (evrp_range_analyzer::pop_to_marker): Likewise.
16436 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
16438 PR tree-optimization/84419
16439 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
16440 with the required type if its current type is compatible but
16443 2018-02-20 Jakub Jelinek <jakub@redhat.com>
16445 PR middle-end/82004
16446 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
16447 after vectorization.
16449 2018-02-20 Martin Liska <mliska@suse.cz>
16452 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
16453 possible values if we don't have a hint.
16455 2018-02-20 Martin Liska <mliska@suse.cz>
16459 * final.c (shorten_branches): Build align_tab array with one
16461 * opts.c (finish_options): Add alignment option limit check.
16462 (MAX_CODE_ALIGN): Likewise.
16463 (MAX_CODE_ALIGN_VALUE): Likewise.
16464 * doc/invoke.texi: Document maximum allowed option value for
16465 all -falign-* options.
16467 2018-02-19 Jakub Jelinek <jakub@redhat.com>
16470 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
16471 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
16472 * var-tracking.c (emit_note_insn_var_location): Remove all references
16473 to NOTE_INSN_CALL_ARG_LOCATION.
16474 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
16475 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
16476 Use copy_rtx_if_shared.
16477 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
16478 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
16479 (dwarf2out_var_location): Remove handling of
16480 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
16482 * final.c (final_scan_insn): Remove all references to
16483 NOTE_INSN_CALL_ARG_LOCATION.
16484 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
16485 before dumping final insns.
16486 * except.c (emit_note_eh_region_end): Remove all references to
16487 NOTE_INSN_CALL_ARG_LOCATION.
16488 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
16489 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
16490 * config/arc/arc.c (hwloop_optimize): Likewise.
16491 * config/arm/arm.c (create_fix_barrier): Likewise.
16492 * config/s390/s390.c (s390_chunkify_start): Likewise.
16493 * config/sh/sh.c (find_barrier): Likewise.
16494 * config/i386/i386.c (rest_of_insert_endbranch,
16495 ix86_seh_fixup_eh_fallthru): Likewise.
16496 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
16497 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
16498 * config/frv/frv.c (frv_function_prologue): Likewise.
16499 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
16501 (note_outside_basic_block_p): Remove all references to
16502 NOTE_INSN_CALL_ARG_LOCATION.
16503 * gengtype.c (adjust_field_rtx_def): Likewise.
16504 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
16506 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
16507 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
16510 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
16513 PR tree-optimization/84452
16514 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
16515 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
16518 2018-02-19 Martin Liska <mliska@suse.cz>
16521 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
16523 2018-02-19 Martin Liska <mliska@suse.cz>
16524 Richard Sandiford <richard.sandiford@linaro.org>
16526 PR tree-optimization/82491
16527 * gimple-fold.c (get_base_constructor): Make earlier bail out
16530 2018-02-19 Carl Love <cel@us.ibm.com>
16532 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
16533 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
16535 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
16536 P8V_BUILTIN_VEC_NEG.
16538 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
16540 * config/rl78/rl78.md (movdf): New define expand.
16542 2018-02-19 Martin Liska <mliska@suse.cz>
16545 * doc/invoke.texi: Fix typo.
16546 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
16548 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
16550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
16551 handle rs6000_single_float and rs6000_double_float specially for
16554 2018-02-16 Jeff Law <law@redhat.com>
16556 * config/rx/rx.c (add_pop_cfi_notes): New function.;
16557 (pop_regs): Use it.
16559 2018-02-16 Jakub Jelinek <jakub@redhat.com>
16562 * ipa-inline.c (inline_small_functions): Fix a typo.
16564 2018-02-16 Nathan Sidwell <nathan@acm.org>
16566 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
16568 2018-02-16 Carl Love <cel@us.ibm.com>
16570 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
16571 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
16572 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
16573 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
16574 expansion to P8V_BUILTIN_VEC_FLOAT2.
16576 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
16578 PR rtl-optimization/70023
16579 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
16580 src_regno into account.
16582 2018-02-16 Carl Love <cel@us.ibm.com>
16584 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
16585 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
16586 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
16587 * config/rs6000/rs6000.c: Remove case statements for
16588 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
16589 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
16590 and P9V_BUILTIN_VEC_VINSERT4B.
16591 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
16592 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
16593 * config/rs6000/vsx.md:
16594 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
16597 2018-02-16 Carl Love <cel@us.ibm.com>
16599 * config/rs6000/altivec.h: Add builtin names vec_extract4b
16601 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
16603 * config/rs6000/rs6000-c.c: Add the definitions for
16604 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
16605 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
16606 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
16607 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
16608 definition for insert4b and define insn *insert3b_internal.
16609 * doc/extend.texi: Add documentation for vec_extract4b.
16611 2018-02-16 Nathan Sidwell <nathan@acm.org>
16613 * doc/extend.texi (Backwards Compatibility): Mention friend
16614 injection. Note for-scope is deprecated.
16615 * doc/invoke.texi (-ffriend-injection): Deprecate.
16617 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
16619 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
16620 that moved to I2, also allow destinations that are a paradoxical
16621 subreg (instead of a normal reg).
16623 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
16626 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
16629 2018-02-16 Richard Biener <rguenther@suse.de>
16631 PR tree-optimization/84037
16632 PR tree-optimization/84016
16634 * config/i386/i386.c (ix86_builtin_vectorization_cost):
16635 Adjust vec_construct for the fact we need additional higher latency
16636 128bit inserts for AVX256 and AVX512 vector builds.
16637 (ix86_add_stmt_cost): Scale vector construction cost for
16640 2018-02-16 Richard Biener <rguenther@suse.de>
16642 PR tree-optimization/84417
16643 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
16644 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
16645 (non_rewritable_lvalue_p): Likewise, use poly-ints.
16647 2018-02-16 Martin Liska <mliska@suse.cz>
16650 * internal-fn.def (ASAN_CHECK): Set proper flags.
16651 (ASAN_MARK): Likewise.
16653 2018-02-16 Julia Koval <julia.koval@intel.com>
16655 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
16656 from PTA_CANNONLAKE.
16658 2018-02-16 Jakub Jelinek <jakub@redhat.com>
16661 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
16662 Use ++iter rather than iter++ for std::list iterators.
16663 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
16664 defer deleting them until all nodes in the forest are processed. Do
16665 free even leaf nodes. Change to_process into auto_vec.
16668 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
16669 * vec.h (vec_default_construct): Use memset instead of placement new
16670 if BROKEN_VALUE_INITIALIZATION is defined.
16671 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
16672 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
16675 PR rtl-optimization/83723
16676 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
16677 * lra.c (lra_substitute_pseudo): Likewise. If true, use
16678 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
16680 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
16682 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
16684 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
16686 PR rtl-optimization/81443
16687 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
16688 from inner REGs to paradoxical SUBREGs.
16690 2018-02-16 Richard Biener <rguenther@suse.de>
16692 PR tree-optimization/84399
16693 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
16694 For operands we can analyze at their definition make sure we can
16695 analyze them at each use as well.
16697 2018-02-16 Richard Biener <rguenther@suse.de>
16699 PR tree-optimization/84190
16700 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
16701 volatile accesses if the decl isn't volatile.
16703 2018-02-15 Jason Merrill <jason@redhat.com>
16705 PR c++/84314 - ICE with templates and fastcall attribute.
16706 * attribs.c (build_type_attribute_qual_variant): Don't clobber
16707 TYPE_CANONICAL on an existing type.
16709 2018-02-15 Jakub Jelinek <jakub@redhat.com>
16711 PR tree-optimization/84383
16712 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
16713 dstoff nor call operand_equal_p if dstbase is NULL.
16715 PR tree-optimization/84334
16716 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
16717 also a CONSTANT_CLASS_P, punt.
16719 2018-02-14 Jim Wilson <jimw@sifive.com>
16721 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
16722 first SMALL_OPERAND check. New local min_second_step. Move assert
16723 to where locals are set. Add TARGET_RVC support.
16724 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
16726 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
16728 * doc/invoke.texi: Correct -Wformat-overflow code sample.
16730 2018-02-14 Martin Sebor <msebor@redhat.com>
16732 PR tree-optimization/83698
16733 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
16734 arrays constrain the offset range to their bounds.
16735 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
16736 (builtin_access::overlap): Avoid setting the size of overlap if it's
16738 (maybe_diag_overlap): Also consider arrays when deciding what values
16739 of offsets to include in diagnostics.
16741 2018-02-14 Martin Sebor <msebor@redhat.com>
16744 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
16745 that correspond to the kind of a declaration.
16747 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
16750 * config/pa/pa.md: Load address of PIC label using the linkage table
16751 if the label is nonlocal.
16753 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
16756 warning message if user requests -maltivec=be.
16757 * doc/invoke.texi: Document deprecation of -maltivec=be.
16759 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
16762 * config/rs6000/rs6000-c.c: Update definitions for
16763 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
16764 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
16766 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
16769 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
16770 add _get_ssp intrinsics. Remove argument from
16771 __builtin_ia32_rdssp[d|q].
16772 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
16773 * config/i386/i386-builtin.def: Remove argument from
16774 __builtin_ia32_rdssp[d|q].
16775 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
16776 ix86_expand_special_args_builtin for _rdssp[d|q].
16777 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
16778 Clear register before usage.
16779 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
16780 Add documentation for new _get_ssp and _inc_ssp intrinsics.
16782 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
16784 PR tree-optimization/84357
16785 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
16786 operand 1 of an ARRAY_REF too.
16788 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
16791 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
16792 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
16794 (set_of_reg): New struct.
16795 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
16796 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
16797 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
16799 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
16800 Split into bitclr, bitset, bitinvert patterns if appropriate.
16801 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
16802 use rx_fuse_in_memory_bitop.
16803 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
16804 to named insn, correct maximum insn length.
16806 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
16809 * machmode.def: Define a complex mode for PARTIAL_INT.
16810 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
16812 * doc/rtl.texi: Document CSPImode.
16813 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
16815 (msp430_hard_regno_nregs_with_padding): Likewise.
16817 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
16820 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
16822 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
16824 PR rtl-optimization/84169
16825 * combine.c (try_combine): New variable split_i2i3. Set it to true if
16826 we generated a parallel as new i3 and we split that to new i2 and i3
16827 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
16828 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
16829 those to i2, not i1. Partially rewrite this scan code.
16831 2018-02-13 Jakub Jelinek <jakub@redhat.com>
16834 * stor-layout.c (place_field): For variable length fields, adjust
16835 offset_align afterwards not just based on the field's alignment,
16836 but also on the size.
16838 PR middle-end/84309
16839 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
16840 of exps and logs in the use_exp2 case.
16842 2018-02-13 Jeff Law <law@redhat.com>
16844 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
16845 entry for "vector".
16847 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
16850 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
16854 * common.opt (gas-loc-support, gas-locview-support): New.
16855 (ginline-points, ginternal-reset-location-views): New.
16856 * doc/invoke.texi: Document them. Use @itemx where intended.
16857 (gvariable-location-views): Adjust.
16858 * target.def (reset_location_view): New.
16859 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
16860 (TARGET_RESET_LOCATION_VIEW): New.
16861 * doc/tm.texi: Rebuilt.
16862 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
16863 (dwarf2out_default_as_locview_support): New.
16864 (output_asm_line_debug_info): Use option variables.
16865 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
16866 (output_loc_list): Likewise.
16867 (add_high_low_attributes): Check option variables.
16868 Don't output entry view attribute in strict mode.
16869 (gen_inlined_subroutine_die): Check option variables.
16870 (dwarf2out_inline_entry): Likewise.
16871 (init_sections_and_labels): Likewise.
16872 (dwarf2out_early_finish): Likewise.
16873 (maybe_reset_location_view): New, from...
16874 (dwarf2out_var_location): ... here. Call it.
16875 * debug.h (dwarf2out_default_as_loc_support): Declare.
16876 (dwarf2out_default_as_locview_support): Declare.
16877 * hooks.c (hook_int_rtx_insn_0): New.
16878 * hooks.h (hook_int_rtx_insn_0): Declare.
16879 * toplev.c (process_options): Take -gas-loc-support and
16880 -gas-locview-support from dwarf2out. Enable
16881 -gvariable-location-views by default only with locview
16882 assembler support. Enable -ginternal-reset-location-views by
16883 default only if the target defines the corresponding hook.
16884 Enable -ginline-points by default if location views are
16885 enabled; force it disabled if statement frontiers are
16887 * tree-inline.c (expand_call_inline): Check option variables.
16888 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
16890 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
16892 PR tree-optimization/84321
16893 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
16894 handling. Also check whether the anti-range contains any values
16895 that satisfy the mask; switch to a VR_RANGE if not.
16897 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
16900 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
16902 2018-02-13 Martin Jambor <mjambor@suse.cz>
16905 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
16906 of call statements, also set location of a load to a temporary.
16908 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
16910 * config/rl78/rl78.c (add_vector_labels): New function.
16911 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
16912 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
16913 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
16914 which checks that no arguments are passed.
16915 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
16916 * doc/extend.texi: Documentation for the new attribute.
16918 2018-02-13 Andreas Schwab <schwab@suse.de>
16920 * config/riscv/linux.h (CPP_SPEC): Define.
16922 2018-02-13 Jakub Jelinek <jakub@redhat.com>
16925 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
16926 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
16927 OPTION_MASK_ISA_AES as first argument to def_builtin_const
16928 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
16929 instead of OPTION_MASK_ISA_PCLMUL as first argument to
16930 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
16931 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
16932 temporarily for AES and PCLMUL builtins.
16934 PR tree-optimization/84339
16935 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
16936 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
16939 PR middle-end/84309
16940 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
16941 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
16942 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
16944 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
16946 * omp-simd-clone.h: New file.
16947 * omp-simd-clone.c: Include omp-simd-clone.h.
16948 (expand_simd_clones): No longer static.
16949 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
16950 cgraph.h and omp-simd-clone.h.
16951 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
16952 (vect_recog_widen_shift_pattern): Formatting fix.
16953 (vect_pattern_recog_1): Don't check optab for calls.
16956 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
16957 operands[2] into a REG before using gen_lowpart on it.
16959 2018-02-12 Jeff Law <law@redhat.com>
16962 * config/sh/sh.c (find_barrier): Consider a sibling call
16965 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
16966 successfully back substituting a reg.
16968 2018-02-12 Richard Biener <rguenther@suse.de>
16970 PR tree-optimization/84037
16971 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
16972 parameter, move visited init to caller.
16973 (vect_slp_analyze_operations): Separate cost from validity
16974 check, initialize visited once for all instances.
16975 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
16977 * tree-vect-stmts.c (vect_model_simple_cost): Make early
16979 (vect_model_promotion_demotion_cost): Likewise.
16980 (vectorizable_bswap): Guard cost modeling with !slp_node
16981 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
16983 (vectorizable_call): Likewise.
16984 (vectorizable_conversion): Likewise.
16985 (vectorizable_assignment): Likewise.
16986 (vectorizable_shift): Likewise.
16987 (vectorizable_operation): Likewise.
16988 (vectorizable_store): Likewise.
16989 (vectorizable_load): Likewise.
16990 (vectorizable_condition): Likewise.
16991 (vectorizable_comparison): Likewise.
16993 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
16996 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
16997 (ASAN_MARK): Fix fnspec to account for return value, change pointer
16998 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
17000 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
17002 PR middle-end/83665
17003 * params.def (inline-min-speedup): Increase from 8 to 15.
17004 (max-inline-insns-auto): Decrease from 40 to 30.
17005 * ipa-split.c (consider_split): Add some buffer for function to
17006 be considered inlining candidate.
17007 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
17010 2018-02-12 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/84037
17013 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
17014 matched stmts if we cannot swap the non-matched ones.
17016 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
17018 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
17019 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
17020 _mm_maskz_scalef_round_ss): New intrinsics.
17021 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
17022 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
17023 __builtin_ia32_scalefss_round): Remove.
17024 (__builtin_ia32_scalefsd_mask_round,
17025 __builtin_ia32_scalefss_mask_round): New intrinsics.
17026 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
17027 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
17028 ((match_operand:VF_128 2 "<round_nimm_predicate>"
17029 "<round_constraint>")): Changed to ...
17030 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
17031 "<round_scalar_constraint>")): ... this.
17032 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
17033 %0, %1, %2<round_op3>}"): Changed to ...
17034 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
17035 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
17036 %2<round_scalar_mask_op3>}"): ... this.
17037 * config/i386/subst.md (round_scalar_nimm_predicate): New.
17039 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
17041 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
17042 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
17043 (_mm_maskz_sqrt_round_ss): New intrinsics.
17044 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
17045 (__builtin_ia32_sqrtsd_mask_round)
17046 (__builtin_ia32_sqrtss_mask_round): New builtins.
17047 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
17048 (__builtin_ia32_sqrtss_round): Remove.
17049 (__builtin_ia32_sqrtsd_mask_round)
17050 (__builtin_ia32_sqrtss_mask_round): New builtins.
17051 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
17052 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
17053 ((match_operand:VF_128 1 "vector_operand"
17054 "xBm,<round_constraint>")): Changed to ...
17055 ((match_operand:VF_128 1 "vector_operand"
17056 "xBm,<round_scalar_constraint>")): ... this.
17057 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
17058 %0, %2, %<iptr>1<round_op3>}): Changed to ...
17059 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
17060 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
17061 %<iptr>1<round_scalar_mask_op3>}): ... this.
17062 ((set_attr "prefix" "<round_prefix>")): Changed to ...
17063 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
17065 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
17068 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
17069 Cast vec_cmpeq result to correct type.
17070 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
17071 Cast vec_cmpgt result to correct type.
17073 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
17075 * final.c (final_scan_insn_1): Renamed from...
17076 (final_scan_insn): ... this. New wrapper, to recover
17077 seen from the outermost call in recursive ones.
17078 * config/sparc/sparc.c (output_return): Drop seen from call.
17079 (output_sibcall): Likewise.
17080 * config/visium/visium.c (output_branch): Likewise.
17082 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
17084 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
17087 2018-02-10 Alan Modra <amodra@gmail.com>
17090 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
17091 Specify LR as an input.
17093 2018-02-10 Jakub Jelinek <jakub@redhat.com>
17096 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
17097 remove_member_access_dummy_vars): New functions.
17098 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
17099 lower_omp_1, execute_lower_omp): Use them.
17101 PR rtl-optimization/84308
17102 * shrink-wrap.c (spread_components): Release todo vector.
17104 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
17106 PR rtl-optimization/57193
17107 * ira-color.c (struct allocno_color_data): Add member
17108 conflict_allocno_hard_prefs.
17109 (update_conflict_allocno_hard_prefs): New.
17110 (bucket_allocno_compare_func): Add a preference based on
17111 conflict_allocno_hard_prefs.
17112 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
17113 (color_allocnos): Remove a dead code. Initiate
17114 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
17116 2018-02-09 Jakub Jelinek <jakub@redhat.com>
17119 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
17120 constraint from =wa to wa. Avoid a subreg on the output operand,
17121 instead use a pseudo and subreg it in a move.
17122 (p9_xxbrd_<mode>): Changed to ...
17123 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
17124 (p9_xxbrd_v2df): New expander.
17125 (p9_xxbrw_<mode>): Changed to ...
17126 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
17127 (p9_xxbrw_v4sf): New expander.
17129 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
17131 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
17133 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
17136 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
17137 multiply in 32-bit mode.
17138 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
17139 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
17142 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
17144 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
17145 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
17146 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
17147 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
17149 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
17152 * dwarf2out.c (is_trivial_indirect_ref): New function.
17153 (dwarf2out_late_global_decl): Do not generate a location
17154 attribute for variables that have a non-trivial DECL_VALUE_EXPR
17155 and that are not defined in the current unit.
17157 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
17159 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
17160 instead of a libcall for UNORDERED.
17162 2018-02-09 Tamar Christina <tamar.christina@arm.com>
17165 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
17166 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
17168 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17171 * config/s390/s390.c (s390_set_current_function): Invoke
17172 s390_indirect_branch_settings also if fndecl didn't change.
17174 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
17176 * config/rs6000/rs6000.md (blockage): Set length to zero.
17178 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
17180 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
17182 2018-02-09 Jakub Jelinek <jakub@redhat.com>
17185 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
17186 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
17190 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
17191 PARALLEL incoming that failed vt_get_decl_and_offset check.
17193 PR middle-end/84237
17194 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
17195 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
17197 (get_variable_section): For decls in named .bss* sections pass true as
17198 second argument to bss_initializer_p.
17200 2018-02-09 Marek Polacek <polacek@redhat.com>
17201 Jakub Jelinek <jakub@redhat.com>
17204 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
17205 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
17206 Sync some changes from cxx_fold_indirect_ref.
17208 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
17210 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
17212 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
17213 (BLOCK_INLINE_ENTRY_LABEL): New.
17214 (dwarf2out_var_location): Disregard inline entry markers.
17215 (inline_entry_data): New struct.
17216 (inline_entry_data_hasher): New hashtable type.
17217 (inline_entry_data_hasher::hash): New.
17218 (inline_entry_data_hasher::equal): New.
17219 (inline_entry_data_table): New variable.
17220 (add_high_low_attributes): Add DW_AT_entry_pc and
17221 DW_AT_GNU_entry_view attributes if a pending entry is found
17222 in inline_entry_data_table. Add old entry_pc attribute only
17223 if debug nonbinding markers are disabled.
17224 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
17225 markers are enabled.
17226 (block_within_block_p, dwarf2out_inline_entry): New.
17227 (dwarf2out_finish): Check that no entries remained in
17228 inline_entry_data_table.
17229 * final.c (reemit_insn_block_notes): Handle inline entry notes.
17230 (final_scan_insn, notice_source_line): Likewise.
17231 (rest_of_clean_state): Skip inline entry markers.
17232 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
17234 * gimple.c (gimple_build_debug_inline_entry): New.
17235 * gimple.h (enum gimple_debug_subcode): Add
17236 GIMPLE_DEBUG_INLINE_ENTRY.
17237 (gimple_build_debug_inline_entry): Declare.
17238 (gimple_debug_inline_entry_p): New.
17239 (gimple_debug_nonbind_marker_p): Adjust.
17240 * insn-notes.def (INLINE_ENTRY): New.
17241 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
17242 inline entry marker notes.
17243 (print_insn): Likewise.
17244 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
17245 (INSN_DEBUG_MARKER_KIND): Likewise.
17246 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
17247 * tree-inline.c (expand_call_inline): Build and insert
17248 debug_inline_entry stmt.
17249 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
17250 inline entry blocks early, if nonbind markers are enabled.
17251 (dump_scope_block): Dump fragment info.
17252 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
17253 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
17254 (gimple_build_debug_inline_entry): New.
17255 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
17256 Enable/disable inline entry points too.
17257 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
17258 (DEBUG_INSN): Describe inline entry markers.
17260 * common.opt (gvariable-location-views): New.
17261 (gvariable-location-views=incompat5): New.
17262 * config.in: Rebuilt.
17263 * configure: Rebuilt.
17264 * configure.ac: Test assembler for view support.
17265 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
17266 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
17267 * dwarf2out.c (var_loc_view): New typedef.
17268 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
17269 (dwarf2out_locviews_in_attribute): New.
17270 (dwarf2out_locviews_in_loclist): New.
17271 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
17272 (enum dw_line_info_opcode): Add LI_adv_address.
17273 (struct dw_line_info_table): Add view.
17274 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
17275 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
17276 (zero_view_p): New variable.
17277 (ZERO_VIEW_P): New macro.
17278 (output_asm_line_debug_info): New.
17279 (struct var_loc_node): Add view.
17280 (add_AT_view_list, AT_loc_list): New.
17281 (add_var_loc_to_decl): Add view param. Test it against last.
17282 (new_loc_list): Add view params. Record them.
17283 (AT_loc_list_ptr): Handle loc and view lists.
17284 (view_list_to_loc_list_val_node): New.
17285 (print_dw_val): Handle dw_val_class_view_list.
17286 (size_of_die): Likewise.
17287 (value_format): Likewise.
17288 (loc_list_has_views): New.
17289 (gen_llsym): Set vl_symbol too.
17290 (maybe_gen_llsym, skip_loc_list_entry): New.
17291 (dwarf2out_maybe_output_loclist_view_pair): New.
17292 (output_loc_list): Output view list or entries too.
17293 (output_view_list_offset): New.
17294 (output_die): Handle dw_val_class_view_list.
17295 (output_dwarf_version): New.
17296 (output_compilation_unit_header): Use it.
17297 (output_skeleton_debug_sections): Likewise.
17298 (output_rnglists, output_line_info): Likewise.
17299 (output_pubnames, output_aranges): Update version comments.
17300 (output_one_line_info_table): Output view numbers in asm comments.
17301 (dw_loc_list): Determine current endview, pass it to new_loc_list.
17302 Call maybe_gen_llsym.
17303 (loc_list_from_tree_1): Adjust.
17304 (add_AT_location_description): Create view list attribute if
17305 needed, check it's absent otherwise.
17306 (convert_cfa_to_fb_loc_list): Adjust.
17307 (maybe_emit_file): Call output_asm_line_debug_info for test.
17308 (dwarf2out_var_location): Reset views as needed. Precompute
17309 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
17310 attribute. Set view.
17311 (new_line_info_table): Reset next view.
17312 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
17313 (dwarf2out_source_line): Likewise. Output view resets and labels to
17314 the assembler, or select appropriate line info opcodes.
17315 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
17316 (optimize_string_length): Catch it. Adjust.
17317 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
17318 dw_val_class_view_list, and remove it if no longer needed.
17319 (hash_loc_list): Hash view numbers.
17320 (loc_list_hasher::equal): Compare them.
17321 (optimize_location_lists): Check whether a view list symbol is
17322 needed, and whether the locview attribute is present, and
17323 whether they match. Remove the locview attribute if no longer
17325 (index_location_lists): Call skip_loc_list_entry for test.
17326 (dwarf2out_finish): Call output_asm_line_debug_info for test.
17327 Use output_dwarf_version.
17328 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
17329 (struct dw_val_node): Add val_view_list.
17330 * final.c (SEEN_NEXT_VIEW): New.
17331 (set_next_view_needed): New.
17332 (clear_next_view_needed): New.
17333 (maybe_output_next_view): New.
17334 (final_start_function): Rename to...
17335 (final_start_function_1): ... this. Take pointer to FIRST,
17336 add SEEN parameter. Emit param bindings in the initial view.
17337 (final_start_function): Reintroduce SEEN-less interface.
17338 (final): Rename to...
17339 (final_1): ... this. Take SEEN parameter. Output final pending
17340 next view at the end.
17341 (final): Reintroduce seen-less interface.
17342 (final_scan_insn): Output pending next view before switching
17343 sections or ending a block. Mark the next view as needed when
17344 outputting variable locations. Notify debug backend of section
17345 changes, and of location view changes.
17346 (rest_of_handle_final): Adjust.
17347 * toplev.c (process_options): Autodetect value for debug variable
17348 location views option. Warn on incompat5 without -gdwarf-5.
17349 * doc/invoke.texi (gvariable-location-views): New.
17350 (gvariable-location-views=incompat5): New.
17351 (gno-variable-location-views): New.
17353 2018-02-08 David Malcolm <dmalcolm@redhat.com>
17355 PR tree-optimization/84136
17356 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
17357 that the result of find_edge is non-NULL.
17359 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
17362 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
17363 storing integer register in SImode. Fix cost of 256 and 512
17364 byte aligned SSE register store.
17366 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
17368 * config/i386/i386.c (ix86_multiplication_cost): Fix
17369 multiplication cost for TARGET_AVX512DQ.
17371 2018-02-08 Marek Polacek <polacek@redhat.com>
17373 PR tree-optimization/84238
17374 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
17377 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
17379 PR tree-optimization/84265
17380 * tree-vect-stmts.c (vectorizable_store): Don't treat
17381 VMAT_CONTIGUOUS accesses as grouped.
17382 (vectorizable_load): Likewise.
17384 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
17386 PR tree-optimization/81635
17387 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
17388 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
17389 (test_round_for_mask): New functions.
17390 (wide_int_cc_tests): Call test_round_for_mask.
17391 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
17392 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
17393 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
17394 range returned by get_range_info.
17396 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
17399 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
17400 * symtab.c: Include builtins.h
17401 (symtab_node::output_to_lto_symbol_table_p): Move here
17402 from lto-streamer-out.c:output_symbol_p.
17403 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
17404 (output_symbol_p): Move all logic to symtab.c
17405 (produce_symtab): Update.
17407 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17409 * config/s390/s390-opts.h (enum indirect_branch): Define.
17410 * config/s390/s390-protos.h (s390_return_addr_from_memory)
17411 (s390_indirect_branch_via_thunk)
17412 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
17413 (enum s390_indirect_branch_type): Define.
17414 * config/s390/s390.c (struct s390_frame_layout, struct
17415 machine_function): Remove.
17416 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
17417 (indirect_branch_table_label_no, indirect_branch_table_name):
17419 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
17420 (enum s390_indirect_branch_option): Define.
17421 (s390_return_addr_from_memory): New function.
17422 (s390_handle_string_attribute): New function.
17423 (s390_attribute_table): Add new attribute handler.
17424 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
17425 (s390_indirect_branch_via_thunk): New function.
17426 (s390_indirect_branch_via_inline_thunk): New function.
17427 (s390_function_ok_for_sibcall): When jumping via thunk disallow
17428 sibling call optimization for non z10 compiles.
17429 (s390_emit_call): Force indirect branch target to be a single
17430 register. Add r1 clobber for non-z10 compiles.
17431 (s390_emit_epilogue): Emit return jump via return_use expander.
17432 (s390_reorg): Handle JUMP_INSNs as execute targets.
17433 (s390_option_override_internal): Perform validity checks for the
17434 new command line options.
17435 (s390_indirect_branch_attrvalue): New function.
17436 (s390_indirect_branch_settings): New function.
17437 (s390_set_current_function): Invoke s390_indirect_branch_settings.
17438 (s390_output_indirect_thunk_function): New function.
17439 (s390_code_end): Implement target hook.
17440 (s390_case_values_threshold): Implement target hook.
17441 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
17443 * config/s390/s390.h (struct s390_frame_layout)
17444 (struct machine_function): Move here from s390.c.
17445 (TARGET_INDIRECT_BRANCH_NOBP_RET)
17446 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
17447 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
17448 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
17449 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
17450 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
17451 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
17452 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
17453 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
17454 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
17455 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
17456 (mnemonic attribute): Add values which aren't recognized
17458 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
17459 pattern for branch conversion. Fix mnemonic attribute.
17460 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
17461 indirect branch via thunk if requested.
17462 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
17463 ("*indirect_jump"): Disable for branch conversion using out of
17465 ("indirect_jump_via_thunk<mode>_z10")
17466 ("indirect_jump_via_thunk<mode>")
17467 ("indirect_jump_via_inlinethunk<mode>_z10")
17468 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
17469 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
17470 ("casesi_jump_via_inlinethunk<mode>_z10")
17471 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
17472 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
17473 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
17474 ("*indirect2_jump"): Disable for branch conversion.
17475 ("casesi_jump"): Turn into expander and expand patterns for branch
17477 ("return_use"): New expander.
17478 ("*return"): Emit return via thunk and rename it to ...
17479 ("*return<mode>"): ... this one.
17480 * config/s390/s390.opt: Add new options and and enum for the
17483 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
17485 * lra-constraints.c (match_reload): Unconditionally use
17486 gen_lowpart_SUBREG, rather than selecting between that
17487 and equivalent gen_rtx_SUBREG code.
17489 2018-02-08 Richard Biener <rguenther@suse.de>
17491 PR tree-optimization/84233
17492 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
17493 changed flag instead of boguously re-using phi_inserted.
17495 2018-02-08 Martin Jambor <mjambor@suse.cz>
17497 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
17498 static local variables.
17500 2018-02-08 Richard Biener <rguenther@suse.de>
17502 PR tree-optimization/84278
17503 * tree-vect-stmts.c (vectorizable_store): When looking for
17504 smaller vector types to perform grouped strided loads/stores
17505 make sure the mode is supported by the target.
17506 (vectorizable_load): Likewise.
17508 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
17510 * config/aarch64/aarch64.c (aarch64_components_for_bb):
17511 Increase LDP/STP opportunities by adding adjacent callee-saves.
17513 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
17515 PR rtl-optimization/84068
17516 PR rtl-optimization/83459
17517 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
17519 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
17521 PR tree-optimization/84224
17522 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
17523 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
17524 non-zero arguments.
17526 2018-02-07 Iain Sandoe <iain@codesourcery.com>
17529 * config/rs6000/altivec.md (*restore_world): Remove LR use.
17530 * config/rs6000/predicates.md (restore_world_operation): Adjust op
17531 count, remove one USE.
17533 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17535 * doc/install.texi (Configuration): Document the
17536 --with-long-double-format={ibm,ieee} PowerPC configuration
17540 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
17541 Convert from define_expand to be define_insn_and_split. Rework
17542 float/double/_Float128 conversions to QI/HI/SImode to work with
17543 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
17544 conversions to QI/HImode types did a store and then a load to
17545 truncate the value. For conversions to VSX registers, don't split
17546 the insn, instead emit the code directly. Use the code iterator
17547 any_fix to combine signed and unsigned conversions.
17548 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
17549 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
17550 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
17551 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
17552 (fix_<mode>di2_hw): Likewise.
17553 (fixuns_<mode>di2_hw): Likewise.
17554 (fix_<mode>si2_hw): Likewise.
17555 (fixuns_<mode>si2_hw): Likewise.
17556 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
17557 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
17558 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
17559 fix<uns>_trunc<SFDF:mode>si2_p8.
17560 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
17561 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17562 (fix<uns>_<mode>_mem): Likewise.
17563 (fctiw<u>z_<mode>_mem): Likewise.
17564 (fix<uns>_<mode>_mem): Likewise.
17565 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
17566 the register allocator from doing a direct move to the GPRs to do
17567 a store, and instead use the ISA 3.0 store byte/half-word from
17568 vector register instruction. For IEEE 128-bit floating point,
17569 also optimize stores of 32-bit ints.
17570 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
17572 2018-02-07 Alan Hayward <alan.hayward@arm.com>
17574 * genextract.c (push_pathstr_operand): New function to support
17576 (walk_rtx): Call push_pathstr_operand.
17577 (print_path): Support [a-zA-Z].
17579 2018-02-07 Richard Biener <rguenther@suse.de>
17581 PR tree-optimization/84037
17582 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
17583 (cse_and_gimplify_to_preheader): Declare.
17584 (vect_get_place_in_interleaving_chain): Likewise.
17585 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
17587 (_loop_vec_info::~_loop_vec_info): Delete it.
17588 (cse_and_gimplify_to_preheader): New function.
17589 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
17590 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
17591 (vectorizable_load): Likewise. For grouped stores always base
17592 the IV on the first element.
17593 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
17594 condition before gimplifying.
17596 2018-02-07 Jakub Jelinek <jakub@redhat.com>
17598 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
17599 *DIV_EXPR and *MOD_EXPR.
17601 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
17604 * config/i386/i386.c (ix86_option_override_internal): Mask out
17605 the CF_SET bit when checking -fcf-protection.
17607 2018-02-07 Tom de Vries <tom@codesourcery.com>
17610 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
17613 2018-02-07 Richard Biener <rguenther@suse.de>
17615 PR tree-optimization/84204
17616 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
17619 PR tree-optimization/84205
17620 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
17621 special-case isl_ast_op_zdiv_r.
17623 PR tree-optimization/84223
17624 * graphite-scop-detection.c (gather_bbs::before_dom_children):
17625 Only add conditions from within the region.
17626 (gather_bbs::after_dom_children): Adjust.
17628 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
17631 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
17632 * config/avr/avr.md: Only post-reload split REG-REG moves if
17633 either register is GENERAL_REG_P.
17635 2018-02-07 Jakub Jelinek <jakub@redhat.com>
17637 PR tree-optimization/84235
17638 * tree-ssa-scopedtables.c
17639 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
17640 if the subtraction is performed in floating point type where NaNs are
17641 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
17642 build 1. Formatting fix.
17644 2018-02-06 Jakub Jelinek <jakub@redhat.com>
17647 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
17648 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
17649 and skip it regardless of bb boundaries. Use CALL_P macro,
17650 don't test INSN_P (insn) together with CALL_P or JUMP_P check
17651 unnecessarily, formatting fix.
17653 2018-02-06 Michael Collison <michael.collison@arm.com>
17655 * config/arm/thumb2.md:
17656 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
17657 (*thumb_mov_notscc): Ditto.
17659 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
17662 * config/rs6000/rs6000.md (su code attribute): Use "u" for
17663 unsigned_fix, not "s".
17665 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17667 * configure.ac (gcc_fn_eh_frame_ro): New function.
17668 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
17669 correct .eh_frame permissions.
17670 * configure: Regenerate.
17672 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
17674 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
17675 irrelevant options.
17677 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17679 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17680 Display warning message for -mno-speculate-indirect-jumps.
17682 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
17684 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
17686 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
17688 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
17690 PR tree-optimization/84225
17691 * tree-eh.c (find_trapping_overflow): Only call
17692 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
17694 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
17697 * config/i386/i386.c: Reimplement the check of possible options
17698 -mibt/-mshstk conbination. Change error messages.
17699 * doc/invoke.texi: Fix a typo: remove extra '='.
17701 2018-02-06 Marek Polacek <polacek@redhat.com>
17703 PR tree-optimization/84228
17704 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
17706 2018-02-06 Tamar Christina <tamar.christina@arm.com>
17709 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
17710 emitted arch directives.
17711 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
17712 __ARM_FEATURE_COPROC before changing architectures.
17714 2018-02-06 Richard Biener <rguenther@suse.de>
17716 * config/i386/i386.c (print_reg): Fix typo.
17717 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
17719 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
17721 * configure: Regenerate.
17723 2018-02-05 Martin Sebor <msebor@redhat.com>
17725 PR tree-optimization/83369
17726 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
17729 2018-02-05 Martin Liska <mliska@suse.cz>
17731 * doc/invoke.texi: Cherry-pick upstream r323995.
17733 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
17735 * ira.c (ira_init_register_move_cost): Adjust comment.
17737 2018-02-05 Martin Liska <mliska@suse.cz>
17739 PR gcov-profile/84137
17740 * doc/gcov.texi: Fix typo in documentation.
17742 2018-02-05 Martin Liska <mliska@suse.cz>
17744 PR gcov-profile/83879
17745 * doc/gcov.texi: Document necessity of --dynamic-list-data when
17746 using dlopen functionality.
17748 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
17750 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
17751 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
17752 _mm_maskz_range_ss, _mm_mask_range_round_ss,
17753 _mm_maskz_range_round_ss): New intrinsics.
17754 (__builtin_ia32_rangesd128_round)
17755 (__builtin_ia32_rangess128_round): Remove.
17756 (__builtin_ia32_rangesd128_mask_round,
17757 __builtin_ia32_rangess128_mask_round): New builtins.
17758 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
17759 __builtin_ia32_rangess128_round): Remove.
17760 (__builtin_ia32_rangesd128_mask_round,
17761 __builtin_ia32_rangess128_mask_round): New builtins.
17762 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
17763 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
17764 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
17765 "<round_saeonly_constraint>")): Changed to ...
17766 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
17767 "<round_saeonly_scalar_constraint>")): ... this.
17768 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17769 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
17770 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
17771 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
17772 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
17774 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
17776 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
17778 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
17779 Remove all values except native, 8540 and 8548.
17781 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
17783 * config/i386/i386.c (ix86_output_function_return): Pass
17784 INVALID_REGNUM, instead of -1, as invalid register number to
17785 indirect_thunk_name and output_indirect_thunk.
17787 2018-02-02 Julia Koval <julia.koval@intel.com>
17789 * config.gcc: Add -march=icelake.
17790 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
17791 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
17792 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
17793 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
17794 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
17795 (processor_target_table): Add icelake.
17796 (ix86_option_override_internal): Handle new PTAs.
17797 (get_builtin_code_for_version): Handle icelake.
17798 (M_INTEL_COREI7_ICELAKE): New.
17799 (fold_builtin_cpu): Handle icelake.
17800 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
17801 * doc/invoke.texi: Add -march=icelake.
17803 2018-02-02 Julia Koval <julia.koval@intel.com>
17805 * config/i386/i386.c (ix86_option_override_internal): Change flags type
17806 to wide_int_bitmask.
17807 * wide-int-bitmask.h: New.
17809 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
17812 * config/i386/i386.md: Replace Pmode with word_mode in
17813 builtin_setjmp_setup and builtin_longjmp to support x32.
17815 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
17819 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
17821 (rs6000_supported_cpu_names): New static variable.
17822 (linux_cpu_translation_table): Likewise.
17823 (elf_platform) <cpu>: Define new static variable and use it.
17824 Translate kernel AT_PLATFORM name to canonical name if needed.
17825 Error if platform name is unknown.
17827 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
17830 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
17832 2018-02-01 Jeff Law <law@redhat.com>
17835 * config/i386/i386.c (release_scratch_register_on_entry): Add new
17836 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
17837 the scratch if RELEASE_VIA_POP is false.
17838 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
17839 If we have to save a temporary register, decrement SIZE appropriately.
17840 Pass new arguments to release_scratch_register_on_entry.
17841 (ix86_adjust_stack_and_probe): Likewise.
17842 (ix86_emit_probe_stack_range): Pass new arguments to
17843 release_scratch_register_on_entry.
17845 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
17847 PR rtl-optimization/84157
17848 * combine.c (change_zero_ext): Use REG_P predicate in
17849 front of HARD_REGISTER_P predicate.
17851 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
17853 * config/avr/avr.c (avr_option_override): Move disabling of
17854 -fdelete-null-pointer-checks to...
17855 * common/config/avr/avr-common.c (avr_option_optimization_table):
17858 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17860 PR tree-optimization/81635
17861 * tree-data-ref.c (split_constant_offset_1): For types that
17862 wrap on overflow, try to use range info to prove that wrapping
17865 2018-02-01 Renlin Li <renlin.li@arm.com>
17868 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
17869 TAILCALL_ADDR_REGS.
17870 (aarch64_register_move_cost): Likewise.
17871 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
17872 TAILCALL_ADDR_REGS.
17873 (REG_CLASS_NAMES): Likewise.
17874 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
17875 TAILCALL_ADDR_REGS. Remove IP registers.
17876 * config/aarch64/aarch64.md (Ucs): Update register constraint.
17878 2018-02-01 Richard Biener <rguenther@suse.de>
17880 * domwalk.h (dom_walker::dom_walker): Add additional constructor
17881 for specifying RPO order and allow NULL for that.
17882 * domwalk.c (dom_walker::dom_walker): Likewise.
17883 (dom_walker::walk): Handle NULL RPO order.
17884 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
17886 (rewrite_update_dom_walker): Likewise.
17887 (mark_def_dom_walker): Likewise.
17889 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17891 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
17892 (aarch64_maybe_expand_sve_subreg_move): Declare.
17893 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
17894 * config/aarch64/predicates.md (aarch64_any_register_operand): New
17896 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
17897 that are semantically a reverse operation.
17898 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
17899 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
17900 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
17902 (aarch64_can_change_mode_class): For big-endian, forbid changes
17903 between two SVE modes if they have different element sizes.
17905 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17907 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
17908 the TImode handling for big-endian targets.
17910 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17912 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
17913 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
17915 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
17916 Remove BSWAP handing for big-endian targets and use the form of
17917 LD1RQ appropariate for the mode.
17919 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17921 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
17922 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
17923 duplicated element.
17925 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
17928 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
17929 check for operands that need to go through aarch64_sve_reload_be.
17931 2018-02-01 Jakub Jelinek <jakub@redhat.com>
17933 PR tree-optimization/81661
17934 PR tree-optimization/84117
17935 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
17936 * tree-eh.c: Include gimplify.h.
17937 (find_trapping_overflow, replace_trapping_overflow,
17938 rewrite_to_non_trapping_overflow): New functions.
17939 * tree-vect-loop.c: Include tree-eh.h.
17940 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
17941 * tree-data-ref.c: Include tree-eh.h.
17942 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
17944 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
17946 PR rtl-optimization/84123
17947 * combine.c (change_zero_ext): Check if hard register satisfies
17948 can_change_dest_mode before calling gen_lowpart_SUBREG.
17950 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
17953 * ira.c (ira_init_register_move_cost): Remove assert.
17955 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
17957 PR rtl-optimization/84071
17958 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
17959 * doc/tm.texi: Regenerate.
17961 2018-01-31 Richard Biener <rguenther@suse.de>
17963 PR tree-optimization/84132
17964 * tree-data-ref.c (analyze_miv_subscript): Properly
17965 check whether evolution_function_is_affine_multivariate_p
17966 before calling gcd_of_steps_may_divide_p.
17968 2018-01-31 Julia Koval <julia.koval@intel.com>
17971 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
17972 * config/i386/i386.md (rdpid_rex64) New.
17973 (rdpid): Make 32bit only.
17975 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
17978 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
17979 an IDENTIFIER_NODE for FUNCTION_TYPE's.
17981 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
17984 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
17986 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
17988 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
17990 PR rtl-optimization/84071
17991 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
17992 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
17994 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
17996 * config/arc/arc.c (arc_handle_aux_attribute): New function.
17997 (arc_attribute_table): Add 'aux' attribute.
17998 (arc_in_small_data_p): Consider aux like variables.
17999 (arc_is_aux_reg_p): New function.
18000 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
18001 (arc_get_aux_arg): New function.
18002 (prepare_move_operands): Handle aux-register access.
18003 (arc_handle_aux_attribute): New function.
18004 * doc/extend.texi (ARC Variable attributes): Add subsection.
18006 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
18008 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
18009 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
18010 (arc_attribute_table): Add 'uncached' attribute.
18011 (arc_print_operand): Print '.di' flag for uncached memory
18013 (arc_in_small_data_p): Do not consider for small data the uncached
18015 (arc_is_uncached_mem_p): New function.
18016 * config/arc/predicates.md (compact_store_memory_operand): Check
18017 for uncached memory accesses.
18018 (nonvol_nonimm_operand): Likewise.
18019 * doc/extend.texi (ARC Type Attribute): New subsection.
18021 2018-01-31 Jakub Jelinek <jakub@redhat.com>
18024 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
18025 falign-loops=): Add Optimization flag.
18027 2018-01-30 Jeff Law <law@redhat.com>
18030 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
18031 INT_REGISTERS_SAVED. Check it prior to calling
18032 get_scratch_register_on_entry.
18033 (ix86_adjust_stack_and_probe): Similarly.
18034 (ix86_emit_probe_stack_range): Similarly.
18035 (ix86_expand_prologue): Corresponding changes.
18037 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18040 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
18041 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
18043 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
18046 * lra-constraints.c (curr_insn_transform): Process AND in the
18049 2018-01-30 Jakub Jelinek <jakub@redhat.com>
18051 PR rtl-optimization/83986
18052 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
18053 dependence against last_pending_memory_flush in addition to
18054 pending_jump_insns.
18056 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
18058 PR tree-optimization/81611
18059 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
18062 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18065 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
18068 2018-01-30 Richard Biener <rguenther@suse.de>
18069 Jakub Jelinek <jakub@redhat.com>
18071 PR tree-optimization/84111
18072 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
18073 inner loops added during recursion, as they don't have up-to-date
18076 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
18079 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
18080 (can_inline_edge_by_limits_p): ... here.
18081 (can_early_inline_edge_p, check_callers,
18082 update_caller_keys, update_callee_keys, recursive_inlining,
18083 add_new_edges_to_heap, speculation_useful_p,
18084 inline_small_functions,
18085 inline_small_functions, flatten_function,
18086 inline_to_all_callers_1): Update.
18088 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
18090 * profile-count.c (profile_count::combine_with_ipa_count): Handle
18093 2018-01-30 Richard Biener <rguenther@suse.de>
18095 PR tree-optimization/83008
18096 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
18097 invariant and constant vector uses in stmts when they need
18098 more than one stmt.
18100 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18103 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
18104 * configure: Regenerate.
18106 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18108 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
18110 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
18111 Use gen_rtx_REG rather than gen_lowpart.
18113 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18115 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
18116 rather than 0 when creating partial subregs.
18118 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18120 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
18123 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18126 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
18127 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
18128 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
18129 flags. This restores the settings used before the 2017-07-24.
18130 Turning off pre increment/decrement/modify allows IVOPTS to
18131 optimize DF/SF loops where the index is an int.
18133 2018-01-29 Richard Biener <rguenther@suse.de>
18134 Kelvin Nilsen <kelvin@gcc.gnu.org>
18137 * tree-vect-stmts.c (vectorizable_call): Don't call
18138 targetm.vectorize_builtin_md_vectorized_function if callee is
18141 2018-01-22 Carl Love <cel@us.ibm.com>
18143 * doc/extend.tex: Fix typo in second arg in
18144 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
18146 2018-01-29 Richard Biener <rguenther@suse.de>
18148 PR tree-optimization/84086
18149 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
18150 (flush_ssaname_freelist): When SSA names were released reset
18151 the SCEV hash table.
18153 2018-01-29 Richard Biener <rguenther@suse.de>
18155 PR tree-optimization/84057
18156 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
18157 removed paths when removing edges.
18159 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
18161 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
18162 -mfunction-return=@var{choice}.
18164 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
18166 PR diagnostic/84034
18167 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
18168 Handle CR like TAB.
18169 (layout::print_source_line): Likewise.
18170 (test_get_line_width_without_trailing_whitespace): Add test cases.
18172 2018-01-27 Jakub Jelinek <jakub@redhat.com>
18174 PR middle-end/84040
18175 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
18178 2018-01-26 Jim Wilson <jimw@sifive.com>
18180 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
18182 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
18185 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18187 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
18188 and CMP + SUB-immediate -> SUBS.
18190 2018-01-26 Martin Sebor <msebor@redhat.com>
18192 PR tree-optimization/83896
18193 * tree-ssa-strlen.c (get_string_len): Rename...
18194 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
18195 Avoid assuming length is constant.
18196 (handle_char_store): Use HOST_WIDE_INT for string length.
18198 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
18201 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
18202 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
18204 2018-01-26 Richard Biener <rguenther@suse.de>
18206 PR rtl-optimization/84003
18207 * dse.c (record_store): Only record redundant stores when
18208 the earlier store aliases at least all accesses the later one does.
18210 2018-01-26 Jakub Jelinek <jakub@redhat.com>
18212 PR rtl-optimization/83985
18213 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
18214 REG_CFA_RESTORE insns.
18215 (delete_unmarked_insns): Don't ignore separate shrink wrapping
18216 REG_CFA_RESTORE insns here.
18219 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
18220 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
18222 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18224 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
18225 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
18226 (arc_init): Likewise.
18227 (arc_override_options): Likewise.
18228 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
18230 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
18232 * config/arc/arc.h (TARGET_DBNZ): Define.
18233 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
18234 properly set the tune attribute.
18235 (dbnz): Use TARGET_DBNZ guard.
18236 * config/arc/arc.opt (mtune): Add core3 option.
18238 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18240 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
18241 recognize new pic like addresses.
18242 (arc_delegitimize_address): Clean up.
18244 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18246 * config/arc/arc-arches.def: Option mrf16 valid for all
18248 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
18249 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
18250 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
18251 * config/arc/arc-tables.opt: Regenerate.
18252 * config/arc/arc.c (arc_conditional_register_usage): Handle
18253 reduced register file case.
18254 (arc_file_start): Set must have build attributes.
18255 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
18256 mrf16 option value.
18257 * config/arc/arc.opt (mrf16): Add new option.
18258 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
18259 * config/arc/genmultilib.awk: Handle new mrf16 option.
18260 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
18261 * config/arc/t-multilib: Regenerate.
18262 * doc/invoke.texi (ARC Options): Document mrf16 option.
18264 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18266 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
18267 * config/arc/arc.c (arc_handle_secure_attribute): New function.
18268 (arc_attribute_table): Add 'secure_call' attribute.
18269 (arc_print_operand): Print secure call operand.
18270 (arc_function_ok_for_sibcall): Don't optimize tail calls when
18272 (arc_is_secure_call_p): New function. * config/arc/arc.md
18273 (call_i): Add support for sjli instruction.
18274 (call_value_i): Likewise.
18275 * config/arc/constraints.md (Csc): New constraint.
18277 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18278 John Eric Martin <John.Martin@emmicro-us.com>
18280 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
18281 * config/arc/arc.c (_arc_jli_section): New struct.
18282 (arc_jli_section): New type.
18283 (rc_jli_sections): New static variable.
18284 (arc_handle_jli_attribute): New function.
18285 (arc_attribute_table): Add jli_always and jli_fixed attribute.
18286 (arc_file_end): New function.
18287 (TARGET_ASM_FILE_END): Define.
18288 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
18289 (arc_add_jli_section): New function.
18290 (jli_call_scan): Likewise.
18291 (arc_reorg): Call jli_call_scan.
18292 (arc_output_addsi): Remove 'S' from printing asm operand.
18293 (arc_is_jli_call_p): New function.
18294 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
18296 (movhi_insn): Likewise.
18297 (movsi_insn): Likewise.
18298 (movsi_set_cc_insn): Likewise.
18299 (loadqi_update): Likewise.
18300 (load_zeroextendqisi_update): Likewise.
18301 (load_signextendqisi_update): Likewise.
18302 (loadhi_update): Likewise.
18303 (load_zeroextendhisi_update): Likewise.
18304 (load_signextendhisi_update): Likewise.
18305 (loadsi_update): Likewise.
18306 (loadsf_update): Likewise.
18307 (movsicc_insn): Likewise.
18308 (bset_insn): Likewise.
18309 (bxor_insn): Likewise.
18310 (bclr_insn): Likewise.
18311 (bmsk_insn): Likewise.
18312 (bicsi3_insn): Likewise.
18313 (cmpsi_cc_c_insn): Likewise.
18314 (movsi_ne): Likewise.
18315 (movsi_cond_exec): Likewise.
18316 (clrsbsi2): Likewise.
18317 (norm_f): Likewise.
18327 (call_i): Remove 'S' asm letter, add jli instruction.
18328 (call_value_i): Likewise.
18329 * config/arc/arc.op (mjli-always): New option.
18330 * config/arc/constraints.md (Cji): New constraint.
18331 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
18333 (subsf3_fpx): Likewise.
18334 (mulsf3_fpx): Likewise.
18335 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
18337 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
18338 function attrbutes.
18339 * doc/invoke.texi (ARC): Document mjli-always option.
18341 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
18343 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
18344 avoid addition with 0 and use incw and decw where possible.
18346 2018-01-26 Richard Biener <rguenther@suse.de>
18348 PR tree-optimization/81082
18349 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
18350 association if it requires casting to unsigned.
18351 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
18352 from fold_plusminus_mult_expr to catch important cases late when
18353 range info is available.
18355 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18357 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
18358 * configure.ac (hidden_linkonce): New test.
18359 * configure: Regenerate.
18360 * config.in: Regenerate.
18362 2018-01-26 Julia Koval <julia.koval@intel.com>
18364 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
18365 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
18366 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
18367 _mm_mask_bitshuffle_epi64_mask): Fix type.
18368 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
18369 USI_FTYPE_V4DI_V4DI_USI): Remove.
18370 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
18371 __builtin_ia32_vpshufbitqmb256_mask,
18372 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
18373 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
18374 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
18376 2018-01-26 Alan Modra <amodra@gmail.com>
18379 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
18380 UNSPEC_VBPERMQ. Sort other unspecs.
18382 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
18384 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
18386 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
18388 PR middle-end/83055
18389 * predict.c (drop_profile): Do not push/pop cfun; update also
18391 (handle_missing_profiles): Fix logic looking for zero profiles.
18393 2018-01-25 Jakub Jelinek <jakub@redhat.com>
18395 PR middle-end/83977
18396 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
18397 on functions with #pragma omp declare simd or functions with simd
18399 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
18400 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
18401 Remove trailing \n from warning_at calls.
18403 2018-01-25 Tom de Vries <tom@codesourcery.com>
18406 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
18407 for neutered workers.
18409 2018-01-24 Joseph Myers <joseph@codesourcery.com>
18412 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
18413 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
18415 2018-01-24 Jeff Law <law@redhat.com>
18418 * i386.c (get_probe_interval): Move to earlier point.
18419 (ix86_compute_frame_layout): If -fstack-clash-protection and
18420 the frame is larger than the probe interval, then use pushes
18421 to save registers rather than reg->mem moves.
18422 (ix86_expand_prologue): Remove conditional for int_registers_saved
18425 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
18428 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
18429 min/max for never referenced object.
18431 2018-01-24 Jakub Jelinek <jakub@redhat.com>
18433 PR middle-end/83977
18434 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
18436 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
18437 attributes from DECL_ATTRIBUTES (decl) without affecting
18438 DECL_ATTRIBUTES (current_function_decl).
18439 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
18440 functions with non-NULL DECL_ABSTRACT_ORIGIN.
18442 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
18444 PR tree-optimization/83979
18445 * fold-const.c (fold_comparison): Use constant_boolean_node
18446 instead of boolean_{true,false}_node.
18448 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
18450 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
18453 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18455 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
18456 Simplify the clause that sets the length attribute.
18457 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
18458 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
18459 clause that sets the length attribute.
18460 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
18462 2018-01-24 Tom de Vries <tom@codesourcery.com>
18465 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
18466 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
18467 Add strict parameter.
18468 (prevent_branch_around_nothing): Insert dummy insn between branch to
18469 label and label with no ptx insn inbetween.
18470 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
18472 2018-01-24 Tom de Vries <tom@codesourcery.com>
18475 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
18476 for neutered threads in warp.
18477 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
18479 2018-01-24 Richard Biener <rguenther@suse.de>
18481 PR tree-optimization/83176
18482 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
18485 2018-01-24 Richard Biener <rguenther@suse.de>
18487 PR tree-optimization/82819
18488 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
18489 code generating pluses that are no-ops in the target precision.
18491 2018-01-24 Richard Biener <rguenther@suse.de>
18493 PR middle-end/84000
18494 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
18496 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
18498 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
18499 to merge probabilities.
18500 * predict.c (probably_never_executed): Also mark as cold functions
18501 with global 0 profile and guessed local profile.
18502 * profile-count.c (profile_probability::combine_with_count): New
18504 * profile-count.h (profile_probability::operator*,
18505 profile_probability::operator*=, profile_probability::operator/,
18506 profile_probability::operator/=): Reduce precision to adjusted
18507 and set value to guessed on contradictory divisions.
18508 (profile_probability::combine_with_freq): Remove.
18509 (profile_probability::combine_wiht_count): Declare.
18510 (profile_count::force_nonzero):: Set to adjusted.
18511 (profile_count::probability_in):: Set quality to adjusted.
18512 * tree-ssa-tail-merge.c (replace_block_by): Use
18513 combine_with_count.
18515 2018-01-23 Andrew Waterman <andrew@sifive.com>
18516 Jim Wilson <jimw@sifive.com>
18518 * config/riscv/riscv.c (riscv_stack_boundary): New.
18519 (riscv_option_override): Set riscv_stack_boundary. Handle
18520 riscv_preferred_stack_boundary_arg.
18521 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
18522 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
18523 (STACK_BOUNDARY): Set to riscv_stack_boundary.
18524 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
18525 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
18526 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
18528 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
18531 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
18532 of struct ix86_frame.
18533 (ix86_expand_epilogue): Likewise. Add a local variable for
18534 the reg_save_offset field in struct ix86_frame.
18536 2018-01-23 Bin Cheng <bin.cheng@arm.com>
18538 PR tree-optimization/82604
18539 * tree-loop-distribution.c (enum partition_kind): New enum item
18540 PKIND_PARTIAL_MEMSET.
18541 (partition_builtin_p): Support above new enum item.
18542 (generate_code_for_partition): Ditto.
18543 (compute_access_range): Differentiate cases that equality can be
18544 proven at all loops, the innermost loops or no loops.
18545 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
18546 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
18547 (finalize_partitions, distribute_loop): Don't fuse partition of
18548 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
18549 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
18550 parloop is enabled.
18552 2018-01-23 Martin Liska <mliska@suse.cz>
18554 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
18555 order to ignore the predictor.
18556 (PRED_POLYMORPHIC_CALL): Likewise.
18557 (PRED_RECURSIVE_CALL): Likewise.
18559 2018-01-23 Martin Liska <mliska@suse.cz>
18561 * tree-profile.c (tree_profiling): Print function header to
18562 aware reader which function we are working on.
18563 * value-prof.c (gimple_find_values_to_profile): Do not print
18564 not interesting value histograms.
18566 2018-01-23 Martin Liska <mliska@suse.cz>
18568 * profile-count.h (enum profile_quality): Add
18569 profile_uninitialized as the first value. Do not number values
18570 as they are zero based.
18571 (profile_count::verify): Update sanity check.
18572 (profile_probability::verify): Likewise.
18574 2018-01-23 Nathan Sidwell <nathan@acm.org>
18576 * doc/invoke.texi (ffor-scope): Deprecate.
18578 2018-01-23 David Malcolm <dmalcolm@redhat.com>
18580 PR tree-optimization/83510
18581 * domwalk.c (set_all_edges_as_executable): New function.
18582 (dom_walker::dom_walker): Convert bool param
18583 "skip_unreachable_blocks" to enum reachability. Move setup of
18584 edge flags to set_all_edges_as_executable and only do it when
18585 reachability is REACHABLE_BLOCKS.
18586 * domwalk.h (enum dom_walker::reachability): New enum.
18587 (dom_walker::dom_walker): Convert bool param
18588 "skip_unreachable_blocks" to enum reachability.
18589 (set_all_edges_as_executable): New decl.
18590 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
18591 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
18593 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
18594 but converting true to REACHABLE_BLOCKS.
18595 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
18597 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
18598 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
18599 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
18601 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
18602 if check_all_array_refs will be called.
18604 2018-01-23 David Malcolm <dmalcolm@redhat.com>
18606 * tree.c (selftest::test_location_wrappers): Add more test
18609 2018-01-23 David Malcolm <dmalcolm@redhat.com>
18611 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
18612 (selftest::test_bit_in_range): Likewise.
18614 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
18617 * doc/sourcebuild.texi (vect_float): Say that the selector
18618 only describes the situation when -funsafe-math-optimizations is on.
18619 (vect_float_strict): Document.
18621 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
18623 PR tree-optimization/83965
18624 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
18625 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
18626 instead of checking only for a reduction.
18627 (vect_recog_widen_sum_pattern): Likewise.
18629 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
18631 * predict.c (probably_never_executed): Only use precise profile info.
18632 (compute_function_frequency): Skip after inlining hack since we now
18633 have quality checking.
18635 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
18637 * profile-count.h (profile_probability::very_unlikely,
18638 profile_probability::unlikely, profile_probability::even): Set
18639 precision to guessed.
18641 2018-01-23 Richard Biener <rguenther@suse.de>
18643 PR tree-optimization/83963
18644 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
18645 Properly terminate dominator walk when crossing the exit edge not
18646 when visiting its source block.
18648 2018-01-23 Jakub Jelinek <jakub@redhat.com>
18651 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
18652 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
18654 2018-01-22 Jakub Jelinek <jakub@redhat.com>
18656 PR tree-optimization/83957
18657 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
18658 semicolon after for body surrounded by braces.
18660 PR tree-optimization/83081
18661 * profile-count.h (profile_probability::split): New method.
18662 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
18663 Use profile_probability::split.
18664 (do_compare_rtx_and_jump): Fix adjustment of probabilities
18665 when splitting a single conditional jump into 2.
18667 2018-01-22 David Malcolm <dmalcolm@redhat.com>
18669 PR tree-optimization/69452
18670 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
18673 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18675 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
18676 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
18677 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
18679 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18681 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
18683 * config/rl78/rl78.md (movdi): New define_expand.
18684 * config/rl78/rl78.c (rl78_split_movdi): New function.
18686 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18689 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
18691 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
18692 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
18693 128-bit to produce an UNSPEC move to get the double word with the
18694 signbit and then a shift directly to do signbit.
18695 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
18696 implementation with a new version that just does either a direct
18697 move or a regular move. Move memory interface to separate insns.
18698 Move insns so they are next to the expander.
18699 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
18700 with signbit move. Split big and little endian case.
18701 (signbit<mode>2_dm_mem_le): Likewise.
18702 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
18703 (signbit<mode>2_dm2): Likewise.
18705 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18707 * config/rl78/rl78.md (anddi3): New define_expand.
18709 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18711 * config/rl78/rl78.md (umindi3): New define_expand.
18713 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18715 * config/rl78/rl78.md (smindi3): New define_expand.
18717 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18719 * config/rl78/rl78.md (smaxdi3): New define_expand.
18721 2018-01-22 Carl Love <cel@us.ibm.com>
18723 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
18724 LVX_V1TI): Add macro expansion.
18725 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
18726 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
18727 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
18728 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
18729 Change check to determine if the instruction is a byte reversing
18730 entry. Fix typo in comment.
18731 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
18732 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
18733 Add def_builtin calls for new builtins.
18734 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
18735 Add define_insn expansion.
18737 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18739 * config/rl78/rl78.md (umaxdi3): New define_expand.
18741 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
18743 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
18744 for non-QImode registers.
18746 2018-01-22 Richard Biener <rguenther@suse.de>
18748 PR tree-optimization/83963
18749 * graphite-scop-detection.c (scop_detection::get_sese): Delay
18750 including the loop exit block.
18751 (scop_detection::merge_sese): Likewise.
18752 (scop_detection::add_scop): Do it here instead.
18754 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18756 * doc/sourcebuild.texi (arm_softfloat): Document.
18758 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
18761 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
18762 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
18763 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
18765 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18766 David Edelsohn <dje.gcc@gmail.com>
18769 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
18770 Change "crset eq" to "crset 2".
18771 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
18772 (*call_indirect_aix<mode>_nospec): Likewise.
18773 (*call_value_indirect_aix<mode>_nospec): Likewise.
18774 (*call_indirect_elfv2<mode>_nospec): Likewise.
18775 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
18776 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
18777 change assembly output from . to $.
18778 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
18779 (indirect_jump<mode>_nospec): Change assembly output from . to $.
18780 (*tablejump<mode>_internal1_nospec): Likewise.
18782 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
18785 * config/sh/sh_optimize_sett_clrt.cc:
18786 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
18788 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
18790 PR tree-optimization/83940
18791 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
18792 offset_dt to vect_constant_def rather than vect_unknown_def_type.
18793 (vect_check_load_store_mask): Add a mask_dt_out parameter and
18794 use it to pass back the definition type.
18795 (vect_check_store_rhs): Likewise rhs_dt_out.
18796 (vect_build_gather_load_calls): Add a mask_dt argument and use
18797 it instead of a call to vect_is_simple_use.
18798 (vectorizable_store): Update calls to vect_check_load_store_mask
18799 and vect_check_store_rhs. Use the dt returned by the latter instead
18800 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
18801 instead of calls to vect_is_simple_use. Pass the scalar rather
18802 than the vector operand to vect_is_simple_use when handling
18803 second and subsequent copies of an rhs value.
18804 (vectorizable_load): Update calls to vect_check_load_store_mask
18805 and vect_build_gather_load_calls. Use the cached mask_dt and
18806 gs_info.offset_dt instead of calls to vect_is_simple_use.
18808 2018-01-20 Jakub Jelinek <jakub@redhat.com>
18810 PR middle-end/83945
18811 * tree-emutls.c: Include gimplify.h.
18812 (lower_emutls_2): New function.
18813 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
18814 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
18815 it before further processing.
18818 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
18819 UINTVAL (trueop1) instead of INTVAL (op1).
18821 2018-01-19 Jakub Jelinek <jakub@redhat.com>
18825 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
18826 INCOMING_FRAME_SP_OFFSET if not defined.
18827 (scan_trace): Add ENTRY argument. If true and
18828 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
18829 emit a note to adjust the CFA offset.
18830 (create_cfi_notes): Adjust scan_trace callers.
18831 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
18832 INCOMING_FRAME_SP_OFFSET in the CIE.
18833 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
18834 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
18836 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
18837 * doc/tm.texi: Regenerated.
18839 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18841 PR rtl-optimization/83147
18842 * lra-constraints.c (remove_inheritance_pseudos): Use
18843 lra_substitute_pseudo_within_insn.
18845 2018-01-19 Tom de Vries <tom@codesourcery.com>
18846 Cesar Philippidis <cesar@codesourcery.com>
18849 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
18851 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
18854 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
18855 spaces for function labels.
18857 2018-01-19 Martin Liska <mliska@suse.cz>
18859 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
18860 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
18861 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
18862 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
18863 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
18864 (PRED_CONST_RETURN): Change from 69 to 65.
18865 (PRED_NULL_RETURN): Change from 91 to 71.
18866 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
18867 (PRED_LOOP_GUARD): Change from 66 to 73.
18869 2018-01-19 Martin Liska <mliska@suse.cz>
18871 * predict.c (predict_insn_def): Add new assert.
18872 (struct branch_predictor): Change type to signed integer.
18873 (test_prediction_value_range): Amend test to cover
18874 PROB_UNINITIALIZED.
18875 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
18876 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
18877 (PRED_LOOP_ITERATIONS_MAX): Likewise.
18878 (PRED_LOOP_IV_COMPARE): Likewise.
18879 * predict.h (PROB_UNINITIALIZED): Define new constant.
18881 2018-01-19 Martin Liska <mliska@suse.cz>
18883 * predict.c (dump_prediction): Add new format for
18884 analyze_brprob.py script which is enabled with -details
18886 * profile-count.h (precise_p): New function.
18888 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
18890 PR tree-optimization/83922
18891 * tree-vect-loop.c (vect_verify_full_masking): Return false if
18892 there are no statements that need masking.
18893 (vect_active_double_reduction_p): New function.
18894 (vect_analyze_loop_operations): Use it when handling phis that
18895 are not in the loop header.
18897 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
18899 PR tree-optimization/83914
18900 * tree-vect-loop.c (vectorizable_induction): Don't convert
18901 init_expr or apply the peeling adjustment for inductions
18902 that are nested within the vectorized loop.
18904 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18906 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
18909 2018-01-18 Jakub Jelinek <jakub@redhat.com>
18913 * function.h (gimplify_parameters): Add gimple_seq * argument.
18914 * function.c: Include gimple.h and options.h.
18915 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
18916 for the added local temporaries if needed.
18917 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
18918 if there are any parameter cleanups, wrap whole body into a
18919 try/finally with the cleanups.
18921 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
18924 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18925 Use GET_MODE_CLASS for scalar floating point.
18927 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
18931 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
18932 Fix call of call_cgraph_insertion_hooks.
18934 2018-01-18 Martin Sebor <msebor@redhat.com>
18936 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
18938 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
18941 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
18944 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
18947 * common.opt: (-ffile-prefix-map): New option.
18948 * opts.c (common_handle_option): Defer it.
18949 * opts-global.c (handle_common_deferred_options): Handle it.
18950 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
18951 * file-prefix-map.h: New file.
18952 (remap_debug_filename, add_debug_prefix_map): ...here.
18953 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
18954 * final.c (debug_prefix_map, add_debug_prefix_map
18955 remap_debug_filename): Move to...
18956 * file-prefix-map.c: New file.
18957 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
18958 generalize, get rid of alloca(), use strrchr() instead of strchr().
18959 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
18960 Implement in terms of add_prefix_map().
18961 (remap_macro_filename, remap_debug_filename): Implement in term of
18963 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
18964 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
18965 * dbxout.c: Include file-prefix-map.h.
18966 * varasm.c: Likewise.
18967 * vmsdbgout.c: Likewise.
18968 * xcoffout.c: Likewise.
18969 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
18970 * doc/cppopts.texi (-fmacro-prefix-map): Document.
18971 * doc/invoke.texi (-ffile-prefix-map): Document.
18972 (-fdebug-prefix-map): Update description.
18974 2018-01-18 Martin Liska <mliska@suse.cz>
18976 * config/i386/i386.c (indirect_thunk_name): Document that also
18978 (output_indirect_thunk): Document why both instructions
18979 (pause and lfence) are generated.
18981 2018-01-18 Richard Biener <rguenther@suse.de>
18983 PR tree-optimization/83887
18984 * graphite-scop-detection.c
18985 (scop_detection::get_nearest_dom_with_single_entry): Remove.
18986 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
18987 (scop_detection::merge_sese): Re-implement with a flood-fill
18988 algorithm that properly finds a SESE region if it exists.
18990 2018-01-18 Jakub Jelinek <jakub@redhat.com>
18993 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
18994 pointer_diff optimizations use view_convert instead of convert.
18996 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18998 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
18999 Generate different code for -mno-speculate-indirect-jumps.
19000 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
19001 (*call_indirect_aix<mode>): Disable for
19002 -mno-speculate-indirect-jumps.
19003 (*call_indirect_aix<mode>_nospec): New define_insn.
19004 (*call_value_indirect_aix<mode>): Disable for
19005 -mno-speculate-indirect-jumps.
19006 (*call_value_indirect_aix<mode>_nospec): New define_insn.
19007 (*sibcall_nonlocal_sysv<mode>): Generate different code for
19008 -mno-speculate-indirect-jumps.
19009 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
19011 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19013 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
19014 long double type, set the flags for noting the default long double
19015 type, even if we don't pass or return a long double type.
19017 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
19020 * ipa-inline.c (flatten_function): Do not overwrite final inlining
19023 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
19025 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
19026 support for merge[hl].
19027 (fold_mergehl_helper): New helper function.
19028 (tree-vector-builder.h): New #include for tree_vector_builder usage.
19029 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
19030 (altivec_vmrglw_direct): Add xxmrglw insn.
19032 2018-01-17 Andrew Waterman <andrew@sifive.com>
19034 * config/riscv/riscv.c (riscv_conditional_register_usage): If
19035 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
19037 2018-01-17 David Malcolm <dmalcolm@redhat.com>
19040 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
19041 call the lto_location_cache before reading the
19042 DECL_SOURCE_LOCATION of the types.
19044 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
19045 Richard Sandiford <richard.sandiford@linaro.org>
19047 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
19048 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
19049 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
19050 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
19051 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
19053 * config/aarch64/constraints.md (aarch64_movti_operand):
19055 * config/aarch64/predicates.md (Uti): Add new constraint.
19057 2018-01-17 Carl Love <cel@us.ibm.com>
19059 * config/rs6000/vsx.md (define_expand xl_len_r,
19060 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
19061 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
19063 (define_expand, define_insn): Move the shift left from the
19064 define_insn to the define_expand for lxvl and stxvl instructions.
19065 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
19066 and XL_LEN_R definitions to PURE.
19068 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
19070 * config/i386/i386.c (indirect_thunk_name): Declare regno
19071 as unsigned int. Compare regno with INVALID_REGNUM.
19072 (output_indirect_thunk): Ditto.
19073 (output_indirect_thunk_function): Ditto.
19074 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
19075 in the call to output_indirect_thunk_function.
19077 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
19079 PR middle-end/83884
19080 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
19081 rather than the size of inner_type to determine the stack slot size
19082 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
19084 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
19087 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
19090 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19092 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
19093 endian Linux systems to optionally enable multilibs for selecting
19094 the long double type if the user configured an explicit type.
19095 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
19096 have no long double multilibs if not defined.
19097 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
19098 warn if the user used -mabi={ieee,ibm}longdouble and we built
19099 multilibs for long double.
19100 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
19101 appropriate multilib option.
19102 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
19104 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
19105 for building long double multilibs.
19106 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
19108 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
19110 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
19113 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
19115 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
19118 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
19121 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
19124 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
19126 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
19127 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
19129 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
19131 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
19132 different rtl trees depending on TARGET_64BIT.
19133 (rs6000_gen_lvx): Likewise.
19135 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
19137 * config/visium/visium.md (nop): Tweak comment.
19138 (hazard_nop): Likewise.
19140 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19142 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
19143 -mspeculate-indirect-jumps.
19144 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
19145 for -mno-speculate-indirect-jumps.
19146 (*call_indirect_elfv2<mode>_nospec): New define_insn.
19147 (*call_value_indirect_elfv2<mode>): Disable for
19148 -mno-speculate-indirect-jumps.
19149 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
19150 (indirect_jump): Emit different RTL for
19151 -mno-speculate-indirect-jumps.
19152 (*indirect_jump<mode>): Disable for
19153 -mno-speculate-indirect-jumps.
19154 (*indirect_jump<mode>_nospec): New define_insn.
19155 (tablejump): Emit different RTL for
19156 -mno-speculate-indirect-jumps.
19157 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
19158 (tablejumpsi_nospec): New define_expand.
19159 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
19160 (tablejumpdi_nospec): New define_expand.
19161 (*tablejump<mode>_internal1): Disable for
19162 -mno-speculate-indirect-jumps.
19163 (*tablejump<mode>_internal1_nospec): New define_insn.
19164 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
19167 2018-01-16 Artyom Skrobov tyomitch@gmail.com
19169 * caller-save.c (insert_save): Drop unnecessary parameter. All
19172 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19173 Richard Biener <rguenth@suse.de>
19176 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
19177 return early, inline manually is_gimple_sizepos. Make sure if we
19178 call gimplify_expr we don't end up with a gimple constant.
19179 * tree.c (variably_modified_type_p): Don't return true for
19180 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
19181 * gimplify.h (is_gimple_sizepos): Remove.
19183 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19185 PR tree-optimization/83857
19186 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
19187 vectorizable_live_operation for pure SLP statements.
19188 (vectorizable_live_operation): Handle PHIs.
19190 2018-01-16 Richard Biener <rguenther@suse.de>
19192 PR tree-optimization/83867
19193 * tree-vect-stmts.c (vect_transform_stmt): Precompute
19194 nested_in_vect_loop_p since the scalar stmt may get invalidated.
19196 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19199 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
19200 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
19201 If off is not INTEGER_CST, issue a may not be aligned warning
19202 rather than isn't aligned. Use isn%'t rather than isn't.
19203 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
19205 <case MULT_EXPR>: Improve the case when bottom and one of the
19206 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
19207 operand, in that case check if the other operand is multiple of
19208 bottom divided by the INTEGER_CST operand.
19210 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19213 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
19214 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
19215 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
19216 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
19217 * config/pa/pa.c (pa_function_arg_advance): Likewise.
19218 (pa_function_arg, pa_arg_partial_bytes): Likewise.
19219 (pa_function_arg_size): New function.
19221 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19223 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
19224 in a separate statement.
19226 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19228 PR tree-optimization/83847
19229 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
19230 group gathers and scatters.
19232 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19234 PR rtl-optimization/86620
19235 * params.def (max-sched-ready-insns): Bump minimum value to 1.
19237 PR rtl-optimization/83213
19238 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
19239 to last if both are JUMP_INSNs.
19241 PR tree-optimization/83843
19242 * gimple-ssa-store-merging.c
19243 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
19244 store_immediate_info for bswap/nop orig_stores.
19246 2018-01-15 Andrew Waterman <andrew@sifive.com>
19248 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
19250 <UDIV>: Increase cost if !TARGET_DIV.
19252 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
19254 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
19255 (define_attr "cr_logical_3op"): New.
19256 (cceq_ior_compare): Adjust.
19257 (cceq_ior_compare_complement): Adjust.
19258 (*cceq_rev_compare): Adjust.
19259 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
19260 (is_cracked_insn): Adjust.
19261 (insn_must_be_first_in_group): Adjust.
19262 * config/rs6000/40x.md: Adjust.
19263 * config/rs6000/440.md: Adjust.
19264 * config/rs6000/476.md: Adjust.
19265 * config/rs6000/601.md: Adjust.
19266 * config/rs6000/603.md: Adjust.
19267 * config/rs6000/6xx.md: Adjust.
19268 * config/rs6000/7450.md: Adjust.
19269 * config/rs6000/7xx.md: Adjust.
19270 * config/rs6000/8540.md: Adjust.
19271 * config/rs6000/cell.md: Adjust.
19272 * config/rs6000/e300c2c3.md: Adjust.
19273 * config/rs6000/e500mc.md: Adjust.
19274 * config/rs6000/e500mc64.md: Adjust.
19275 * config/rs6000/e5500.md: Adjust.
19276 * config/rs6000/e6500.md: Adjust.
19277 * config/rs6000/mpc.md: Adjust.
19278 * config/rs6000/power4.md: Adjust.
19279 * config/rs6000/power5.md: Adjust.
19280 * config/rs6000/power6.md: Adjust.
19281 * config/rs6000/power7.md: Adjust.
19282 * config/rs6000/power8.md: Adjust.
19283 * config/rs6000/power9.md: Adjust.
19284 * config/rs6000/rs64.md: Adjust.
19285 * config/rs6000/titan.md: Adjust.
19287 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19289 * config/i386/predicates.md (indirect_branch_operand): Rewrite
19290 ix86_indirect_branch_register logic.
19292 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19294 * config/i386/constraints.md (Bs): Update
19295 ix86_indirect_branch_register check. Don't check
19296 ix86_indirect_branch_register with GOT_memory_operand.
19298 * config/i386/predicates.md (GOT_memory_operand): Don't check
19299 ix86_indirect_branch_register here.
19300 (GOT32_symbol_operand): Likewise.
19302 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19304 * config/i386/predicates.md (constant_call_address_operand):
19305 Rewrite ix86_indirect_branch_register logic.
19306 (sibcall_insn_operand): Likewise.
19308 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19310 * config/i386/constraints.md (Bs): Replace
19311 ix86_indirect_branch_thunk_register with
19312 ix86_indirect_branch_register.
19314 * config/i386/i386.md (indirect_jump): Likewise.
19315 (tablejump): Likewise.
19316 (*sibcall_memory): Likewise.
19317 (*sibcall_value_memory): Likewise.
19318 Peepholes of indirect call and jump via memory: Likewise.
19319 * config/i386/i386.opt: Likewise.
19320 * config/i386/predicates.md (indirect_branch_operand): Likewise.
19321 (GOT_memory_operand): Likewise.
19322 (call_insn_operand): Likewise.
19323 (sibcall_insn_operand): Likewise.
19324 (GOT32_symbol_operand): Likewise.
19326 2018-01-15 Jakub Jelinek <jakub@redhat.com>
19328 PR middle-end/83837
19329 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
19330 type rather than type addr's type points to.
19331 (expand_omp_atomic_mutex): Likewise.
19332 (expand_omp_atomic): Likewise.
19334 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19337 * config/i386/i386.c (output_indirect_thunk_function): Use
19338 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
19339 for __x86_return_thunk.
19341 2018-01-15 Richard Biener <rguenther@suse.de>
19343 PR middle-end/83850
19344 * expmed.c (extract_bit_field_1): Fix typo.
19346 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19349 * config/arm/iterators.md (VF): New mode iterator.
19350 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
19351 Remove integer-related logic from pattern.
19352 (neon_vabd<mode>_3): Likewise.
19354 2018-01-15 Jakub Jelinek <jakub@redhat.com>
19356 PR middle-end/82694
19357 * common.opt (fstrict-overflow): No longer an alias.
19358 (fwrapv-pointer): New option.
19359 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
19360 also for pointer types based on flag_wrapv_pointer.
19361 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
19362 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
19363 opts->x_flag_wrapv got set.
19364 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
19365 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
19366 POINTER_TYPE_OVERFLOW_UNDEFINED.
19367 * match.pd: Likewise in address comparison pattern.
19368 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
19370 2018-01-15 Richard Biener <rguenther@suse.de>
19373 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
19374 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
19375 Reset type names to their identifier if their TYPE_DECL doesn't
19376 have linkage (and thus is used for ODR and devirt).
19377 (save_debug_info_for_decl): Remove.
19378 (save_debug_info_for_type): Likewise.
19379 (add_tree_to_fld_list): Adjust.
19380 * tree-pretty-print.c (dump_generic_node): Make dumping of
19381 type names more robust.
19383 2018-01-15 Richard Biener <rguenther@suse.de>
19385 * BASE-VER: Bump to 8.0.1.
19387 2018-01-14 Martin Sebor <msebor@redhat.com>
19390 * builtins.c (check_access): Avoid warning when the no-warning bit
19393 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
19395 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
19396 * ira-color (allocno_hard_regs_compare): Likewise.
19398 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
19401 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
19402 Use .pushsection/.popsection.
19404 2018-01-14 Martin Sebor <msebor@redhat.com>
19407 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
19409 2018-01-14 Jakub Jelinek <jakub@redhat.com>
19411 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
19412 entry from extra_headers.
19413 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
19414 extra_headers, make the list bitwise identical to the i?86-*-* one.
19416 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
19418 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
19419 -mcmodel=large with -mindirect-branch=thunk,
19420 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
19421 -mfunction-return=thunk-extern.
19422 * doc/invoke.texi: Document -mcmodel=large is incompatible with
19423 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
19424 -mfunction-return=thunk and -mfunction-return=thunk-extern.
19426 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
19428 * config/i386/i386.c (print_reg): Print the name of the full
19429 integer register without '%'.
19430 (ix86_print_operand): Handle 'V'.
19431 * doc/extend.texi: Document 'V' modifier.
19433 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
19435 * config/i386/constraints.md (Bs): Disallow memory operand for
19436 -mindirect-branch-register.
19438 * config/i386/predicates.md (indirect_branch_operand): Likewise.
19439 (GOT_memory_operand): Likewise.
19440 (call_insn_operand): Likewise.
19441 (sibcall_insn_operand): Likewise.
19442 (GOT32_symbol_operand): Likewise.
19443 * config/i386/i386.md (indirect_jump): Call convert_memory_address
19444 for -mindirect-branch-register.
19445 (tablejump): Likewise.
19446 (*sibcall_memory): Likewise.
19447 (*sibcall_value_memory): Likewise.
19448 Disallow peepholes of indirect call and jump via memory for
19449 -mindirect-branch-register.
19450 (*call_pop): Replace m with Bw.
19451 (*call_value_pop): Likewise.
19452 (*sibcall_pop_memory): Replace m with Bs.
19453 * config/i386/i386.opt (mindirect-branch-register): New option.
19454 * doc/invoke.texi: Document -mindirect-branch-register option.
19456 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
19458 * config/i386/i386-protos.h (ix86_output_function_return): New.
19459 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
19460 set function_return_type.
19461 (indirect_thunk_name): Add ret_p to indicate thunk for function
19463 (output_indirect_thunk_function): Pass false to
19464 indirect_thunk_name.
19465 (ix86_output_indirect_branch_via_reg): Likewise.
19466 (ix86_output_indirect_branch_via_push): Likewise.
19467 (output_indirect_thunk_function): Create alias for function
19468 return thunk if regno < 0.
19469 (ix86_output_function_return): New function.
19470 (ix86_handle_fndecl_attribute): Handle function_return.
19471 (ix86_attribute_table): Add function_return.
19472 * config/i386/i386.h (machine_function): Add
19473 function_return_type.
19474 * config/i386/i386.md (simple_return_internal): Use
19475 ix86_output_function_return.
19476 (simple_return_internal_long): Likewise.
19477 * config/i386/i386.opt (mfunction-return=): New option.
19478 (indirect_branch): Mention -mfunction-return=.
19479 * doc/extend.texi: Document function_return function attribute.
19480 * doc/invoke.texi: Document -mfunction-return= option.
19482 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
19484 * config/i386/i386-opts.h (indirect_branch): New.
19485 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
19486 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
19487 with local indirect jump when converting indirect call and jump.
19488 (ix86_set_indirect_branch_type): New.
19489 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
19490 (indirectlabelno): New.
19491 (indirect_thunk_needed): Likewise.
19492 (indirect_thunk_bnd_needed): Likewise.
19493 (indirect_thunks_used): Likewise.
19494 (indirect_thunks_bnd_used): Likewise.
19495 (INDIRECT_LABEL): Likewise.
19496 (indirect_thunk_name): Likewise.
19497 (output_indirect_thunk): Likewise.
19498 (output_indirect_thunk_function): Likewise.
19499 (ix86_output_indirect_branch_via_reg): Likewise.
19500 (ix86_output_indirect_branch_via_push): Likewise.
19501 (ix86_output_indirect_branch): Likewise.
19502 (ix86_output_indirect_jmp): Likewise.
19503 (ix86_code_end): Call output_indirect_thunk_function if needed.
19504 (ix86_output_call_insn): Call ix86_output_indirect_branch if
19506 (ix86_handle_fndecl_attribute): Handle indirect_branch.
19507 (ix86_attribute_table): Add indirect_branch.
19508 * config/i386/i386.h (machine_function): Add indirect_branch_type
19509 and has_local_indirect_jump.
19510 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
19512 (tablejump): Likewise.
19513 (*indirect_jump): Use ix86_output_indirect_jmp.
19514 (*tablejump_1): Likewise.
19515 (simple_return_indirect_internal): Likewise.
19516 * config/i386/i386.opt (mindirect-branch=): New option.
19517 (indirect_branch): New.
19520 (thunk-inline): Likewise.
19521 (thunk-extern): Likewise.
19522 * doc/extend.texi: Document indirect_branch function attribute.
19523 * doc/invoke.texi: Document -mindirect-branch= option.
19525 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
19528 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
19530 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
19532 * ipa-inline.c (want_inline_small_function_p): Return false if
19533 inlining has already failed with CIF_FINAL_ERROR.
19534 (update_caller_keys): Call want_inline_small_function_p before
19536 (update_callee_keys): Likewise.
19538 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19540 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
19542 (rs6000_quadword_masked_address_p): Likewise.
19543 (quad_aligned_load_p): Likewise.
19544 (quad_aligned_store_p): Likewise.
19545 (const_load_sequence_p): Add comment to describe the outer-most loop.
19546 (mimic_memory_attributes_and_flags): New function.
19547 (rs6000_gen_stvx): Likewise.
19548 (replace_swapped_aligned_store): Likewise.
19549 (rs6000_gen_lvx): Likewise.
19550 (replace_swapped_aligned_load): Likewise.
19551 (replace_swapped_load_constant): Capitalize argument name in
19552 comment describing this function.
19553 (rs6000_analyze_swaps): Add a third pass to search for vector loads
19554 and stores that access quad-word aligned addresses and replace
19555 with stvx or lvx instructions when appropriate.
19556 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
19557 New function prototype.
19558 (rs6000_quadword_masked_address_p): Likewise.
19559 (rs6000_gen_lvx): Likewise.
19560 (rs6000_gen_stvx): Likewise.
19561 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
19562 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
19563 when memory address is aligned.
19564 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
19565 this split to select lvx instruction when memory address is aligned.
19566 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
19567 instruction when memory address is aligned.
19568 (*vsx_le_perm_load_v16qi): Likewise.
19569 (four unnamed splitters): Modify to select the stvx instruction
19570 when memory is aligned.
19572 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
19574 * predict.c (determine_unlikely_bbs): Handle correctly BBs
19575 which appears in the queue multiple times.
19577 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19578 Alan Hayward <alan.hayward@arm.com>
19579 David Sherwood <david.sherwood@arm.com>
19581 * tree-vectorizer.h (vec_lower_bound): New structure.
19582 (_loop_vec_info): Add check_nonzero and lower_bounds.
19583 (LOOP_VINFO_CHECK_NONZERO): New macro.
19584 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
19585 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
19586 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
19587 fields. Make seg_len the distance travelled, not including the
19589 (dr_direction_indicator): Declare.
19590 (dr_zero_step_indicator): Likewise.
19591 (dr_known_forward_stride_p): Likewise.
19592 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
19594 (runtime_alias_check_p): Allow runtime alias checks with
19596 (operator ==): Compare access_size and align.
19597 (prune_runtime_alias_test_list): Rework for new distinction between
19598 the access_size and seg_len.
19599 (create_intersect_range_checks_index): Likewise. Cope with polynomial
19601 (get_segment_min_max): New function.
19602 (create_intersect_range_checks): Use it.
19603 (dr_step_indicator): New function.
19604 (dr_direction_indicator): Likewise.
19605 (dr_zero_step_indicator): Likewise.
19606 (dr_known_forward_stride_p): Likewise.
19607 * tree-loop-distribution.c (data_ref_segment_size): Return
19608 DR_STEP * (niters - 1).
19609 (compute_alias_check_pairs): Update call to the dr_with_seg_len
19611 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
19612 (vect_preserves_scalar_order_p): New function, split out from...
19613 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
19614 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
19615 (vect_vfa_access_size): New function.
19616 (vect_vfa_align): Likewise.
19617 (vect_compile_time_alias): Take access_size_a and access_b arguments.
19618 (dump_lower_bound): New function.
19619 (vect_check_lower_bound): Likewise.
19620 (vect_small_gap_p): Likewise.
19621 (vectorizable_with_step_bound_p): Likewise.
19622 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
19623 depencies if the vectorization factor is 1. Convert the checks
19624 for nonzero steps into checks on the bounds of DR_STEP. Try using
19625 a bunds check for variable steps if the minimum required step is
19626 relatively small. Update calls to the dr_with_seg_len
19627 constructor and to vect_compile_time_alias.
19628 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
19630 (vect_loop_versioning): Call it.
19631 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
19633 (vect_estimate_min_profitable_iters): Account for any bounds checks.
19635 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19636 Alan Hayward <alan.hayward@arm.com>
19637 David Sherwood <david.sherwood@arm.com>
19639 * doc/sourcebuild.texi (vect_scatter_store): Document.
19640 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
19642 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
19644 * genopinit.c (main): Add supports_vec_scatter_store and
19645 supports_vec_scatter_store_cached to target_optabs.
19646 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
19647 IFN_MASK_SCATTER_STORE.
19648 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
19650 * internal-fn.h (internal_store_fn_p): Declare.
19651 (internal_fn_stored_value_index): Likewise.
19652 * internal-fn.c (scatter_store_direct): New macro.
19653 (expand_scatter_store_optab_fn): New function.
19654 (direct_scatter_store_optab_supported_p): New macro.
19655 (internal_store_fn_p): New function.
19656 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
19657 IFN_MASK_SCATTER_STORE.
19658 (internal_fn_mask_index): Likewise.
19659 (internal_fn_stored_value_index): New function.
19660 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
19661 for scatter stores.
19662 * optabs-query.h (supports_vec_scatter_store_p): Declare.
19663 * optabs-query.c (supports_vec_scatter_store_p): New function.
19664 * tree-vectorizer.h (vect_get_store_rhs): Declare.
19665 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
19666 true for scatter stores.
19667 (vect_gather_scatter_fn_p): Handle scatter stores too.
19668 (vect_check_gather_scatter): Consider using scatter stores if
19669 supports_vec_scatter_store_p.
19670 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
19671 scatter stores too.
19672 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
19673 internal_fn_stored_value_index.
19674 (check_load_store_masking): Handle scatter stores too.
19675 (vect_get_store_rhs): Make public.
19676 (vectorizable_call): Use internal_store_fn_p.
19677 (vectorizable_store): Handle scatter store internal functions.
19678 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
19679 when deciding whether the end of the group has been reached.
19680 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
19681 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
19682 (mask_scatter_store<mode>): New insns.
19684 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19685 Alan Hayward <alan.hayward@arm.com>
19686 David Sherwood <david.sherwood@arm.com>
19688 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
19689 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
19690 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
19692 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
19693 Use vect_truncate_gather_scatter_offset if we can't treat the
19694 operation as a normal gather load or scatter store.
19695 (get_group_load_store_type): Take the gather_scatter_info
19696 as argument. Try using a gather load or scatter store for
19697 single-element groups.
19698 (get_load_store_type): Update calls to get_group_load_store_type
19699 and vect_use_strided_gather_scatters_p.
19701 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19702 Alan Hayward <alan.hayward@arm.com>
19703 David Sherwood <david.sherwood@arm.com>
19705 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
19706 optional tree argument.
19707 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
19709 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
19710 but continue to use the current value as a fallback.
19711 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
19712 to compare the updates.
19713 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
19714 (get_load_store_type): Use it when handling a strided access.
19715 (vect_get_strided_load_store_ops): New function.
19716 (vect_get_data_ptr_increment): Likewise.
19717 (vectorizable_load): Handle strided gather loads. Always pass
19718 a step to vect_create_data_ref_ptr and bump_vector_ptr.
19720 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19721 Alan Hayward <alan.hayward@arm.com>
19722 David Sherwood <david.sherwood@arm.com>
19724 * doc/md.texi (gather_load@var{m}): Document.
19725 (mask_gather_load@var{m}): Likewise.
19726 * genopinit.c (main): Add supports_vec_gather_load and
19727 supports_vec_gather_load_cached to target_optabs.
19728 * optabs-tree.c (init_tree_optimization_optabs): Use
19729 ggc_cleared_alloc to allocate target_optabs.
19730 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
19731 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
19733 * internal-fn.h (internal_load_fn_p): Declare.
19734 (internal_gather_scatter_fn_p): Likewise.
19735 (internal_fn_mask_index): Likewise.
19736 (internal_gather_scatter_fn_supported_p): Likewise.
19737 * internal-fn.c (gather_load_direct): New macro.
19738 (expand_gather_load_optab_fn): New function.
19739 (direct_gather_load_optab_supported_p): New macro.
19740 (direct_internal_fn_optab): New function.
19741 (internal_load_fn_p): Likewise.
19742 (internal_gather_scatter_fn_p): Likewise.
19743 (internal_fn_mask_index): Likewise.
19744 (internal_gather_scatter_fn_supported_p): Likewise.
19745 * optabs-query.c (supports_at_least_one_mode_p): New function.
19746 (supports_vec_gather_load_p): Likewise.
19747 * optabs-query.h (supports_vec_gather_load_p): Declare.
19748 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
19749 and memory_type field.
19750 (NUM_PATTERNS): Bump to 15.
19751 * tree-vect-data-refs.c: Include internal-fn.h.
19752 (vect_gather_scatter_fn_p): New function.
19753 (vect_describe_gather_scatter_call): Likewise.
19754 (vect_check_gather_scatter): Try using internal functions for
19755 gather loads. Recognize existing calls to a gather load function.
19756 (vect_analyze_data_refs): Consider using gather loads if
19757 supports_vec_gather_load_p.
19758 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
19759 (vect_get_gather_scatter_offset_type): Likewise.
19760 (vect_convert_mask_for_vectype): Likewise.
19761 (vect_add_conversion_to_patterm): Likewise.
19762 (vect_try_gather_scatter_pattern): Likewise.
19763 (vect_recog_gather_scatter_pattern): New pattern recognizer.
19764 (vect_vect_recog_func_ptrs): Add it.
19765 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
19766 internal_fn_mask_index and internal_gather_scatter_fn_p.
19767 (check_load_store_masking): Take the gather_scatter_info as an
19768 argument and handle gather loads.
19769 (vect_get_gather_scatter_ops): New function.
19770 (vectorizable_call): Check internal_load_fn_p.
19771 (vectorizable_load): Likewise. Handle gather load internal
19773 (vectorizable_store): Update call to check_load_store_masking.
19774 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
19775 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
19776 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
19777 (aarch64_gather_scale_operand_d): New predicates.
19778 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
19779 (mask_gather_load<mode>): New insns.
19781 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19782 Alan Hayward <alan.hayward@arm.com>
19783 David Sherwood <david.sherwood@arm.com>
19785 * optabs.def (fold_left_plus_optab): New optab.
19786 * doc/md.texi (fold_left_plus_@var{m}): Document.
19787 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
19788 * internal-fn.c (fold_left_direct): Define.
19789 (expand_fold_left_optab_fn): Likewise.
19790 (direct_fold_left_optab_supported_p): Likewise.
19791 * fold-const-call.c (fold_const_fold_left): New function.
19792 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
19793 * tree-parloops.c (valid_reduction_p): New function.
19794 (gather_scalar_reductions): Use it.
19795 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
19796 (vect_finish_replace_stmt): Declare.
19797 * tree-vect-loop.c (fold_left_reduction_fn): New function.
19798 (needs_fold_left_reduction_p): New function, split out from...
19799 (vect_is_simple_reduction): ...here. Accept reductions that
19800 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
19801 (vect_force_simple_reduction): Also store the reduction type in
19802 the assignment's STMT_VINFO_REDUC_TYPE.
19803 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
19804 (merge_with_identity): New function.
19805 (vect_expand_fold_left): Likewise.
19806 (vectorize_fold_left_reduction): Likewise.
19807 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
19808 scalar phi in place for it. Check for target support and reject
19809 cases that would reassociate the operation. Defer the transform
19810 phase to vectorize_fold_left_reduction.
19811 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
19812 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
19813 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
19815 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19817 * tree-if-conv.c (predicate_mem_writes): Remove redundant
19818 call to ifc_temp_var.
19820 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19821 Alan Hayward <alan.hayward@arm.com>
19822 David Sherwood <david.sherwood@arm.com>
19824 * target.def (legitimize_address_displacement): Take the original
19825 offset as a poly_int.
19826 * targhooks.h (default_legitimize_address_displacement): Update
19828 * targhooks.c (default_legitimize_address_displacement): Likewise.
19829 * doc/tm.texi: Regenerate.
19830 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
19831 as an argument, moving assert of ad->disp == ad->disp_term to...
19832 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
19833 Try calling targetm.legitimize_address_displacement before expanding
19834 the address rather than afterwards, and adjust for the new interface.
19835 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
19836 Match the new hook interface. Handle SVE addresses.
19837 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
19838 new hook interface.
19840 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19842 * Makefile.in (OBJS): Add early-remat.o.
19843 * target.def (select_early_remat_modes): New hook.
19844 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
19845 * doc/tm.texi: Regenerate.
19846 * targhooks.h (default_select_early_remat_modes): Declare.
19847 * targhooks.c (default_select_early_remat_modes): New function.
19848 * timevar.def (TV_EARLY_REMAT): New timevar.
19849 * passes.def (pass_early_remat): New pass.
19850 * tree-pass.h (make_pass_early_remat): Declare.
19851 * early-remat.c: New file.
19852 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
19854 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
19856 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19857 Alan Hayward <alan.hayward@arm.com>
19858 David Sherwood <david.sherwood@arm.com>
19860 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
19861 vfm1 with a bound_epilog parameter.
19862 (vect_do_peeling): Update calls accordingly, and move the prologue
19863 call earlier in the function. Treat the base bound_epilog as 0 for
19864 fully-masked loops and retain vf - 1 for other loops. Add 1 to
19865 this base when peeling for gaps.
19866 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
19867 with fully-masked loops.
19868 (vect_estimate_min_profitable_iters): Handle the single peeled
19869 iteration in that case.
19871 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19872 Alan Hayward <alan.hayward@arm.com>
19873 David Sherwood <david.sherwood@arm.com>
19875 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
19876 single-element interleaving even if the size is not a power of 2.
19877 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
19878 accesses for single-element interleaving if the group size is
19881 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19882 Alan Hayward <alan.hayward@arm.com>
19883 David Sherwood <david.sherwood@arm.com>
19885 * doc/md.texi (fold_extract_last_@var{m}): Document.
19886 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
19887 * optabs.def (fold_extract_last_optab): New optab.
19888 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
19889 * internal-fn.c (fold_extract_direct): New macro.
19890 (expand_fold_extract_optab_fn): Likewise.
19891 (direct_fold_extract_optab_supported_p): Likewise.
19892 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
19893 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19894 EXTRACT_LAST_REDUCTION.
19895 (get_initial_def_for_reduction): Do not create an initial vector
19896 for EXTRACT_LAST_REDUCTION reductions.
19897 (vectorizable_reduction): Leave the scalar phi in place for
19898 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
19899 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
19900 epilogue code for EXTRACT_LAST_REDUCTION and defer the
19901 transform phase to vectorizable_condition.
19902 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
19904 (vect_finish_stmt_generation): ...here.
19905 (vect_finish_replace_stmt): New function.
19906 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
19907 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
19909 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
19911 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19912 Alan Hayward <alan.hayward@arm.com>
19913 David Sherwood <david.sherwood@arm.com>
19915 * doc/md.texi (extract_last_@var{m}): Document.
19916 * optabs.def (extract_last_optab): New optab.
19917 * internal-fn.def (EXTRACT_LAST): New internal function.
19918 * internal-fn.c (cond_unary_direct): New macro.
19919 (expand_cond_unary_optab_fn): Likewise.
19920 (direct_cond_unary_optab_supported_p): Likewise.
19921 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
19922 loops using EXTRACT_LAST.
19923 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
19924 (extract_last_<mode>): ...this optab.
19925 (vec_extract<mode><Vel>): Update accordingly.
19927 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19928 Alan Hayward <alan.hayward@arm.com>
19929 David Sherwood <david.sherwood@arm.com>
19931 * target.def (empty_mask_is_expensive): New hook.
19932 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
19933 * doc/tm.texi: Regenerate.
19934 * targhooks.h (default_empty_mask_is_expensive): Declare.
19935 * targhooks.c (default_empty_mask_is_expensive): New function.
19936 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
19937 if the target says that empty masks are expensive.
19938 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
19940 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
19942 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19943 Alan Hayward <alan.hayward@arm.com>
19944 David Sherwood <david.sherwood@arm.com>
19946 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
19947 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
19948 (vect_use_loop_mask_for_alignment_p): New function.
19949 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
19950 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
19951 niters_skip argument. Make sure that the first niters_skip elements
19952 of the first iteration are inactive.
19953 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
19954 Update call to vect_set_loop_masks_directly.
19955 (get_misalign_in_elems): New function, split out from...
19956 (vect_gen_prolog_loop_niters): ...here.
19957 (vect_update_init_of_dr): Take a code argument that specifies whether
19958 the adjustment should be added or subtracted.
19959 (vect_update_init_of_drs): Likewise.
19960 (vect_prepare_for_masked_peels): New function.
19961 (vect_do_peeling): Skip prologue peeling if we're using a mask
19962 instead. Update call to vect_update_inits_of_drs.
19963 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19965 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
19966 alignment. Do not include the number of peeled iterations in
19967 the minimum threshold in that case.
19968 (vectorizable_induction): Adjust the start value down by
19969 LOOP_VINFO_MASK_SKIP_NITERS iterations.
19970 (vect_transform_loop): Call vect_prepare_for_masked_peels.
19971 Take the number of skipped iterations into account when calculating
19973 * tree-vect-stmts.c (vect_gen_while_not): New function.
19975 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19976 Alan Hayward <alan.hayward@arm.com>
19977 David Sherwood <david.sherwood@arm.com>
19979 * doc/sourcebuild.texi (vect_fully_masked): Document.
19980 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
19981 default value to 0.
19982 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
19984 (vect_analyze_loop_2): ...here. Don't check the vectorization
19985 factor against the number of loop iterations if the loop is
19988 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
19989 Alan Hayward <alan.hayward@arm.com>
19990 David Sherwood <david.sherwood@arm.com>
19992 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
19993 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
19994 (dump_groups): Update accordingly.
19995 (iv_use::mem_type): New member variable.
19996 (address_p): New function.
19997 (record_use): Add a mem_type argument and initialize the new
19999 (record_group_use): Add a mem_type argument. Use address_p.
20000 Remove obsolete null checks of base_object. Update call to record_use.
20001 (find_interesting_uses_op): Update call to record_group_use.
20002 (find_interesting_uses_cond): Likewise.
20003 (find_interesting_uses_address): Likewise.
20004 (get_mem_type_for_internal_fn): New function.
20005 (find_address_like_use): Likewise.
20006 (find_interesting_uses_stmt): Try find_address_like_use before
20007 calling find_interesting_uses_op.
20008 (addr_offset_valid_p): Use the iv mem_type field as the type
20009 of the addressed memory.
20010 (add_autoinc_candidates): Likewise.
20011 (get_address_cost): Likewise.
20012 (split_small_address_groups_p): Use address_p.
20013 (split_address_groups): Likewise.
20014 (add_iv_candidate_for_use): Likewise.
20015 (autoinc_possible_for_pair): Likewise.
20016 (rewrite_groups): Likewise.
20017 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
20018 (determine_group_iv_cost): Update after split of USE_ADDRESS.
20019 (get_alias_ptr_type_for_ptr_address): New function.
20020 (rewrite_use_address): Rewrite address uses in calls that were
20021 identified by find_address_like_use.
20023 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20024 Alan Hayward <alan.hayward@arm.com>
20025 David Sherwood <david.sherwood@arm.com>
20027 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
20029 * gimple-expr.h (is_gimple_addressable: Likewise.
20030 * gimple-expr.c (is_gimple_address): Likewise.
20031 * internal-fn.c (expand_call_mem_ref): New function.
20032 (expand_mask_load_optab_fn): Use it.
20033 (expand_mask_store_optab_fn): Likewise.
20035 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20036 Alan Hayward <alan.hayward@arm.com>
20037 David Sherwood <david.sherwood@arm.com>
20039 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
20040 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
20041 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
20042 (cond_umax@var{mode}): Document.
20043 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
20044 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
20045 (cond_umin_optab, cond_umax_optab): New optabs.
20046 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
20047 (COND_IOR, COND_XOR): New internal functions.
20048 * internal-fn.h (get_conditional_internal_fn): Declare.
20049 * internal-fn.c (cond_binary_direct): New macro.
20050 (expand_cond_binary_optab_fn): Likewise.
20051 (direct_cond_binary_optab_supported_p): Likewise.
20052 (get_conditional_internal_fn): New function.
20053 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
20054 Cope with reduction statements that are vectorized as calls rather
20056 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
20057 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
20058 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
20059 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20060 (UNSPEC_COND_EOR): New unspecs.
20061 (optab): Add mappings for them.
20062 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
20063 (sve_int_op, sve_fp_op): New int attributes.
20065 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20066 Alan Hayward <alan.hayward@arm.com>
20067 David Sherwood <david.sherwood@arm.com>
20069 * optabs.def (while_ult_optab): New optab.
20070 * doc/md.texi (while_ult@var{m}@var{n}): Document.
20071 * internal-fn.def (WHILE_ULT): New internal function.
20072 * internal-fn.h (direct_internal_fn_supported_p): New override
20073 that takes two types as argument.
20074 * internal-fn.c (while_direct): New macro.
20075 (expand_while_optab_fn): New function.
20076 (convert_optab_supported_p): Likewise.
20077 (direct_while_optab_supported_p): New macro.
20078 * wide-int.h (wi::udiv_ceil): New function.
20079 * tree-vectorizer.h (rgroup_masks): New structure.
20080 (vec_loop_masks): New typedef.
20081 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
20082 and fully_masked_p.
20083 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
20084 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
20085 (vect_max_vf): New function.
20086 (slpeel_make_loop_iterate_ntimes): Delete.
20087 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
20088 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
20089 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
20090 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
20091 internal-fn.h, stor-layout.h and optabs-query.h.
20092 (vect_set_loop_mask): New function.
20093 (add_preheader_seq): Likewise.
20094 (add_header_seq): Likewise.
20095 (interleave_supported_p): Likewise.
20096 (vect_maybe_permute_loop_masks): Likewise.
20097 (vect_set_loop_masks_directly): Likewise.
20098 (vect_set_loop_condition_masked): Likewise.
20099 (vect_set_loop_condition_unmasked): New function, split out from
20100 slpeel_make_loop_iterate_ntimes.
20101 (slpeel_make_loop_iterate_ntimes): Rename to..
20102 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
20103 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
20104 (vect_do_peeling): Update call accordingly.
20105 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
20107 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
20108 mask_compare_type, can_fully_mask_p and fully_masked_p.
20109 (release_vec_loop_masks): New function.
20110 (_loop_vec_info): Use it to free the loop masks.
20111 (can_produce_all_loop_masks_p): New function.
20112 (vect_get_max_nscalars_per_iter): Likewise.
20113 (vect_verify_full_masking): Likewise.
20114 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
20115 retries, and free the mask rgroups before retrying. Check loop-wide
20116 reasons for disallowing fully-masked loops. Make the final decision
20117 about whether use a fully-masked loop or not.
20118 (vect_estimate_min_profitable_iters): Do not assume that peeling
20119 for the number of iterations will be needed for fully-masked loops.
20120 (vectorizable_reduction): Disable fully-masked loops.
20121 (vectorizable_live_operation): Likewise.
20122 (vect_halve_mask_nunits): New function.
20123 (vect_double_mask_nunits): Likewise.
20124 (vect_record_loop_mask): Likewise.
20125 (vect_get_loop_mask): Likewise.
20126 (vect_transform_loop): Handle the case in which the final loop
20127 iteration might handle a partial vector. Call vect_set_loop_condition
20128 instead of slpeel_make_loop_iterate_ntimes.
20129 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
20130 (check_load_store_masking): New function.
20131 (prepare_load_store_mask): Likewise.
20132 (vectorizable_store): Handle fully-masked loops.
20133 (vectorizable_load): Likewise.
20134 (supportable_widening_operation): Use vect_halve_mask_nunits for
20136 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
20137 (vect_gen_while): New function.
20138 * config/aarch64/aarch64.md (umax<mode>3): New expander.
20139 (aarch64_uqdec<mode>): New insn.
20141 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20142 Alan Hayward <alan.hayward@arm.com>
20143 David Sherwood <david.sherwood@arm.com>
20145 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
20146 (reduc_xor_scal_optab): New optabs.
20147 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
20148 (reduc_xor_scal_@var{m}): Document.
20149 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
20150 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
20151 internal functions.
20152 * fold-const-call.c (fold_const_call): Handle them.
20153 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
20154 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
20155 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
20156 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
20157 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
20158 (UNSPEC_XORV): New unspecs.
20159 (optab): Add entries for them.
20160 (BITWISEV): New int iterator.
20161 (bit_reduc_op): New int attributes.
20163 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20164 Alan Hayward <alan.hayward@arm.com>
20165 David Sherwood <david.sherwood@arm.com>
20167 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
20168 * internal-fn.def (VEC_SHL_INSERT): New internal function.
20169 * optabs.def (vec_shl_insert_optab): New optab.
20170 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
20171 (duplicate_and_interleave): Likewise.
20172 * tree-vect-loop.c: Include internal-fn.h.
20173 (neutral_op_for_slp_reduction): New function, split out from
20174 get_initial_defs_for_reduction.
20175 (get_initial_def_for_reduction): Handle option 2 for variable-length
20176 vectors by loading the neutral value into a vector and then shifting
20177 the initial value into element 0.
20178 (get_initial_defs_for_reduction): Replace the code argument with
20179 the neutral value calculated by neutral_op_for_slp_reduction.
20180 Use gimple_build_vector for constant-length vectors.
20181 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
20182 but the first group_size elements have a neutral value.
20183 Use duplicate_and_interleave otherwise.
20184 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
20185 Update call to get_initial_defs_for_reduction. Handle SLP
20186 reductions for variable-length vectors by creating one vector
20187 result for each scalar result, with the elements associated
20188 with other scalar results stubbed out with the neutral value.
20189 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
20190 Require IFN_VEC_SHL_INSERT for double reductions on
20191 variable-length vectors, or SLP reductions that have
20192 a neutral value. Require can_duplicate_and_interleave_p
20193 support for variable-length unchained SLP reductions if there
20194 is no neutral value, such as for MIN/MAX reductions. Also require
20195 the number of vector elements to be a multiple of the number of
20196 SLP statements when doing variable-length unchained SLP reductions.
20197 Update call to vect_create_epilog_for_reduction.
20198 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
20199 and remove initial values.
20200 (duplicate_and_interleave): Make public.
20201 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
20202 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
20204 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20205 Alan Hayward <alan.hayward@arm.com>
20206 David Sherwood <david.sherwood@arm.com>
20208 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
20209 (can_duplicate_and_interleave_p): New function.
20210 (vect_get_and_check_slp_defs): Take the vector of statements
20211 rather than just the current one. Remove excess parentheses.
20212 Restriction rejectinon of vect_constant_def and vect_external_def
20213 for variable-length vectors to boolean types, or types for which
20214 can_duplicate_and_interleave_p is false.
20215 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
20216 (duplicate_and_interleave): New function.
20217 (vect_get_constant_vectors): Use gimple_build_vector for
20218 constant-length vectors and suitable variable-length constant
20219 vectors. Use duplicate_and_interleave for other variable-length
20220 vectors. Don't defer the update when inserting new statements.
20222 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20223 Alan Hayward <alan.hayward@arm.com>
20224 David Sherwood <david.sherwood@arm.com>
20226 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
20227 min_profitable_iters doesn't go negative.
20229 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20230 Alan Hayward <alan.hayward@arm.com>
20231 David Sherwood <david.sherwood@arm.com>
20233 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
20234 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
20235 * optabs.def (vec_mask_load_lanes_optab): New optab.
20236 (vec_mask_store_lanes_optab): Likewise.
20237 * internal-fn.def (MASK_LOAD_LANES): New internal function.
20238 (MASK_STORE_LANES): Likewise.
20239 * internal-fn.c (mask_load_lanes_direct): New macro.
20240 (mask_store_lanes_direct): Likewise.
20241 (expand_mask_load_optab_fn): Handle masked operations.
20242 (expand_mask_load_lanes_optab_fn): New macro.
20243 (expand_mask_store_optab_fn): Handle masked operations.
20244 (expand_mask_store_lanes_optab_fn): New macro.
20245 (direct_mask_load_lanes_optab_supported_p): Likewise.
20246 (direct_mask_store_lanes_optab_supported_p): Likewise.
20247 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
20249 (vect_load_lanes_supported): Likewise.
20250 * tree-vect-data-refs.c (strip_conversion): New function.
20251 (can_group_stmts_p): Likewise.
20252 (vect_analyze_data_ref_accesses): Use it instead of checking
20253 for a pair of assignments.
20254 (vect_store_lanes_supported): Take a masked_p parameter.
20255 (vect_load_lanes_supported): Likewise.
20256 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
20257 vect_store_lanes_supported and vect_load_lanes_supported.
20258 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
20259 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
20260 parameter. Don't allow gaps for masked accesses.
20261 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
20262 and vect_load_lanes_supported.
20263 (get_load_store_type): Take a masked_p parameter and update
20264 call to get_group_load_store_type.
20265 (vectorizable_store): Update call to get_load_store_type.
20266 Handle IFN_MASK_STORE_LANES.
20267 (vectorizable_load): Update call to get_load_store_type.
20268 Handle IFN_MASK_LOAD_LANES.
20270 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20271 Alan Hayward <alan.hayward@arm.com>
20272 David Sherwood <david.sherwood@arm.com>
20274 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
20276 * config/aarch64/aarch64-protos.h
20277 (aarch64_sve_struct_memory_operand_p): Declare.
20278 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
20279 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
20280 (VPRED, vpred): Handle SVE structure modes.
20281 * config/aarch64/constraints.md (Utx): New constraint.
20282 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
20283 (aarch64_sve_struct_nonimmediate_operand): New predicates.
20284 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
20285 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
20286 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
20287 structure modes. Split into pieces after RA.
20288 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
20289 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
20291 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
20292 SVE structure modes.
20293 (aarch64_classify_address): Likewise.
20294 (sizetochar): Move earlier in file.
20295 (aarch64_print_operand): Handle SVE register lists.
20296 (aarch64_array_mode): New function.
20297 (aarch64_sve_struct_memory_operand_p): Likewise.
20298 (TARGET_ARRAY_MODE): Redefine.
20300 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20301 Alan Hayward <alan.hayward@arm.com>
20302 David Sherwood <david.sherwood@arm.com>
20304 * target.def (array_mode): New target hook.
20305 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
20306 * doc/tm.texi: Regenerate.
20307 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
20308 * hooks.c (hook_optmode_mode_uhwi_none): New function.
20309 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
20310 targetm.array_mode.
20311 * stor-layout.c (mode_for_array): Likewise. Support polynomial
20314 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20315 Alan Hayward <alan.hayward@arm.com>
20316 David Sherwood <david.sherwood@arm.com>
20318 * fold-const.c (fold_binary_loc): Check the argument types
20319 rather than the result type when testing for a vector operation.
20321 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20323 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
20324 * doc/tm.texi: Regenerate.
20326 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20327 Alan Hayward <alan.hayward@arm.com>
20328 David Sherwood <david.sherwood@arm.com>
20330 * doc/invoke.texi (-msve-vector-bits=): Document new option.
20331 (sve): Document new AArch64 extension.
20332 * doc/md.texi (w): Extend the description of the AArch64
20333 constraint to include SVE vectors.
20334 (Upl, Upa): Document new AArch64 predicate constraints.
20335 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
20337 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
20338 (msve-vector-bits=): New option.
20339 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
20340 SVE when these are disabled.
20341 (sve): New extension.
20342 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
20343 modes. Adjust their number of units based on aarch64_sve_vg.
20344 (MAX_BITSIZE_MODE_ANY_MODE): Define.
20345 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
20346 aarch64_addr_query_type.
20347 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
20348 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
20349 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
20350 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
20351 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
20352 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
20353 (aarch64_simd_imm_zero_p): Delete.
20354 (aarch64_check_zero_based_sve_index_immediate): Declare.
20355 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
20356 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
20357 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
20358 (aarch64_sve_float_mul_immediate_p): Likewise.
20359 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
20360 rather than an rtx.
20361 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
20362 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
20363 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
20364 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
20365 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
20366 (aarch64_regmode_natural_size): Likewise.
20367 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
20368 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
20370 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
20371 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
20372 for VG and the SVE predicate registers.
20373 (V_ALIASES): Add a "z"-prefixed alias.
20374 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
20375 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
20376 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
20377 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
20378 (REG_CLASS_NAMES): Add entries for them.
20379 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
20380 and the predicate registers.
20381 (aarch64_sve_vg): Declare.
20382 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
20383 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
20384 (REGMODE_NATURAL_SIZE): Define.
20385 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
20387 * config/aarch64/aarch64.c: Include cfgrtl.h.
20388 (simd_immediate_info): Add a constructor for series vectors,
20389 and an associated step field.
20390 (aarch64_sve_vg): New variable.
20391 (aarch64_dbx_register_number): Handle VG and the predicate registers.
20392 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
20393 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
20394 (VEC_ANY_DATA, VEC_STRUCT): New constants.
20395 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
20396 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
20397 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
20398 (aarch64_get_mask_mode): New functions.
20399 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
20400 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
20401 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
20402 predicate modes and predicate registers. Explicitly restrict
20403 GPRs to modes of 16 bytes or smaller. Only allow FP registers
20404 to store a vector mode if it is recognized by
20405 aarch64_classify_vector_mode.
20406 (aarch64_regmode_natural_size): New function.
20407 (aarch64_hard_regno_caller_save_mode): Return the original mode
20409 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
20410 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
20411 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
20412 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
20414 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
20415 does not overlap dest if the function is frame-related. Handle
20417 (aarch64_split_add_offset): New function.
20418 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
20419 them aarch64_add_offset.
20420 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
20421 and update call to aarch64_sub_sp.
20422 (aarch64_add_cfa_expression): New function.
20423 (aarch64_expand_prologue): Pass extra temporary registers to the
20424 functions above. Handle the case in which we need to emit new
20425 DW_CFA_expressions for registers that were originally saved
20426 relative to the stack pointer, but now have to be expressed
20427 relative to the frame pointer.
20428 (aarch64_output_mi_thunk): Pass extra temporary registers to the
20430 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
20431 IP0 and IP1 values for SVE frames.
20432 (aarch64_expand_vec_series): New function.
20433 (aarch64_expand_sve_widened_duplicate): Likewise.
20434 (aarch64_expand_sve_const_vector): Likewise.
20435 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
20436 Handle SVE constants. Use emit_move_insn to move a force_const_mem
20437 into the register, rather than emitting a SET directly.
20438 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
20439 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
20440 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
20441 (offset_9bit_signed_scaled_p): New functions.
20442 (aarch64_replicate_bitmask_imm): New function.
20443 (aarch64_bitmask_imm): Use it.
20444 (aarch64_cannot_force_const_mem): Reject expressions involving
20445 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
20446 (aarch64_classify_index): Handle SVE indices, by requiring
20447 a plain register index with a scale that matches the element size.
20448 (aarch64_classify_address): Handle SVE addresses. Assert that
20449 the mode of the address is VOIDmode or an integer mode.
20450 Update call to aarch64_classify_symbol.
20451 (aarch64_classify_symbolic_expression): Update call to
20452 aarch64_classify_symbol.
20453 (aarch64_const_vec_all_in_range_p): New function.
20454 (aarch64_print_vector_float_operand): Likewise.
20455 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
20456 "vN" for FP registers with SVE modes. Handle (const ...) vectors
20457 and the FP immediates 1.0 and 0.5.
20458 (aarch64_print_address_internal): Handle SVE addresses.
20459 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
20460 (aarch64_regno_regclass): Handle predicate registers.
20461 (aarch64_secondary_reload): Handle big-endian reloads of SVE
20463 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
20464 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
20465 (aarch64_convert_sve_vector_bits): New function.
20466 (aarch64_override_options): Use it to handle -msve-vector-bits=.
20467 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
20468 rather than an rtx.
20469 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
20470 Handle SVE vector and predicate modes. Accept VL-based constants
20471 that need only one temporary register, and VL offsets that require
20472 no temporary registers.
20473 (aarch64_conditional_register_usage): Mark the predicate registers
20474 as fixed if SVE isn't available.
20475 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
20476 Return true for SVE vector and predicate modes.
20477 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
20478 rather than an unsigned int. Handle SVE modes.
20479 (aarch64_preferred_simd_mode): Update call accordingly. Handle
20481 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
20483 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
20484 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
20485 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
20486 (aarch64_sve_float_mul_immediate_p): New functions.
20487 (aarch64_sve_valid_immediate): New function.
20488 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
20489 Explicitly reject structure modes. Check for INDEX constants.
20490 Handle PTRUE and PFALSE constants.
20491 (aarch64_check_zero_based_sve_index_immediate): New function.
20492 (aarch64_simd_imm_zero_p): Delete.
20493 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
20494 vector modes. Accept constants in the range of CNT[BHWD].
20495 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
20496 ask for an Advanced SIMD mode.
20497 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
20498 (aarch64_simd_vector_alignment): Handle SVE predicates.
20499 (aarch64_vectorize_preferred_vector_alignment): New function.
20500 (aarch64_simd_vector_alignment_reachable): Use it instead of
20502 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
20503 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
20505 (MAX_VECT_LEN): Delete.
20506 (expand_vec_perm_d): Add a vec_flags field.
20507 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
20508 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
20509 (aarch64_evpc_ext): Don't apply a big-endian lane correction
20511 (aarch64_evpc_rev): Rename to...
20512 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
20513 (aarch64_evpc_rev_global): New function.
20514 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
20515 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
20517 (aarch64_evpc_sve_tbl): New function.
20518 (aarch64_expand_vec_perm_const_1): Update after rename of
20519 aarch64_evpc_rev. Handle SVE permutes too, trying
20520 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
20521 than aarch64_evpc_tbl.
20522 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
20523 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
20524 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
20525 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
20526 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
20527 (aarch64_expand_sve_vcond): New functions.
20528 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
20529 of aarch64_vector_mode_p.
20530 (aarch64_dwarf_poly_indeterminate_value): New function.
20531 (aarch64_compute_pressure_classes): Likewise.
20532 (aarch64_can_change_mode_class): Likewise.
20533 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
20534 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
20535 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
20536 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
20537 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
20538 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
20539 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
20540 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
20542 (Dn, Dl, Dr): Accept const as well as const_vector.
20543 (Dz): Likewise. Compare against CONST0_RTX.
20544 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
20545 of "vector" where appropriate.
20546 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
20547 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
20548 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
20549 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
20550 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
20551 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
20552 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
20553 (v_int_equiv): Extend to SVE modes.
20554 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
20556 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
20557 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
20558 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
20559 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
20560 (SVE_COND_FP_CMP): New int iterators.
20561 (perm_hilo): Handle the new unpack unspecs.
20562 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
20564 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
20565 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
20566 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
20567 (aarch64_equality_operator, aarch64_constant_vector_operand)
20568 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
20569 (aarch64_sve_nonimmediate_operand): Likewise.
20570 (aarch64_sve_general_operand): Likewise.
20571 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
20572 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
20573 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
20574 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
20575 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
20576 (aarch64_sve_float_arith_immediate): Likewise.
20577 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
20578 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
20579 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
20580 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
20581 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
20582 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
20583 (aarch64_sve_float_arith_operand): Likewise.
20584 (aarch64_sve_float_arith_with_sub_operand): Likewise.
20585 (aarch64_sve_float_mul_operand): Likewise.
20586 (aarch64_sve_vec_perm_operand): Likewise.
20587 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
20588 (aarch64_mov_operand): Accept const_poly_int and const_vector.
20589 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
20590 as well as const_vector.
20591 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
20592 in file. Use CONST0_RTX and CONSTM1_RTX.
20593 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
20594 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
20595 Use aarch64_simd_imm_zero.
20596 * config/aarch64/aarch64-sve.md: New file.
20597 * config/aarch64/aarch64.md: Include it.
20598 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
20599 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
20600 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
20601 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
20602 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
20603 (sve): New attribute.
20604 (enabled): Disable instructions with the sve attribute unless
20606 (movqi, movhi): Pass CONST_POLY_INT operaneds through
20607 aarch64_expand_mov_immediate.
20608 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
20609 CNT[BHSD] immediates.
20610 (movti): Split CONST_POLY_INT moves into two halves.
20611 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
20612 Split additions that need a temporary here if the destination
20613 is the stack pointer.
20614 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
20615 (*add<mode>3_poly_1): New instruction.
20616 (set_clobber_cc): New expander.
20618 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20620 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
20621 parameter and use it instead of GET_MODE_SIZE (innermode). Use
20622 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
20623 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
20624 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
20625 Change innermode from fixed_mode_size to machine_mode.
20626 (simplify_subreg): Update call accordingly. Handle a constant-sized
20627 subreg of a variable-length CONST_VECTOR.
20629 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20630 Alan Hayward <alan.hayward@arm.com>
20631 David Sherwood <david.sherwood@arm.com>
20633 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
20634 (add_offset_to_base): New function, split out from...
20635 (create_mem_ref): ...here. When handling a scale other than 1,
20636 check first whether the address is valid without the offset.
20637 Add it into the base if so, leaving the index and scale as-is.
20639 2018-01-12 Jakub Jelinek <jakub@redhat.com>
20642 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
20643 fold_for_warn before checking if arg2 is INTEGER_CST.
20645 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
20647 * config/rs6000/predicates.md (load_multiple_operation): Delete.
20648 (store_multiple_operation): Delete.
20649 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
20650 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
20651 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
20652 guarded by TARGET_STRING.
20653 (rs6000_output_load_multiple): Delete.
20654 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20655 OPTION_MASK_STRING / TARGET_STRING handling.
20656 (print_operand) <'N', 'O'>: Add comment that these are unused now.
20657 (const rs6000_opt_masks) <"string">: Change mask to 0.
20658 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
20659 (MASK_STRING): Delete.
20660 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
20662 (load_multiple): Delete.
20669 (store_multiple): Delete.
20676 (movmemsi_8reg): Delete.
20677 (corresponding unnamed define_insn): Delete.
20678 (movmemsi_6reg): Delete.
20679 (corresponding unnamed define_insn): Delete.
20680 (movmemsi_4reg): Delete.
20681 (corresponding unnamed define_insn): Delete.
20682 (movmemsi_2reg): Delete.
20683 (corresponding unnamed define_insn): Delete.
20684 (movmemsi_1reg): Delete.
20685 (corresponding unnamed define_insn): Delete.
20686 * config/rs6000/rs6000.opt (mno-string): New.
20687 (mstring): Replace by deprecation warning stub.
20688 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
20690 2018-01-12 Jakub Jelinek <jakub@redhat.com>
20692 * regrename.c (regrename_do_replace): If replacing the same
20693 reg multiple times, try to reuse last created gen_raw_REG.
20696 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
20697 main to workaround a bug in GDB.
20699 2018-01-12 Tom de Vries <tom@codesourcery.com>
20702 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
20704 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
20706 PR rtl-optimization/80481
20707 * ira-color.c (get_cap_member): New function.
20708 (allocnos_conflict_by_live_ranges_p): Use it.
20709 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
20710 (setup_slot_coalesced_allocno_live_ranges): Ditto.
20712 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
20715 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
20716 (*saddl_se_1): Ditto.
20717 (*ssubsi_1): Ditto.
20718 (*ssubl_se_1): Ditto.
20720 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
20722 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
20723 rather than wi::to_widest for DR_INITs.
20724 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
20725 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
20726 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
20728 (vect_analyze_group_access_1): Note that here.
20730 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
20732 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
20733 polynomial type sizes.
20735 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
20737 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
20738 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
20739 (gimple_add_tmp_var): Likewise.
20741 2018-01-12 Martin Liska <mliska@suse.cz>
20743 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
20744 (gimple_alloc_sizes): Likewise.
20745 (dump_gimple_statistics): Use PRIu64 in printf format.
20746 * gimple.h: Change uint64_t to int.
20748 2018-01-12 Martin Liska <mliska@suse.cz>
20750 * tree-core.h: Use uint64_t instead of int.
20751 * tree.c (tree_node_counts): Likewise.
20752 (tree_node_sizes): Likewise.
20753 (dump_tree_statistics): Use PRIu64 in printf format.
20755 2018-01-12 Martin Liska <mliska@suse.cz>
20757 * Makefile.in: As qsort_chk is implemented in vec.c, add
20758 vec.o to linkage of gencfn-macros.
20759 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
20760 passing the info to record_node_allocation_statistics.
20761 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
20763 * ggc-common.c (struct ggc_usage): Add operator== and use
20764 it in operator< and compare function.
20765 * mem-stats.h (struct mem_usage): Likewise.
20766 * vec.c (struct vec_usage): Remove operator< and compare
20767 function. Can be simply inherited.
20769 2018-01-12 Martin Jambor <mjambor@suse.cz>
20772 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
20773 * tree-ssa-math-opts.c: Include domwalk.h.
20774 (convert_mult_to_fma_1): New function.
20775 (fma_transformation_info): New type.
20776 (fma_deferring_state): Likewise.
20777 (cancel_fma_deferring): New function.
20778 (result_of_phi): Likewise.
20779 (last_fma_candidate_feeds_initial_phi): Likewise.
20780 (convert_mult_to_fma): Added deferring logic, split actual
20781 transformation to convert_mult_to_fma_1.
20782 (math_opts_dom_walker): New type.
20783 (math_opts_dom_walker::after_dom_children): New method, body moved
20784 here from pass_optimize_widening_mul::execute, added deferring logic
20786 (pass_optimize_widening_mul::execute): Moved most of code to
20787 math_opts_dom_walker::after_dom_children.
20788 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
20789 * config/i386/i386.c (ix86_option_override_internal): Added
20790 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
20792 2018-01-12 Richard Biener <rguenther@suse.de>
20795 * dwarf2out.c (gen_variable_die): Do not reset old_die for
20796 inline instance vars.
20798 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
20801 * config/rx/rx.c (rx_is_restricted_memory_address):
20802 Handle SUBREG case.
20804 2018-01-12 Richard Biener <rguenther@suse.de>
20806 PR tree-optimization/80846
20807 * target.def (split_reduction): New target hook.
20808 * targhooks.c (default_split_reduction): New function.
20809 * targhooks.h (default_split_reduction): Declare.
20810 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
20811 target requests first reduce vectors by combining low and high
20813 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
20814 (get_vectype_for_scalar_type_and_size): Export.
20815 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
20816 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
20817 * doc/tm.texi: Regenerate.
20818 * config/i386/i386.c (ix86_split_reduction): Implement
20819 TARGET_VECTORIZE_SPLIT_REDUCTION.
20821 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20824 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
20825 in PIC mode except for TARGET_VXWORKS_RTP.
20826 * config/sparc/sparc.c: Include cfgrtl.h.
20827 (TARGET_INIT_PIC_REG): Define.
20828 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
20829 (sparc_pic_register_p): New predicate.
20830 (sparc_legitimate_address_p): Use it.
20831 (sparc_legitimize_pic_address): Likewise.
20832 (sparc_delegitimize_address): Likewise.
20833 (sparc_mode_dependent_address_p): Likewise.
20834 (gen_load_pcrel_sym): Remove 4th parameter.
20835 (load_got_register): Adjust call to above. Remove obsolete stuff.
20836 (sparc_expand_prologue): Do not call load_got_register here.
20837 (sparc_flat_expand_prologue): Likewise.
20838 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
20839 (sparc_use_pseudo_pic_reg): New function.
20840 (sparc_init_pic_reg): Likewise.
20841 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
20842 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
20844 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
20846 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20847 Add item for branch_cost.
20849 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20851 PR rtl-optimization/83565
20852 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
20853 not extend the result to a larger mode for rotate operations.
20854 (num_sign_bit_copies1): Likewise.
20856 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20859 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
20861 Use values-Xc.o for -pedantic.
20862 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
20864 2018-01-12 Martin Liska <mliska@suse.cz>
20867 * ipa-devirt.c (final_warning_record::grow_type_warnings):
20869 (possible_polymorphic_call_targets): Use it.
20870 (ipa_devirt): Likewise.
20872 2018-01-12 Martin Liska <mliska@suse.cz>
20874 * profile-count.h (enum profile_quality): Use 0 as invalid
20875 enum value of profile_quality.
20877 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
20879 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
20880 -mext-string options.
20882 2018-01-12 Richard Biener <rguenther@suse.de>
20884 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
20885 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
20886 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20888 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20890 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20892 * configure.ac (--with-long-double-format): Add support for the
20893 configuration option to change the default long double format on
20895 * config.gcc (powerpc*-linux*-*): Likewise.
20896 * configure: Regenerate.
20897 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
20898 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
20899 used without modification.
20901 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20903 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
20904 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
20905 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
20906 MISC_BUILTIN_SPEC_BARRIER.
20907 (rs6000_init_builtins): Likewise.
20908 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
20910 (speculation_barrier): New define_insn.
20911 * doc/extend.texi: Document __builtin_speculation_barrier.
20913 2018-01-11 Jakub Jelinek <jakub@redhat.com>
20916 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
20917 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
20918 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
20920 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
20921 integral modes instead of "ss" and "sd".
20922 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
20923 vectors with 32-bit and 64-bit elements.
20924 (vecdupssescalarmodesuffix): New mode attribute.
20925 (vec_dup<mode>): Use it.
20927 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
20930 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
20931 frame if argument is passed on stack.
20933 2018-01-11 Jakub Jelinek <jakub@redhat.com>
20936 * ree.c (combine_reaching_defs): Optimize also
20937 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
20938 reg2=any_extend(exp); reg1=reg2;, formatting fix.
20940 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
20942 PR middle-end/83189
20943 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
20945 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
20947 PR middle-end/83718
20948 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
20949 after they are computed.
20951 2018-01-11 Bin Cheng <bin.cheng@arm.com>
20953 PR tree-optimization/83695
20954 * gimple-loop-linterchange.cc
20955 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
20956 reset cached scev information after interchange.
20957 (pass_linterchange::execute): Remove call to scev_reset_htab.
20959 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20961 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
20962 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
20963 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
20964 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
20965 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
20966 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
20967 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
20968 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
20969 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
20970 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
20971 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
20972 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
20973 (V_lane_reg): Likewise.
20974 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
20976 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
20977 (vfmal_lane_low<mode>_intrinsic,
20978 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
20979 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
20980 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
20981 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
20982 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
20983 vfmsl_lane_high<mode>_intrinsic): New define_insns.
20985 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20987 * config/arm/arm-cpus.in (fp16fml): New feature.
20988 (ALL_SIMD): Add fp16fml.
20989 (armv8.2-a): Add fp16fml as an option.
20990 (armv8.3-a): Likewise.
20991 (armv8.4-a): Add fp16fml as part of fp16.
20992 * config/arm/arm.h (TARGET_FP16FML): Define.
20993 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
20995 * config/arm/arm-modes.def (V2HF): Define.
20996 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
20997 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
20998 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
20999 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
21000 vfmsl_low, vfmsl_high): New set of builtins.
21001 * config/arm/iterators.md (PLUSMINUS): New code iterator.
21002 (vfml_op): New code attribute.
21003 (VFMLHALVES): New int iterator.
21004 (VFML, VFMLSEL): New mode attributes.
21005 (V_reg): Define mapping for V2HF.
21006 (V_hi, V_lo): New mode attributes.
21007 (VF_constraint): Likewise.
21008 (vfml_half, vfml_half_selector): New int attributes.
21009 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
21011 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
21012 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
21014 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
21015 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
21016 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
21017 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
21019 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
21020 Document new effective target and option set.
21022 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21024 * config/arm/arm-cpus.in (armv8_4): New feature.
21025 (ARMv8_4a): New fgroup.
21026 (armv8.4-a): New arch.
21027 * config/arm/arm-tables.opt: Regenerate.
21028 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
21029 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
21030 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
21031 Add matching rules for -march=armv8.4-a and extensions.
21032 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
21034 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
21037 * config/rx/rx.md (BW): New mode attribute.
21038 (sync_lock_test_and_setsi): Add mode suffix to insn output.
21040 2018-01-11 Richard Biener <rguenther@suse.de>
21042 PR tree-optimization/83435
21043 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
21044 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
21045 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
21047 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21048 Alan Hayward <alan.hayward@arm.com>
21049 David Sherwood <david.sherwood@arm.com>
21051 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
21053 (aarch64_classify_address): Initialize it. Track polynomial offsets.
21054 (aarch64_print_address_internal): Use it to check for a zero offset.
21056 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21057 Alan Hayward <alan.hayward@arm.com>
21058 David Sherwood <david.sherwood@arm.com>
21060 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
21061 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
21062 Return a poly_int64 rather than a HOST_WIDE_INT.
21063 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
21064 rather than a HOST_WIDE_INT.
21065 * config/aarch64/aarch64.h (aarch64_frame): Protect with
21066 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
21067 hard_fp_offset, frame_size, initial_adjust, callee_offset and
21068 final_offset from HOST_WIDE_INT to poly_int64.
21069 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
21070 to_constant when getting the number of units in an Advanced SIMD
21072 (aarch64_builtin_vectorized_function): Check for a constant number
21074 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
21076 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
21077 attribute instead of GET_MODE_NUNITS.
21078 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
21079 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
21080 GET_MODE_SIZE for fixed-size registers.
21081 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
21082 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
21083 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
21084 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
21085 (aarch64_print_operand, aarch64_print_address_internal)
21086 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
21087 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
21088 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
21089 Handle polynomial GET_MODE_SIZE.
21090 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
21091 wider than SImode without modification.
21092 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
21093 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
21094 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
21095 passing and returning SVE modes.
21096 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
21097 rather than GEN_INT.
21098 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
21099 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
21100 (aarch64_allocate_and_probe_stack_space): Likewise.
21101 (aarch64_layout_frame): Cope with polynomial offsets.
21102 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
21103 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
21104 polynomial offsets.
21105 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
21106 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
21107 poly_int64 rather than a HOST_WIDE_INT.
21108 (aarch64_get_separate_components, aarch64_process_components)
21109 (aarch64_expand_prologue, aarch64_expand_epilogue)
21110 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
21111 (aarch64_anchor_offset): New function, split out from...
21112 (aarch64_legitimize_address): ...here.
21113 (aarch64_builtin_vectorization_cost): Handle polynomial
21114 TYPE_VECTOR_SUBPARTS.
21115 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
21117 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
21118 number of elements from the PARALLEL rather than the mode.
21119 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
21120 rather than GET_MODE_BITSIZE.
21121 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
21122 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
21123 (aarch64_expand_vec_perm_const_1): Handle polynomial
21124 d->perm.length () and d->perm elements.
21125 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
21126 Apply to_constant to d->perm elements.
21127 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
21128 polynomial CONST_VECTOR_NUNITS.
21129 (aarch64_move_pointer): Take amount as a poly_int64 rather
21131 (aarch64_progress_pointer): Avoid temporary variable.
21132 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
21133 the mode attribute instead of GET_MODE.
21135 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21136 Alan Hayward <alan.hayward@arm.com>
21137 David Sherwood <david.sherwood@arm.com>
21139 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
21140 x exists before using it.
21141 (aarch64_add_constant_internal): Rename to...
21142 (aarch64_add_offset_1): ...this. Replace regnum with separate
21143 src and dest rtxes. Handle the case in which they're different,
21144 including when the offset is zero. Replace scratchreg with an rtx.
21145 Use 2 additions if there is no spare register into which we can
21146 move a 16-bit constant.
21147 (aarch64_add_constant): Delete.
21148 (aarch64_add_offset): Replace reg with separate src and dest
21149 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
21150 Use aarch64_add_offset_1.
21151 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
21152 an rtx rather than an int. Take the delta as a poly_int64
21153 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
21154 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
21155 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
21156 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
21157 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
21158 and aarch64_add_sp.
21159 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
21160 aarch64_add_constant.
21162 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21164 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21165 Use scalar_float_mode.
21167 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21169 * config/aarch64/aarch64-simd.md
21170 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
21171 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
21172 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
21173 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
21174 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
21175 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
21176 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
21177 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
21178 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
21179 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
21181 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21184 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
21185 targ_options->x_arm_arch_string is non NULL.
21187 2018-01-11 Tamar Christina <tamar.christina@arm.com>
21189 * config/aarch64/aarch64.h
21190 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
21192 2018-01-11 Sudakshina Das <sudi.das@arm.com>
21195 * expmed.c (emit_store_flag_force): Swap if const op0
21196 and change VOIDmode to mode of op0.
21198 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21200 PR rtl-optimization/83761
21201 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
21202 than bytes to mode_for_size.
21204 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
21206 PR middle-end/83189
21207 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
21208 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
21211 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
21213 PR middle-end/83575
21214 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
21215 when in layout mode.
21216 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
21217 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
21220 2018-01-10 Michael Collison <michael.collison@arm.com>
21222 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
21223 * config/aarch64/aarch64-option-extension.def: Add
21224 AARCH64_OPT_EXTENSION of 'fp16fml'.
21225 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
21226 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
21227 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
21228 * config/aarch64/constraints.md (Ui7): New constraint.
21229 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
21230 (VFMLA_SEL_W): Ditto.
21233 (VFMLA16_LOW): New int iterator.
21234 (VFMLA16_HIGH): Ditto.
21235 (UNSPEC_FMLAL): New unspec.
21236 (UNSPEC_FMLSL): Ditto.
21237 (UNSPEC_FMLAL2): Ditto.
21238 (UNSPEC_FMLSL2): Ditto.
21239 (f16mac): New code attribute.
21240 * config/aarch64/aarch64-simd-builtins.def
21241 (aarch64_fmlal_lowv2sf): Ditto.
21242 (aarch64_fmlsl_lowv2sf): Ditto.
21243 (aarch64_fmlalq_lowv4sf): Ditto.
21244 (aarch64_fmlslq_lowv4sf): Ditto.
21245 (aarch64_fmlal_highv2sf): Ditto.
21246 (aarch64_fmlsl_highv2sf): Ditto.
21247 (aarch64_fmlalq_highv4sf): Ditto.
21248 (aarch64_fmlslq_highv4sf): Ditto.
21249 (aarch64_fmlal_lane_lowv2sf): Ditto.
21250 (aarch64_fmlsl_lane_lowv2sf): Ditto.
21251 (aarch64_fmlal_laneq_lowv2sf): Ditto.
21252 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
21253 (aarch64_fmlalq_lane_lowv4sf): Ditto.
21254 (aarch64_fmlsl_lane_lowv4sf): Ditto.
21255 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
21256 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
21257 (aarch64_fmlal_lane_highv2sf): Ditto.
21258 (aarch64_fmlsl_lane_highv2sf): Ditto.
21259 (aarch64_fmlal_laneq_highv2sf): Ditto.
21260 (aarch64_fmlsl_laneq_highv2sf): Ditto.
21261 (aarch64_fmlalq_lane_highv4sf): Ditto.
21262 (aarch64_fmlsl_lane_highv4sf): Ditto.
21263 (aarch64_fmlalq_laneq_highv4sf): Ditto.
21264 (aarch64_fmlsl_laneq_highv4sf): Ditto.
21265 * config/aarch64/aarch64-simd.md:
21266 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
21267 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
21268 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
21269 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
21270 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
21271 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
21272 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
21273 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
21274 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
21275 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
21276 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
21277 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
21278 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
21279 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
21280 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
21281 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
21282 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
21283 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
21284 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
21285 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
21286 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
21287 (vfmlsl_low_u32): Ditto.
21288 (vfmlalq_low_u32): Ditto.
21289 (vfmlslq_low_u32): Ditto.
21290 (vfmlal_high_u32): Ditto.
21291 (vfmlsl_high_u32): Ditto.
21292 (vfmlalq_high_u32): Ditto.
21293 (vfmlslq_high_u32): Ditto.
21294 (vfmlal_lane_low_u32): Ditto.
21295 (vfmlsl_lane_low_u32): Ditto.
21296 (vfmlal_laneq_low_u32): Ditto.
21297 (vfmlsl_laneq_low_u32): Ditto.
21298 (vfmlalq_lane_low_u32): Ditto.
21299 (vfmlslq_lane_low_u32): Ditto.
21300 (vfmlalq_laneq_low_u32): Ditto.
21301 (vfmlslq_laneq_low_u32): Ditto.
21302 (vfmlal_lane_high_u32): Ditto.
21303 (vfmlsl_lane_high_u32): Ditto.
21304 (vfmlal_laneq_high_u32): Ditto.
21305 (vfmlsl_laneq_high_u32): Ditto.
21306 (vfmlalq_lane_high_u32): Ditto.
21307 (vfmlslq_lane_high_u32): Ditto.
21308 (vfmlalq_laneq_high_u32): Ditto.
21309 (vfmlslq_laneq_high_u32): Ditto.
21310 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
21311 (AARCH64_FL_FOR_ARCH8_4): New.
21312 (AARCH64_ISA_F16FML): New ISA flag.
21313 (TARGET_F16FML): New feature flag for fp16fml.
21314 (doc/invoke.texi): Document new fp16fml option.
21316 2018-01-10 Michael Collison <michael.collison@arm.com>
21318 * config/aarch64/aarch64-builtins.c:
21319 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
21320 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
21321 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
21322 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
21323 (AARCH64_ISA_SHA3): New ISA flag.
21324 (TARGET_SHA3): New feature flag for sha3.
21325 * config/aarch64/iterators.md (sha512_op): New int attribute.
21326 (CRYPTO_SHA512): New int iterator.
21327 (UNSPEC_SHA512H): New unspec.
21328 (UNSPEC_SHA512H2): Ditto.
21329 (UNSPEC_SHA512SU0): Ditto.
21330 (UNSPEC_SHA512SU1): Ditto.
21331 * config/aarch64/aarch64-simd-builtins.def
21332 (aarch64_crypto_sha512hqv2di): New builtin.
21333 (aarch64_crypto_sha512h2qv2di): Ditto.
21334 (aarch64_crypto_sha512su0qv2di): Ditto.
21335 (aarch64_crypto_sha512su1qv2di): Ditto.
21336 (aarch64_eor3qv8hi): Ditto.
21337 (aarch64_rax1qv2di): Ditto.
21338 (aarch64_xarqv2di): Ditto.
21339 (aarch64_bcaxqv8hi): Ditto.
21340 * config/aarch64/aarch64-simd.md:
21341 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
21342 (aarch64_crypto_sha512su0qv2di): Ditto.
21343 (aarch64_crypto_sha512su1qv2di): Ditto.
21344 (aarch64_eor3qv8hi): Ditto.
21345 (aarch64_rax1qv2di): Ditto.
21346 (aarch64_xarqv2di): Ditto.
21347 (aarch64_bcaxqv8hi): Ditto.
21348 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
21349 (vsha512h2q_u64): Ditto.
21350 (vsha512su0q_u64): Ditto.
21351 (vsha512su1q_u64): Ditto.
21352 (veor3q_u16): Ditto.
21353 (vrax1q_u64): Ditto.
21354 (vxarq_u64): Ditto.
21355 (vbcaxq_u16): Ditto.
21356 * config/arm/types.md (crypto_sha512): New type attribute.
21357 (crypto_sha3): Ditto.
21358 (doc/invoke.texi): Document new sha3 option.
21360 2018-01-10 Michael Collison <michael.collison@arm.com>
21362 * config/aarch64/aarch64-builtins.c:
21363 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
21364 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
21365 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
21366 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
21367 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
21368 (AARCH64_ISA_SM4): New ISA flag.
21369 (TARGET_SM4): New feature flag for sm4.
21370 * config/aarch64/aarch64-simd-builtins.def
21371 (aarch64_sm3ss1qv4si): Ditto.
21372 (aarch64_sm3tt1aq4si): Ditto.
21373 (aarch64_sm3tt1bq4si): Ditto.
21374 (aarch64_sm3tt2aq4si): Ditto.
21375 (aarch64_sm3tt2bq4si): Ditto.
21376 (aarch64_sm3partw1qv4si): Ditto.
21377 (aarch64_sm3partw2qv4si): Ditto.
21378 (aarch64_sm4eqv4si): Ditto.
21379 (aarch64_sm4ekeyqv4si): Ditto.
21380 * config/aarch64/aarch64-simd.md:
21381 (aarch64_sm3ss1qv4si): Ditto.
21382 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
21383 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
21384 (aarch64_sm4eqv4si): Ditto.
21385 (aarch64_sm4ekeyqv4si): Ditto.
21386 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
21387 (sm3part_op): Ditto.
21388 (CRYPTO_SM3TT): Ditto.
21389 (CRYPTO_SM3PART): Ditto.
21390 (UNSPEC_SM3SS1): New unspec.
21391 (UNSPEC_SM3TT1A): Ditto.
21392 (UNSPEC_SM3TT1B): Ditto.
21393 (UNSPEC_SM3TT2A): Ditto.
21394 (UNSPEC_SM3TT2B): Ditto.
21395 (UNSPEC_SM3PARTW1): Ditto.
21396 (UNSPEC_SM3PARTW2): Ditto.
21397 (UNSPEC_SM4E): Ditto.
21398 (UNSPEC_SM4EKEY): Ditto.
21399 * config/aarch64/constraints.md (Ui2): New constraint.
21400 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
21401 * config/arm/types.md (crypto_sm3): New type attribute.
21402 (crypto_sm4): Ditto.
21403 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
21404 (vsm3tt1aq_u32): Ditto.
21405 (vsm3tt1bq_u32): Ditto.
21406 (vsm3tt2aq_u32): Ditto.
21407 (vsm3tt2bq_u32): Ditto.
21408 (vsm3partw1q_u32): Ditto.
21409 (vsm3partw2q_u32): Ditto.
21410 (vsm4eq_u32): Ditto.
21411 (vsm4ekeyq_u32): Ditto.
21412 (doc/invoke.texi): Document new sm4 option.
21414 2018-01-10 Michael Collison <michael.collison@arm.com>
21416 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
21417 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
21418 (AARCH64_FL_FOR_ARCH8_4): New.
21419 (AARCH64_FL_V8_4): New flag.
21420 (doc/invoke.texi): Document new armv8.4-a option.
21422 2018-01-10 Michael Collison <michael.collison@arm.com>
21424 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
21425 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
21426 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
21427 * config/aarch64/aarch64-option-extension.def: Add
21428 AARCH64_OPT_EXTENSION of 'sha2'.
21429 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
21430 (crypto): Disable sha2 and aes if crypto disabled.
21431 (crypto): Enable aes and sha2 if enabled.
21432 (simd): Disable sha2 and aes if simd disabled.
21433 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
21435 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
21436 (TARGET_SHA2): New feature flag for sha2.
21437 (TARGET_AES): New feature flag for aes.
21438 * config/aarch64/aarch64-simd.md:
21439 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
21440 conditional on TARGET_AES.
21441 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
21442 (aarch64_crypto_sha1hsi): Make pattern conditional
21444 (aarch64_crypto_sha1hv4si): Ditto.
21445 (aarch64_be_crypto_sha1hv4si): Ditto.
21446 (aarch64_crypto_sha1su1v4si): Ditto.
21447 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
21448 (aarch64_crypto_sha1su0v4si): Ditto.
21449 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
21450 (aarch64_crypto_sha256su0v4si): Ditto.
21451 (aarch64_crypto_sha256su1v4si): Ditto.
21452 (doc/invoke.texi): Document new aes and sha2 options.
21454 2018-01-10 Martin Sebor <msebor@redhat.com>
21456 PR tree-optimization/83781
21457 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
21460 2018-01-11 Martin Sebor <msebor@gmail.com>
21461 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21463 PR tree-optimization/83501
21464 PR tree-optimization/81703
21466 * tree-ssa-strlen.c (get_string_cst): Rename...
21467 (get_string_len): ...to this. Handle global constants.
21468 (handle_char_store): Adjust.
21470 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
21471 Jim Wilson <jimw@sifive.com>
21473 * config/riscv/riscv-protos.h (riscv_output_return): New.
21474 * config/riscv/riscv.c (struct machine_function): New naked_p field.
21475 (riscv_attribute_table, riscv_output_return),
21476 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
21477 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
21478 (riscv_compute_frame_info): Only compute frame->mask if not a naked
21480 (riscv_expand_prologue): Add early return for naked function.
21481 (riscv_expand_epilogue): Likewise.
21482 (riscv_function_ok_for_sibcall): Return false for naked function.
21483 (riscv_set_current_function): New.
21484 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
21485 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
21486 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
21487 * doc/extend.texi (RISC-V Function Attributes): New.
21489 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
21491 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
21492 check for 128-bit long double before checking TCmode.
21493 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
21494 128-bit long doubles before checking TFmode or TCmode.
21495 (FLOAT128_IBM_P): Likewise.
21497 2018-01-10 Martin Sebor <msebor@redhat.com>
21499 PR tree-optimization/83671
21500 * builtins.c (c_strlen): Unconditionally return zero for the empty
21502 Use -Warray-bounds for warnings.
21503 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
21504 for non-constant array indices with COMPONENT_REF, arrays of
21505 arrays, and pointers to arrays.
21506 (gimple_fold_builtin_strlen): Determine and set length range for
21507 non-constant character arrays.
21509 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
21511 PR middle-end/81897
21512 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
21515 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
21517 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
21519 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
21522 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
21523 VECTOR_MEM_ALTIVEC_OR_VSX_P.
21524 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
21525 indexed_or_indirect_operand predicate.
21526 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
21527 (*vsx_le_perm_load_v8hi): Likewise.
21528 (*vsx_le_perm_load_v16qi): Likewise.
21529 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
21530 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
21531 (*vsx_le_perm_store_v8hi): Likewise.
21532 (*vsx_le_perm_store_v16qi): Likewise.
21533 (eight unnamed splitters): Likewise.
21535 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
21537 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
21538 * config/rs6000/emmintrin.h: Likewise.
21539 * config/rs6000/mmintrin.h: Likewise.
21540 * config/rs6000/xmmintrin.h: Likewise.
21542 2018-01-10 David Malcolm <dmalcolm@redhat.com>
21545 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
21547 * tree.c (tree_nop_conversion): Return true for location wrapper
21549 (maybe_wrap_with_location): New function.
21550 (selftest::check_strip_nops): New function.
21551 (selftest::test_location_wrappers): New function.
21552 (selftest::tree_c_tests): Call it.
21553 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
21554 (maybe_wrap_with_location): New decl.
21555 (EXPR_LOCATION_WRAPPER_P): New macro.
21556 (location_wrapper_p): New inline function.
21557 (tree_strip_any_location_wrapper): New inline function.
21559 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
21562 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
21563 stack_realign_offset for the largest alignment of stack slot
21565 (ix86_find_max_used_stack_alignment): New function.
21566 (ix86_finalize_stack_frame_flags): Use it. Set
21567 max_used_stack_alignment if we don't realign stack.
21568 * config/i386/i386.h (machine_function): Add
21569 max_used_stack_alignment.
21571 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
21573 * config/arm/arm.opt (-mbranch-cost): New option.
21574 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
21577 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
21580 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
21581 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
21583 2018-01-10 Richard Biener <rguenther@suse.de>
21586 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
21587 early out so it also covers the case where we have a non-NULL
21590 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
21592 PR tree-optimization/83753
21593 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
21594 for non-strided grouped accesses if the number of elements is 1.
21596 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
21599 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
21600 * i386.h (TARGET_USE_GATHER): Define.
21601 * x86-tune.def (X86_TUNE_USE_GATHER): New.
21603 2018-01-10 Martin Liska <mliska@suse.cz>
21606 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
21607 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
21609 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
21610 CLEANUP_NO_PARTITIONING is not set.
21612 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
21614 * doc/rtl.texi: Remove documentation of (const ...) wrappers
21615 for vectors, as a partial revert of r254296.
21616 * rtl.h (const_vec_p): Delete.
21617 (const_vec_duplicate_p): Don't test for vector CONSTs.
21618 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
21619 * expmed.c (make_tree): Likewise.
21622 * common.md (E, F): Use CONSTANT_P instead of checking for
21624 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
21625 checking for CONST_VECTOR.
21627 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
21629 PR middle-end/83575
21630 * predict.c (force_edge_cold): Handle in more sane way edges
21631 with no prediction.
21633 2018-01-09 Carl Love <cel@us.ibm.com>
21635 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
21637 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
21638 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
21639 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
21640 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
21641 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
21642 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
21643 * config/rs6000/rs6000-protos.h: Add extern defition for
21644 rs6000_generate_float2_double_code.
21645 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
21647 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
21648 (float2_v2df): Add define_expand.
21650 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
21653 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
21654 op_mode in the force_to_mode call.
21656 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
21658 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
21659 instead of checking each element individually.
21660 (aarch64_evpc_uzp): Likewise.
21661 (aarch64_evpc_zip): Likewise.
21662 (aarch64_evpc_ext): Likewise.
21663 (aarch64_evpc_rev): Likewise.
21664 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
21665 instead of checking each element individually. Return true without
21667 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
21668 whether all selected elements come from the same input, instead of
21669 checking each element individually. Remove calls to gen_rtx_REG,
21670 start_sequence and end_sequence and instead assert that no rtl is
21673 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
21675 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
21676 order of HIGH and CONST checks.
21678 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
21680 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
21681 if the destination isn't an SSA_NAME.
21683 2018-01-09 Richard Biener <rguenther@suse.de>
21685 PR tree-optimization/83668
21686 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
21688 (canonicalize_loop_form): ... here, renamed from ...
21689 (canonicalize_loop_closed_ssa_form): ... this and amended to
21690 swap successor edges for loop exit blocks to make us use
21691 the RPO order we need for initial schedule generation.
21693 2018-01-09 Joseph Myers <joseph@codesourcery.com>
21695 PR tree-optimization/64811
21696 * match.pd: When optimizing comparisons with Inf, avoid
21697 introducing or losing exceptions from comparisons with NaN.
21699 2018-01-09 Martin Liska <mliska@suse.cz>
21702 * asan.c (shadow_mem_size): Add gcc_assert.
21704 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
21706 Don't save registers in main().
21709 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
21710 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
21711 * config/avr/avr.c (avr_set_current_function): Don't error if
21712 naked, OS_task or OS_main are specified at the same time.
21713 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
21715 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
21717 * common/config/avr/avr-common.c (avr_option_optimization_table):
21718 Switch on -mmain-is-OS_task for optimizing compilations.
21720 2018-01-09 Richard Biener <rguenther@suse.de>
21722 PR tree-optimization/83572
21723 * graphite.c: Include cfganal.h.
21724 (graphite_transform_loops): Connect infinite loops to exit
21725 and remove fake edges at the end.
21727 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
21729 * ipa-inline.c (edge_badness): Revert accidental checkin.
21731 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
21734 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
21735 symbols; not inline clones.
21737 2018-01-09 Jakub Jelinek <jakub@redhat.com>
21740 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
21741 hard registers. Formatting fixes.
21743 PR preprocessor/83722
21744 * gcc.c (try_generate_repro): Pass
21745 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
21746 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
21749 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
21750 Kito Cheng <kito.cheng@gmail.com>
21752 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
21753 (riscv_leaf_function_p): Delete.
21754 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
21756 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21758 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
21760 (do_ifelse): New function.
21761 (do_isel): New function.
21762 (do_sub3): New function.
21763 (do_add3): New function.
21764 (do_load_mask_compare): New function.
21765 (do_overlap_load_compare): New function.
21766 (expand_compare_loop): New function.
21767 (expand_block_compare): Call expand_compare_loop() when appropriate.
21768 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
21769 option description.
21770 (-mblock-compare-inline-loop-limit): New option.
21772 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21775 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
21776 Reverse order of second and third operands in first alternative.
21777 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
21778 of first and second elements in UNSPEC_VPERMR vector.
21779 (altivec_expand_vec_perm_le): Likewise.
21781 2018-01-08 Jeff Law <law@redhat.com>
21783 PR rtl-optimizatin/81308
21784 * tree-switch-conversion.c (cfg_altered): New file scoped static.
21785 (process_switch): If group_case_labels makes a change, then set
21787 (pass_convert_switch::execute): If a switch is converted, then
21788 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
21790 PR rtl-optimization/81308
21791 * recog.c (split_all_insns): Conditionally cleanup the CFG after
21794 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
21796 PR target/83663 - Revert r255946
21797 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
21798 generation for cases where splatting a value is not useful.
21799 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
21800 across a vec_duplicate and a paradoxical subreg forming a vector
21801 mode to a vec_concat.
21803 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21805 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
21806 -march=armv8.3-a variants.
21807 * config/arm/t-multilib: Likewise.
21808 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
21810 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21812 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
21814 (cceq_ior_compare_complement): Give it a name so I can use it, and
21815 change boolean_or_operator predicate to boolean_operator so it can
21816 be used to generate a crand.
21817 (eqne): New code iterator.
21818 (bd/bd_neg): New code_attrs.
21819 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
21820 a single define_insn.
21821 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
21822 decrement (bdnzt/bdnzf/bdzt/bdzf).
21823 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
21824 with the new names of the branch decrement patterns, and added the
21825 names of the branch decrement conditional patterns.
21827 2018-01-08 Richard Biener <rguenther@suse.de>
21829 PR tree-optimization/83563
21830 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
21833 2018-01-08 Richard Biener <rguenther@suse.de>
21835 PR middle-end/83713
21836 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
21838 2018-01-08 Richard Biener <rguenther@suse.de>
21840 PR tree-optimization/83685
21841 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
21842 references to abnormals.
21844 2018-01-08 Richard Biener <rguenther@suse.de>
21847 * dwarf2out.c (output_indirect_strings): Handle empty
21848 skeleton_debug_str_hash.
21849 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
21851 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
21853 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
21854 (emit_store_direct): Likewise.
21855 (arc_trampoline_adjust_address): Likewise.
21856 (arc_asm_trampoline_template): New function.
21857 (arc_initialize_trampoline): Use asm_trampoline_template.
21858 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
21859 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
21860 * config/arc/arc.md (flush_icache): Delete pattern.
21862 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
21864 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
21865 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
21868 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
21871 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
21872 by not USED_FOR_TARGET.
21873 (make_pass_resolve_sw_modes): Likewise.
21875 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
21877 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
21880 2018-01-08 Richard Biener <rguenther@suse.de>
21882 PR middle-end/83580
21883 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
21885 2018-01-08 Richard Biener <rguenther@suse.de>
21887 PR middle-end/83517
21888 * match.pd ((t * 2) / 2) -> t): Add missing :c.
21890 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
21892 PR middle-end/81897
21893 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
21894 basic blocks with a small number of successors.
21895 (convert_control_dep_chain_into_preds): Improve handling of
21897 (dump_predicates): Split apart into...
21898 (dump_pred_chain): ...here...
21899 (dump_pred_info): ...and here.
21900 (can_one_predicate_be_invalidated_p): Add debugging printfs.
21901 (can_chain_union_be_invalidated_p): Improve check for invalidation
21903 (uninit_uses_cannot_happen): Avoid unnecessary if
21904 convert_control_dep_chain_into_preds yielded nothing.
21906 2018-01-06 Martin Sebor <msebor@redhat.com>
21908 PR tree-optimization/83640
21909 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
21910 subtracting negative offset from size.
21911 (builtin_access::overlap): Adjust offset bounds of the access to fall
21912 within the size of the object if possible.
21914 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
21916 PR rtl-optimization/83699
21917 * expmed.c (extract_bit_field_1): Restrict the vector usage of
21918 extract_bit_field_as_subreg to cases in which the extracted
21919 value is also a vector.
21921 * lra-constraints.c (process_alt_operands): Test for the equivalence
21922 substitutions when detecting a possible reload cycle.
21924 2018-01-06 Jakub Jelinek <jakub@redhat.com>
21927 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
21928 by default if flag_selective_schedling{,2}. Formatting fixes.
21930 PR rtl-optimization/83682
21931 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
21932 if it has non-VECTOR_MODE element mode.
21933 (vec_duplicate_p): Likewise.
21935 PR middle-end/83694
21936 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
21937 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
21939 2018-01-05 Jakub Jelinek <jakub@redhat.com>
21942 * config/i386/i386-builtin.def
21943 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
21944 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
21945 Require also OPTION_MASK_ISA_AVX512F in addition to
21946 OPTION_MASK_ISA_GFNI.
21947 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
21948 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
21949 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
21950 to OPTION_MASK_ISA_GFNI.
21951 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
21952 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
21953 OPTION_MASK_ISA_AVX512BW.
21954 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
21955 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
21956 addition to OPTION_MASK_ISA_GFNI.
21957 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
21958 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
21959 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
21960 to OPTION_MASK_ISA_GFNI.
21961 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
21962 a requirement for all ISAs rather than any of them with a few
21964 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
21966 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
21967 bitmasks to be enabled with 3 exceptions, instead of requiring any
21968 enabled ISA with lots of exceptions.
21969 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
21970 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
21971 Change avx512bw in isa attribute to avx512f.
21972 * config/i386/sgxintrin.h: Add license boilerplate.
21973 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
21974 to __AVX512F__ and __AVX512VL to __AVX512VL__.
21975 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
21976 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
21978 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
21979 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
21980 temporarily sse2 rather than sse if not enabled already.
21983 * config/i386/sse.md (VI248_VLBW): Rename to ...
21984 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
21985 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
21986 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
21987 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
21988 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
21989 mode iterator instead of VI248_VLBW.
21991 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
21993 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
21994 (record_modified): Skip clobbers; add debug output.
21995 (param_change_prob): Use sreal frequencies.
21997 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
21999 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22000 punt for user-aligned variables.
22002 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
22004 * tree-chrec.c (chrec_contains_symbols): Return true for
22007 2018-01-05 Sudakshina Das <sudi.das@arm.com>
22010 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
22011 of (x|y) == x for BICS pattern.
22013 2018-01-05 Jakub Jelinek <jakub@redhat.com>
22015 PR tree-optimization/83605
22016 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
22017 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
22020 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
22022 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
22023 * config/epiphany/rtems.h: New file.
22025 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22026 Uros Bizjak <ubizjak@gmail.com>
22029 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
22030 QIreg_operand instead of register_operand predicate.
22031 * config/i386/i386.c (ix86_rop_should_change_byte_p,
22032 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
22033 comments instead of -fmitigate[-_]rop.
22035 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22038 * cgraphunit.c (symbol_table::compile): Switch to text_section
22039 before calling assembly_start debug hook.
22040 * run-rtl-passes.c (run_rtl_passes): Likewise.
22043 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22045 * tree-vrp.c (extract_range_from_binary_expr_1): Check
22046 range_int_cst_p rather than !symbolic_range_p before calling
22047 extract_range_from_multiplicative_op_1.
22049 2018-01-04 Jeff Law <law@redhat.com>
22051 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
22052 redundant test in assertion.
22054 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22056 * doc/rtl.texi: Document machine_mode wrapper classes.
22058 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22060 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
22061 using tree_to_uhwi.
22063 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22065 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
22066 the VEC_PERM_EXPR fold to fail.
22068 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22071 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
22072 to switched_sections.
22074 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22077 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
22078 test for d.testing.
22080 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
22083 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
22084 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
22086 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22089 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
22090 is BLKmode and bitpos not zero or mode change is needed.
22092 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22095 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
22098 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
22101 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
22102 instead of MULT rtx. Update all corresponding splitters.
22103 (*saddl_se): Ditto.
22104 (*ssub<modesuffix>): Ditto.
22105 (*ssubl_se): Ditto.
22106 (*cmp_sadd_di): Update split patterns.
22107 (*cmp_sadd_si): Ditto.
22108 (*cmp_sadd_sidi): Ditto.
22109 (*cmp_ssub_di): Ditto.
22110 (*cmp_ssub_si): Ditto.
22111 (*cmp_ssub_sidi): Ditto.
22112 * config/alpha/predicates.md (const23_operand): New predicate.
22113 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
22114 Look for ASHIFT, not MULT inner operand.
22115 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
22117 2018-01-04 Martin Liska <mliska@suse.cz>
22119 PR gcov-profile/83669
22120 * gcov.c (output_intermediate_file): Add version to intermediate
22122 * doc/gcov.texi: Document new field 'version' in intermediate
22123 file format. Fix location of '-k' option of gcov command.
22125 2018-01-04 Martin Liska <mliska@suse.cz>
22128 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
22130 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22132 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
22134 2018-01-03 Martin Sebor <msebor@redhat.com>
22136 PR tree-optimization/83655
22137 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
22138 checking calls with invalid arguments.
22140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22142 * tree-vect-stmts.c (vect_get_store_rhs): New function.
22143 (vectorizable_mask_load_store): Delete.
22144 (vectorizable_call): Return false for masked loads and stores.
22145 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
22146 instead of gimple_assign_rhs1.
22147 (vectorizable_load): Handle IFN_MASK_LOAD.
22148 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
22150 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22152 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
22154 (vectorizable_mask_load_store): ...here.
22155 (vectorizable_load): ...and here.
22157 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22159 * tree-vect-stmts.c (vect_build_all_ones_mask)
22160 (vect_build_zero_merge_argument): New functions, split out from...
22161 (vectorizable_load): ...here.
22163 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22165 * tree-vect-stmts.c (vect_check_store_rhs): New function,
22167 (vectorizable_mask_load_store): ...here.
22168 (vectorizable_store): ...and here.
22170 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22172 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
22174 (vectorizable_mask_load_store): ...here.
22176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22178 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
22179 (vect_model_store_cost): Take a vec_load_store_type instead of a
22181 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
22182 (vect_model_store_cost): Take a vec_load_store_type instead of a
22184 (vectorizable_mask_load_store): Update accordingly.
22185 (vectorizable_store): Likewise.
22186 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
22188 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22190 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
22191 IFN_MASK_LOAD calls here rather than...
22192 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
22194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22195 Alan Hayward <alan.hayward@arm.com>
22196 David Sherwood <david.sherwood@arm.com>
22198 * expmed.c (extract_bit_field_1): For vector extracts,
22199 fall back to extract_bit_field_as_subreg if vec_extract
22202 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22203 Alan Hayward <alan.hayward@arm.com>
22204 David Sherwood <david.sherwood@arm.com>
22206 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
22207 they are variable or constant sized.
22208 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
22209 slots for constant-sized data.
22211 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22212 Alan Hayward <alan.hayward@arm.com>
22213 David Sherwood <david.sherwood@arm.com>
22215 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
22216 handling COND_EXPRs with boolean comparisons, try to find a better
22217 basis for the mask type than the boolean itself.
22219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22221 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
22222 is calculated and how it can be overridden.
22223 * genmodes.c (max_bitsize_mode_any_mode): New variable.
22224 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
22226 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
22229 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22230 Alan Hayward <alan.hayward@arm.com>
22231 David Sherwood <david.sherwood@arm.com>
22233 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
22234 Remove the mode argument.
22235 (aarch64_simd_valid_immediate): Remove the mode and inverse
22237 * config/aarch64/iterators.md (bitsize): New iterator.
22238 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
22239 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
22240 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
22241 aarch64_simd_valid_immediate.
22242 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
22243 (aarch64_reg_or_bic_imm): Likewise.
22244 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
22245 with an insn_type enum and msl with a modifier_type enum.
22246 Replace element_width with a scalar_mode. Change the shift
22247 to unsigned int. Add constructors for scalar_float_mode and
22248 scalar_int_mode elements.
22249 (aarch64_vect_float_const_representable_p): Delete.
22250 (aarch64_can_const_movi_rtx_p)
22251 (aarch64_simd_scalar_immediate_valid_for_move)
22252 (aarch64_simd_make_constant): Update call to
22253 aarch64_simd_valid_immediate.
22254 (aarch64_advsimd_valid_immediate_hs): New function.
22255 (aarch64_advsimd_valid_immediate): Likewise.
22256 (aarch64_simd_valid_immediate): Remove mode and inverse
22257 arguments. Rewrite to use the above. Use const_vec_duplicate_p
22258 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
22259 and aarch64_float_const_representable_p on the result.
22260 (aarch64_output_simd_mov_immediate): Remove mode argument.
22261 Update call to aarch64_simd_valid_immediate and use of
22262 simd_immediate_info.
22263 (aarch64_output_scalar_simd_mov_immediate): Update call
22266 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22267 Alan Hayward <alan.hayward@arm.com>
22268 David Sherwood <david.sherwood@arm.com>
22270 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
22271 (mode_nunits): Likewise CONST_MODE_NUNITS.
22272 * machmode.def (ADJUST_NUNITS): Document.
22273 * genmodes.c (mode_data::need_nunits_adj): New field.
22274 (blank_mode): Update accordingly.
22275 (adj_nunits): New variable.
22276 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
22278 (emit_mode_size_inline): Set need_bytesize_adj for all modes
22279 listed in adj_nunits.
22280 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
22281 listed in adj_nunits. Don't emit case statements for such modes.
22282 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
22283 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
22284 nothing if adj_nunits is nonnull.
22285 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
22286 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
22287 (emit_mode_fbit): Update use of print_maybe_const_decl.
22288 (emit_move_size): Likewise. Treat the array as non-const
22290 (emit_mode_adjustments): Handle adj_nunits.
22292 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22294 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
22295 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
22296 (VECTOR_MODES): Use it.
22297 (make_vector_modes): Take the prefix as an argument.
22299 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22300 Alan Hayward <alan.hayward@arm.com>
22301 David Sherwood <david.sherwood@arm.com>
22303 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
22304 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
22305 for MODE_VECTOR_BOOL.
22306 * machmode.def (VECTOR_BOOL_MODE): Document.
22307 * genmodes.c (VECTOR_BOOL_MODE): New macro.
22308 (make_vector_bool_mode): New function.
22309 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
22311 * lto-streamer-in.c (lto_input_mode_table): Likewise.
22312 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
22314 * stor-layout.c (int_mode_for_mode): Likewise.
22315 * tree.c (build_vector_type_for_mode): Likewise.
22316 * varasm.c (output_constant_pool_2): Likewise.
22317 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
22318 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
22319 for MODE_VECTOR_BOOL.
22320 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
22321 of mode class checks.
22322 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
22323 instead of a list of mode class checks.
22324 (expand_vector_scalar_condition): Likewise.
22325 (type_for_widest_vector_mode): Handle BImode as an inner mode.
22327 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22328 Alan Hayward <alan.hayward@arm.com>
22329 David Sherwood <david.sherwood@arm.com>
22331 * machmode.h (mode_size): Change from unsigned short to
22333 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
22334 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
22335 or if measurement_type is not polynomial.
22336 (fixed_size_mode::includes_p): Check for constant-sized modes.
22337 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
22338 return a poly_uint16 rather than an unsigned short.
22339 (emit_mode_size): Change the type of mode_size from unsigned short
22340 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
22341 (emit_mode_adjustments): Cope with polynomial vector sizes.
22342 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
22344 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
22346 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
22347 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
22348 * caller-save.c (setup_save_areas): Likewise.
22349 (replace_reg_with_saved_mem): Likewise.
22350 * calls.c (emit_library_call_value_1): Likewise.
22351 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
22352 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
22353 (gen_lowpart_for_combine): Likewise.
22354 * convert.c (convert_to_integer_1): Likewise.
22355 * cse.c (equiv_constant, cse_insn): Likewise.
22356 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
22357 (cselib_subst_to_values): Likewise.
22358 * dce.c (word_dce_process_block): Likewise.
22359 * df-problems.c (df_word_lr_mark_ref): Likewise.
22360 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
22361 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
22362 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
22363 (rtl_for_decl_location): Likewise.
22364 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
22365 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
22366 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
22367 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
22368 (expand_expr_real_1): Likewise.
22369 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
22370 (pad_below): Likewise.
22371 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
22372 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
22373 * ira.c (get_subreg_tracking_sizes): Likewise.
22374 * ira-build.c (ira_create_allocno_objects): Likewise.
22375 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
22376 (ira_sort_regnos_for_alter_reg): Likewise.
22377 * ira-costs.c (record_operand_costs): Likewise.
22378 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
22379 (resolve_simple_move): Likewise.
22380 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
22381 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
22382 (lra_constraints): Likewise.
22383 (CONST_POOL_OK_P): Reject variable-sized modes.
22384 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
22385 (add_pseudo_to_slot, lra_spill): Likewise.
22386 * omp-low.c (omp_clause_aligned_alignment): Likewise.
22387 * optabs-query.c (get_best_extraction_insn): Likewise.
22388 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
22389 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
22390 (expand_mult_highpart, valid_multiword_target_p): Likewise.
22391 * recog.c (offsettable_address_addr_space_p): Likewise.
22392 * regcprop.c (maybe_mode_change): Likewise.
22393 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
22394 * regrename.c (build_def_use): Likewise.
22395 * regstat.c (dump_reg_info): Likewise.
22396 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
22397 (find_reloads, find_reloads_subreg_address): Likewise.
22398 * reload1.c (eliminate_regs_1): Likewise.
22399 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
22400 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
22401 (simplify_binary_operation_1, simplify_subreg): Likewise.
22402 * targhooks.c (default_function_arg_padding): Likewise.
22403 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
22404 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
22405 (verify_gimple_assign_ternary): Likewise.
22406 * tree-inline.c (estimate_move_cost): Likewise.
22407 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
22408 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
22409 (get_address_cost_ainc): Likewise.
22410 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22411 (vect_supportable_dr_alignment): Likewise.
22412 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22413 (vectorizable_reduction): Likewise.
22414 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
22415 (vectorizable_operation, vectorizable_load): Likewise.
22416 * tree.c (build_same_sized_truth_vector_type): Likewise.
22417 * valtrack.c (cleanup_auto_inc_dec): Likewise.
22418 * var-tracking.c (emit_note_insn_var_location): Likewise.
22419 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
22420 (ADDR_VEC_ALIGN): Likewise.
22422 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22423 Alan Hayward <alan.hayward@arm.com>
22424 David Sherwood <david.sherwood@arm.com>
22426 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
22428 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
22429 or if measurement_type is polynomial.
22430 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
22431 * combine.c (make_extraction): Likewise.
22432 * dse.c (find_shift_sequence): Likewise.
22433 * dwarf2out.c (mem_loc_descriptor): Likewise.
22434 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
22435 (extract_bit_field, extract_low_bits): Likewise.
22436 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
22437 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
22438 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
22439 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
22440 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
22441 * reload.c (find_reloads): Likewise.
22442 * reload1.c (alter_reg): Likewise.
22443 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
22444 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
22445 * tree-if-conv.c (predicate_mem_writes): Likewise.
22446 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
22447 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
22448 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22449 * valtrack.c (dead_debug_insert_temp): Likewise.
22450 * varasm.c (mergeable_constant_section): Likewise.
22451 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
22453 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22454 Alan Hayward <alan.hayward@arm.com>
22455 David Sherwood <david.sherwood@arm.com>
22457 * expr.c (expand_assignment): Cope with polynomial mode sizes
22458 when assigning to a CONCAT.
22460 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22461 Alan Hayward <alan.hayward@arm.com>
22462 David Sherwood <david.sherwood@arm.com>
22464 * machmode.h (mode_precision): Change from unsigned short to
22466 (mode_to_precision): Return a poly_uint16 rather than an unsigned
22468 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
22469 or if measurement_type is not polynomial.
22470 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
22471 in which the mode is already known to be a scalar_int_mode.
22472 * genmodes.c (emit_mode_precision): Change the type of mode_precision
22473 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
22475 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
22476 for GET_MODE_PRECISION.
22477 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
22478 for GET_MODE_PRECISION.
22479 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
22481 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
22482 (expand_field_assignment, make_extraction): Likewise.
22483 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
22484 (get_last_value): Likewise.
22485 * convert.c (convert_to_integer_1): Likewise.
22486 * cse.c (cse_insn): Likewise.
22487 * expr.c (expand_expr_real_1): Likewise.
22488 * lra-constraints.c (simplify_operand_subreg): Likewise.
22489 * optabs-query.c (can_atomic_load_p): Likewise.
22490 * optabs.c (expand_atomic_load): Likewise.
22491 (expand_atomic_store): Likewise.
22492 * ree.c (combine_reaching_defs): Likewise.
22493 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
22494 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
22495 * tree.h (type_has_mode_precision_p): Likewise.
22496 * ubsan.c (instrument_si_overflow): Likewise.
22498 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22499 Alan Hayward <alan.hayward@arm.com>
22500 David Sherwood <david.sherwood@arm.com>
22502 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
22503 polynomial numbers of units.
22504 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
22505 (valid_vector_subparts_p): New function.
22506 (build_vector_type): Remove temporary shim and take the number
22507 of units as a poly_uint64 rather than an int.
22508 (build_opaque_vector_type): Take the number of units as a
22509 poly_uint64 rather than an int.
22510 * tree.c (build_vector_from_ctor): Handle polynomial
22511 TYPE_VECTOR_SUBPARTS.
22512 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
22513 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
22514 (build_vector_from_val): If the number of units is variable,
22515 use build_vec_duplicate_cst for constant operands and
22516 VEC_DUPLICATE_EXPR otherwise.
22517 (make_vector_type): Remove temporary is_constant ().
22518 (build_vector_type, build_opaque_vector_type): Take the number of
22519 units as a poly_uint64 rather than an int.
22520 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
22522 * cfgexpand.c (expand_debug_expr): Likewise.
22523 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
22524 (store_constructor, expand_expr_real_1): Likewise.
22525 (const_scalar_mask_from_tree): Likewise.
22526 * fold-const-call.c (fold_const_reduction): Likewise.
22527 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
22528 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
22529 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
22530 (fold_relational_const): Likewise.
22531 (native_interpret_vector): Likewise. Change the size from an
22532 int to an unsigned int.
22533 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
22534 TYPE_VECTOR_SUBPARTS.
22535 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
22536 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
22537 duplicating a non-constant operand into a variable-length vector.
22538 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
22539 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
22540 * ipa-icf.c (sem_variable::equals): Likewise.
22541 * match.pd: Likewise.
22542 * omp-simd-clone.c (simd_clone_subparts): Likewise.
22543 * print-tree.c (print_node): Likewise.
22544 * stor-layout.c (layout_type): Likewise.
22545 * targhooks.c (default_builtin_vectorization_cost): Likewise.
22546 * tree-cfg.c (verify_gimple_comparison): Likewise.
22547 (verify_gimple_assign_binary): Likewise.
22548 (verify_gimple_assign_ternary): Likewise.
22549 (verify_gimple_assign_single): Likewise.
22550 * tree-pretty-print.c (dump_generic_node): Likewise.
22551 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
22552 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
22553 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
22554 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
22555 (vect_shift_permute_load_chain): Likewise.
22556 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
22557 (expand_vector_condition, optimize_vector_constructor): Likewise.
22558 (lower_vec_perm, get_compute_type): Likewise.
22559 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22560 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
22561 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
22562 (vect_recog_mask_conversion_pattern): Likewise.
22563 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
22564 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
22565 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
22566 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
22567 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
22568 (vectorizable_shift, vectorizable_operation, vectorizable_store)
22569 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
22570 (supportable_widening_operation): Likewise.
22571 (supportable_narrowing_operation): Likewise.
22572 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
22574 * varasm.c (output_constant): Likewise.
22576 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22577 Alan Hayward <alan.hayward@arm.com>
22578 David Sherwood <david.sherwood@arm.com>
22580 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
22581 so that both the length == 3 and length != 3 cases set up their
22582 own permute vectors. Add comments explaining why we know the
22583 number of elements is constant.
22584 (vect_permute_load_chain): Likewise.
22586 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22587 Alan Hayward <alan.hayward@arm.com>
22588 David Sherwood <david.sherwood@arm.com>
22590 * machmode.h (mode_nunits): Change from unsigned char to
22592 (ONLY_FIXED_SIZE_MODES): New macro.
22593 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
22594 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
22595 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
22597 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
22598 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
22599 or if measurement_type is not polynomial.
22600 * genmodes.c (ZERO_COEFFS): New macro.
22601 (emit_mode_nunits_inline): Make mode_nunits_inline return a
22603 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
22604 Use ZERO_COEFFS when emitting initializers.
22605 * data-streamer.h (bp_pack_poly_value): New function.
22606 (bp_unpack_poly_value): Likewise.
22607 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
22608 for GET_MODE_NUNITS.
22609 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
22610 for GET_MODE_NUNITS.
22611 * tree.c (make_vector_type): Remove temporary shim and make
22612 the real function take the number of units as a poly_uint64
22613 rather than an int.
22614 (build_vector_type_for_mode): Handle polynomial nunits.
22615 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
22616 * emit-rtl.c (const_vec_series_p_1): Likewise.
22617 (gen_rtx_CONST_VECTOR): Likewise.
22618 * fold-const.c (test_vec_duplicate_folding): Likewise.
22619 * genrecog.c (validate_pattern): Likewise.
22620 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
22621 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
22622 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
22623 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
22624 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
22625 * rtlanal.c (subreg_get_info): Likewise.
22626 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
22627 (vect_grouped_load_supported): Likewise.
22628 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
22629 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
22630 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
22631 (simplify_const_unary_operation, simplify_binary_operation_1)
22632 (simplify_const_binary_operation, simplify_ternary_operation)
22633 (test_vector_ops_duplicate, test_vector_ops): Likewise.
22634 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
22635 instead of CONST_VECTOR_NUNITS.
22636 * varasm.c (output_constant_pool_2): Likewise.
22637 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
22638 explicit-encoded elements in the XVEC for variable-length vectors.
22640 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22642 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
22644 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22645 Alan Hayward <alan.hayward@arm.com>
22646 David Sherwood <david.sherwood@arm.com>
22648 * coretypes.h (fixed_size_mode): Declare.
22649 (fixed_size_mode_pod): New typedef.
22650 * builtins.h (target_builtins::x_apply_args_mode)
22651 (target_builtins::x_apply_result_mode): Change type to
22652 fixed_size_mode_pod.
22653 * builtins.c (apply_args_size, apply_result_size, result_vector)
22654 (expand_builtin_apply_args_1, expand_builtin_apply)
22655 (expand_builtin_return): Update accordingly.
22657 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22659 * cse.c (hash_rtx_cb): Hash only the encoded elements.
22660 * cselib.c (cselib_hash_rtx): Likewise.
22661 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
22662 CONST_VECTOR encoding.
22664 2018-01-03 Jakub Jelinek <jakub@redhat.com>
22665 Jeff Law <law@redhat.com>
22668 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
22669 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
22670 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
22671 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
22674 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
22675 explicitly probe *sp in a noreturn function if there were any callee
22676 register saves or frame pointer is needed.
22678 2018-01-03 Jakub Jelinek <jakub@redhat.com>
22681 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
22682 BLKmode for ternary, binary or unary expressions.
22685 * var-tracking.c (delete_vta_debug_insn): New inline function.
22686 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
22687 insns from get_insns () to NULL instead of each bb separately.
22688 Use delete_vta_debug_insn. No longer static.
22689 (vt_debug_insns_local, variable_tracking_main_1): Adjust
22690 delete_vta_debug_insns callers.
22691 * rtl.h (delete_vta_debug_insns): Declare.
22692 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
22693 instead of variable_tracking_main.
22695 2018-01-03 Martin Sebor <msebor@redhat.com>
22697 PR tree-optimization/83603
22698 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
22699 arguments past the endof the argument list in functions declared
22700 without a prototype.
22701 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
22702 Avoid checking when arguments are null.
22704 2018-01-03 Martin Sebor <msebor@redhat.com>
22707 * doc/extend.texi (attribute const): Fix a typo.
22708 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
22709 issuing -Wsuggest-attribute for void functions.
22711 2018-01-03 Martin Sebor <msebor@redhat.com>
22713 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
22714 offset_int::from instead of wide_int::to_shwi.
22715 (maybe_diag_overlap): Remove assertion.
22716 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
22717 * gimple-ssa-sprintf.c (format_directive): Same.
22718 (parse_directive): Same.
22719 (sprintf_dom_walker::compute_format_length): Same.
22720 (try_substitute_return_value): Same.
22722 2018-01-03 Jeff Law <law@redhat.com>
22724 PR middle-end/83654
22725 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
22726 non-constant residual for zero at runtime and avoid probing in
22727 that case. Reorganize code for trailing problem to mirror handling
22730 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22732 PR tree-optimization/83501
22733 * tree-ssa-strlen.c (get_string_cst): New.
22734 (handle_char_store): Call get_string_cst.
22736 2018-01-03 Martin Liska <mliska@suse.cz>
22738 PR tree-optimization/83593
22739 * tree-ssa-strlen.c: Include tree-cfg.h.
22740 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
22741 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
22742 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
22744 (strlen_dom_walker::before_dom_children): Call
22745 gimple_purge_dead_eh_edges. Dump tranformation with details
22747 (strlen_dom_walker::before_dom_children): Update call by adding
22748 new argument cleanup_eh.
22749 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
22751 2018-01-03 Martin Liska <mliska@suse.cz>
22754 * cif-code.def (VARIADIC_THUNK): New enum value.
22755 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
22758 2018-01-03 Jan Beulich <jbeulich@suse.com>
22760 * sse.md (mov<mode>_internal): Tighten condition for when to use
22761 vmovdqu<ssescalarsize> for TI and OI modes.
22763 2018-01-03 Jakub Jelinek <jakub@redhat.com>
22765 Update copyright years.
22767 2018-01-03 Martin Liska <mliska@suse.cz>
22770 * ipa-visibility.c (function_and_variable_visibility): Skip
22771 functions with noipa attribure.
22773 2018-01-03 Jakub Jelinek <jakub@redhat.com>
22775 * gcc.c (process_command): Update copyright notice dates.
22776 * gcov-dump.c (print_version): Ditto.
22777 * gcov.c (print_version): Ditto.
22778 * gcov-tool.c (print_version): Ditto.
22779 * gengtype.c (create_file): Ditto.
22780 * doc/cpp.texi: Bump @copying's copyright year.
22781 * doc/cppinternals.texi: Ditto.
22782 * doc/gcc.texi: Ditto.
22783 * doc/gccint.texi: Ditto.
22784 * doc/gcov.texi: Ditto.
22785 * doc/install.texi: Ditto.
22786 * doc/invoke.texi: Ditto.
22788 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22790 * vector-builder.h (vector_builder::m_full_nelts): Change from
22791 unsigned int to poly_uint64.
22792 (vector_builder::full_nelts): Update prototype accordingly.
22793 (vector_builder::new_vector): Likewise.
22794 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
22795 (vector_builder::operator ==): Likewise.
22796 (vector_builder::finalize): Likewise.
22797 * int-vector-builder.h (int_vector_builder::int_vector_builder):
22798 Take the number of elements as a poly_uint64 rather than an
22800 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
22801 from unsigned int to poly_uint64.
22802 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
22803 (vec_perm_indices::new_vector): Likewise.
22804 (vec_perm_indices::length): Likewise.
22805 (vec_perm_indices::nelts_per_input): Likewise.
22806 (vec_perm_indices::input_nelts): Likewise.
22807 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
22808 number of elements per input as a poly_uint64 rather than an
22809 unsigned int. Use the original encoding for variable-length
22810 vectors, rather than clamping each individual element.
22811 For the second and subsequent elements in each pattern,
22812 clamp the step and base before clamping their sum.
22813 (vec_perm_indices::series_p): Handle polynomial element counts.
22814 (vec_perm_indices::all_in_range_p): Likewise.
22815 (vec_perm_indices_to_tree): Likewise.
22816 (vec_perm_indices_to_rtx): Likewise.
22817 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
22818 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
22819 (tree_vector_builder::new_binary_operation): Handle polynomial
22820 element counts. Return false if we need to know the number
22821 of elements at compile time.
22822 * fold-const.c (fold_vec_perm): Punt if the number of elements
22823 isn't known at compile time.
22825 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22827 * vec-perm-indices.h (vec_perm_builder): Change element type
22828 from HOST_WIDE_INT to poly_int64.
22829 (vec_perm_indices::element_type): Update accordingly.
22830 (vec_perm_indices::clamp): Handle polynomial element_types.
22831 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
22832 (vec_perm_indices::all_in_range_p): Likewise.
22833 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
22835 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
22836 polynomial vec_perm_indices element types.
22837 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
22838 * fold-const.c (fold_vec_perm): Likewise.
22839 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
22840 * tree-vect-generic.c (lower_vec_perm): Likewise.
22841 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
22842 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
22843 element type to HOST_WIDE_INT.
22845 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22846 Alan Hayward <alan.hayward@arm.com>
22847 David Sherwood <david.sherwood@arm.com>
22849 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
22850 rather than an int. Use plus_constant.
22851 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
22852 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
22854 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22855 Alan Hayward <alan.hayward@arm.com>
22856 David Sherwood <david.sherwood@arm.com>
22858 * calls.c (emit_call_1, expand_call): Change struct_value_size from
22859 a HOST_WIDE_INT to a poly_int64.
22861 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22862 Alan Hayward <alan.hayward@arm.com>
22863 David Sherwood <david.sherwood@arm.com>
22865 * calls.c (load_register_parameters): Cope with polynomial
22866 mode sizes. Require a constant size for BLKmode parameters
22867 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
22868 forces a parameter to be padded at the lsb end in order to
22869 fill a complete number of words, require the parameter size
22870 to be ordered wrt UNITS_PER_WORD.
22872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22873 Alan Hayward <alan.hayward@arm.com>
22874 David Sherwood <david.sherwood@arm.com>
22876 * reload1.c (spill_stack_slot_width): Change element type
22877 from unsigned int to poly_uint64_pod.
22878 (alter_reg): Treat mode sizes as polynomial.
22880 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22881 Alan Hayward <alan.hayward@arm.com>
22882 David Sherwood <david.sherwood@arm.com>
22884 * reload.c (complex_word_subreg_p): New function.
22885 (reload_inner_reg_of_subreg, push_reload): Use it.
22887 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22888 Alan Hayward <alan.hayward@arm.com>
22889 David Sherwood <david.sherwood@arm.com>
22891 * lra-constraints.c (process_alt_operands): Reject matched
22892 operands whose sizes aren't ordered.
22893 (match_reload): Refer to this check here.
22895 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22896 Alan Hayward <alan.hayward@arm.com>
22897 David Sherwood <david.sherwood@arm.com>
22899 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
22900 that the mode size is in the set {1, 2, 4, 8, 16}.
22902 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22903 Alan Hayward <alan.hayward@arm.com>
22904 David Sherwood <david.sherwood@arm.com>
22906 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
22907 Use plus_constant instead of gen_rtx_PLUS.
22909 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22910 Alan Hayward <alan.hayward@arm.com>
22911 David Sherwood <david.sherwood@arm.com>
22913 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
22914 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
22915 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
22916 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
22917 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
22918 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
22919 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
22920 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
22921 * config/i386/i386.c (ix86_push_rounding): ...this new function.
22922 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
22924 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
22925 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
22926 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
22927 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
22928 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
22929 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
22930 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
22931 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
22932 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
22933 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
22935 * expr.c (emit_move_resolve_push): Treat the input and result
22936 of PUSH_ROUNDING as a poly_int64.
22937 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
22938 (emit_push_insn): Likewise.
22939 * lra-eliminations.c (mark_not_eliminable): Likewise.
22940 * recog.c (push_operand): Likewise.
22941 * reload1.c (elimination_effects): Likewise.
22942 * rtlanal.c (nonzero_bits1): Likewise.
22943 * calls.c (store_one_arg): Likewise. Require the padding to be
22944 known at compile time.
22946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22947 Alan Hayward <alan.hayward@arm.com>
22948 David Sherwood <david.sherwood@arm.com>
22950 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
22951 Use plus_constant instead of gen_rtx_PLUS.
22953 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22954 Alan Hayward <alan.hayward@arm.com>
22955 David Sherwood <david.sherwood@arm.com>
22957 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
22958 rather than an int.
22960 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22961 Alan Hayward <alan.hayward@arm.com>
22962 David Sherwood <david.sherwood@arm.com>
22964 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
22965 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
22966 via stack temporaries. Treat the mode size as polynomial too.
22968 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22969 Alan Hayward <alan.hayward@arm.com>
22970 David Sherwood <david.sherwood@arm.com>
22972 * expr.c (expand_expr_real_2): When handling conversions involving
22973 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
22974 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
22975 as a poly_uint64 too.
22977 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22978 Alan Hayward <alan.hayward@arm.com>
22979 David Sherwood <david.sherwood@arm.com>
22981 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
22983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22984 Alan Hayward <alan.hayward@arm.com>
22985 David Sherwood <david.sherwood@arm.com>
22987 * combine.c (can_change_dest_mode): Handle polynomial
22988 REGMODE_NATURAL_SIZE.
22989 * expmed.c (store_bit_field_1): Likewise.
22990 * expr.c (store_constructor): Likewise.
22991 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
22992 and polynomial REGMODE_NATURAL_SIZE.
22993 (gen_lowpart_common): Likewise.
22994 * reginfo.c (record_subregs_of_mode): Likewise.
22995 * rtlanal.c (read_modify_subreg_p): Likewise.
22997 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22998 Alan Hayward <alan.hayward@arm.com>
22999 David Sherwood <david.sherwood@arm.com>
23001 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
23002 numbers of elements.
23004 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23005 Alan Hayward <alan.hayward@arm.com>
23006 David Sherwood <david.sherwood@arm.com>
23008 * match.pd: Cope with polynomial numbers of vector elements.
23010 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23011 Alan Hayward <alan.hayward@arm.com>
23012 David Sherwood <david.sherwood@arm.com>
23014 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
23015 in a POINTER_PLUS_EXPR.
23017 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23018 Alan Hayward <alan.hayward@arm.com>
23019 David Sherwood <david.sherwood@arm.com>
23021 * omp-simd-clone.c (simd_clone_subparts): New function.
23022 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
23023 (ipa_simd_modify_function_body): Likewise.
23025 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23026 Alan Hayward <alan.hayward@arm.com>
23027 David Sherwood <david.sherwood@arm.com>
23029 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
23030 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
23031 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
23032 (expand_vector_condition, vector_element): Likewise.
23033 (subparts_gt): New function.
23034 (get_compute_type): Use subparts_gt.
23035 (count_type_subparts): Delete.
23036 (expand_vector_operations_1): Use subparts_gt instead of
23037 count_type_subparts.
23039 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23040 Alan Hayward <alan.hayward@arm.com>
23041 David Sherwood <david.sherwood@arm.com>
23043 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
23044 (vect_compile_time_alias): ...this new function. Do the calculation
23045 on poly_ints rather than trees.
23046 (vect_prune_runtime_alias_test_list): Update call accordingly.
23048 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23049 Alan Hayward <alan.hayward@arm.com>
23050 David Sherwood <david.sherwood@arm.com>
23052 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
23054 (vect_schedule_slp_instance): Likewise.
23056 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23057 Alan Hayward <alan.hayward@arm.com>
23058 David Sherwood <david.sherwood@arm.com>
23060 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
23061 constant and extern definitions for variable-length vectors.
23062 (vect_get_constant_vectors): Note that the number of units
23063 is known to be constant.
23065 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23066 Alan Hayward <alan.hayward@arm.com>
23067 David Sherwood <david.sherwood@arm.com>
23069 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
23070 of units as polynomial. Choose between WIDE and NARROW based
23073 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23074 Alan Hayward <alan.hayward@arm.com>
23075 David Sherwood <david.sherwood@arm.com>
23077 * tree-vect-stmts.c (simd_clone_subparts): New function.
23078 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
23080 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23081 Alan Hayward <alan.hayward@arm.com>
23082 David Sherwood <david.sherwood@arm.com>
23084 * tree-vect-stmts.c (vectorizable_call): Treat the number of
23085 vectors as polynomial. Use build_index_vector for
23086 IFN_GOMP_SIMD_LANE.
23088 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23089 Alan Hayward <alan.hayward@arm.com>
23090 David Sherwood <david.sherwood@arm.com>
23092 * tree-vect-stmts.c (get_load_store_type): Treat the number of
23093 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
23094 for variable-length vectors.
23095 (vectorizable_mask_load_store): Treat the number of units as
23096 polynomial, asserting that it is constant if the condition has
23097 already been enforced.
23098 (vectorizable_store, vectorizable_load): Likewise.
23100 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23101 Alan Hayward <alan.hayward@arm.com>
23102 David Sherwood <david.sherwood@arm.com>
23104 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
23105 of units as polynomial. Punt if we can't tell at compile time
23106 which vector contains the final result.
23108 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23109 Alan Hayward <alan.hayward@arm.com>
23110 David Sherwood <david.sherwood@arm.com>
23112 * tree-vect-loop.c (vectorizable_induction): Treat the number
23113 of units as polynomial. Punt on SLP inductions. Use an integer
23114 VEC_SERIES_EXPR for variable-length integer reductions. Use a
23115 cast of such a series for variable-length floating-point
23118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23119 Alan Hayward <alan.hayward@arm.com>
23120 David Sherwood <david.sherwood@arm.com>
23122 * tree.h (build_index_vector): Declare.
23123 * tree.c (build_index_vector): New function.
23124 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
23125 of units as polynomial, forcibly converting it to a constant if
23126 vectorizable_reduction has already enforced the condition.
23127 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
23128 to create a {1,2,3,...} vector.
23129 (vectorizable_reduction): Treat the number of units as polynomial.
23130 Choose vectype_in based on the largest scalar element size rather
23131 than the smallest number of units. Enforce the restrictions
23134 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23135 Alan Hayward <alan.hayward@arm.com>
23136 David Sherwood <david.sherwood@arm.com>
23138 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
23139 number of units as polynomial.
23141 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23142 Alan Hayward <alan.hayward@arm.com>
23143 David Sherwood <david.sherwood@arm.com>
23145 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
23146 * target.def (autovectorize_vector_sizes): Return the vector sizes
23147 by pointer, using vector_sizes rather than a bitmask.
23148 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
23149 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
23150 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
23152 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
23153 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
23154 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
23155 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
23156 * omp-general.c (omp_max_vf): Likewise.
23157 * omp-low.c (omp_clause_aligned_alignment): Likewise.
23158 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
23159 * tree-vect-loop.c (vect_analyze_loop): Likewise.
23160 * tree-vect-slp.c (vect_slp_bb): Likewise.
23161 * doc/tm.texi: Regenerate.
23162 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
23164 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
23165 the vector size as a poly_uint64 rather than an unsigned int.
23166 (current_vector_size): Change from an unsigned int to a poly_uint64.
23167 (get_vectype_for_scalar_type): Update accordingly.
23168 * tree.h (build_truth_vector_type): Take the size and number of
23169 units as a poly_uint64 rather than an unsigned int.
23170 (build_vector_type): Add a temporary overload that takes
23171 the number of units as a poly_uint64 rather than an unsigned int.
23172 * tree.c (make_vector_type): Likewise.
23173 (build_truth_vector_type): Take the number of units as a poly_uint64
23174 rather than an unsigned int.
23176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23177 Alan Hayward <alan.hayward@arm.com>
23178 David Sherwood <david.sherwood@arm.com>
23180 * target.def (get_mask_mode): Take the number of units and length
23181 as poly_uint64s rather than unsigned ints.
23182 * targhooks.h (default_get_mask_mode): Update accordingly.
23183 * targhooks.c (default_get_mask_mode): Likewise.
23184 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
23185 * doc/tm.texi: Regenerate.
23187 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23188 Alan Hayward <alan.hayward@arm.com>
23189 David Sherwood <david.sherwood@arm.com>
23191 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
23192 * omp-general.c (omp_max_vf): Likewise.
23193 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
23194 (expand_omp_simd): Handle polynomial safelen.
23195 * omp-low.c (omplow_simd_context): Add a default constructor.
23196 (omplow_simd_context::max_vf): Change from int to poly_uint64.
23197 (lower_rec_simd_input_clauses): Update accordingly.
23198 (lower_rec_input_clauses): Likewise.
23200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23201 Alan Hayward <alan.hayward@arm.com>
23202 David Sherwood <david.sherwood@arm.com>
23204 * tree-vectorizer.h (vect_nunits_for_cost): New function.
23205 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
23206 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
23207 (vect_analyze_slp_cost): Likewise.
23208 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
23209 (vect_model_load_cost): Likewise.
23211 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23212 Alan Hayward <alan.hayward@arm.com>
23213 David Sherwood <david.sherwood@arm.com>
23215 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
23216 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
23217 from an unsigned int * to a poly_uint64_pod *.
23218 (calculate_unrolling_factor): New function.
23219 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
23221 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23222 Alan Hayward <alan.hayward@arm.com>
23223 David Sherwood <david.sherwood@arm.com>
23225 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
23226 from an unsigned int to a poly_uint64.
23227 (_loop_vec_info::slp_unrolling_factor): Likewise.
23228 (_loop_vec_info::vectorization_factor): Change from an int
23230 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
23231 (vect_get_num_vectors): New function.
23232 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
23233 (vect_get_num_copies): Use vect_get_num_vectors.
23234 (vect_analyze_data_ref_dependences): Change max_vf from an int *
23235 to an unsigned int *.
23236 (vect_analyze_data_refs): Change min_vf from an int * to a
23238 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
23239 than an unsigned HOST_WIDE_INT.
23240 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
23241 (vect_analyze_data_ref_dependence): Change max_vf from an int *
23242 to an unsigned int *.
23243 (vect_analyze_data_ref_dependences): Likewise.
23244 (vect_compute_data_ref_alignment): Handle polynomial vf.
23245 (vect_enhance_data_refs_alignment): Likewise.
23246 (vect_prune_runtime_alias_test_list): Likewise.
23247 (vect_shift_permute_load_chain): Likewise.
23248 (vect_supportable_dr_alignment): Likewise.
23249 (dependence_distance_ge_vf): Take the vectorization factor as a
23250 poly_uint64 rather than an unsigned HOST_WIDE_INT.
23251 (vect_analyze_data_refs): Change min_vf from an int * to a
23253 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
23254 vfm1 as a poly_uint64 rather than an int. Make the same change
23255 for the returned bound_scalar.
23256 (vect_gen_vector_loop_niters): Handle polynomial vf.
23257 (vect_do_peeling): Likewise. Update call to
23258 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
23259 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
23261 * tree-vect-loop.c (vect_determine_vectorization_factor)
23262 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
23263 (vect_get_known_peeling_cost): Likewise.
23264 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
23265 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
23266 (vect_transform_loop): Likewise. Use the lowest possible VF when
23267 updating the upper bounds of the loop.
23268 (vect_min_worthwhile_factor): Make static. Return an unsigned int
23269 rather than an int.
23270 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
23271 polynomial unroll factors.
23272 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
23273 (vect_make_slp_decision): Likewise.
23274 (vect_supported_load_permutation_p): Likewise, and polynomial
23276 (vect_analyze_slp_cost): Handle polynomial vf.
23277 (vect_slp_analyze_node_operations): Likewise.
23278 (vect_slp_analyze_bb_1): Likewise.
23279 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
23280 than an unsigned HOST_WIDE_INT.
23281 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
23282 (vectorizable_load): Handle polynomial vf.
23283 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
23285 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
23287 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23288 Alan Hayward <alan.hayward@arm.com>
23289 David Sherwood <david.sherwood@arm.com>
23291 * match.pd: Handle bit operations involving three constants
23292 and try to fold one pair.
23294 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23296 * tree-vect-loop-manip.c: Include gimple-fold.h.
23297 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
23298 niters_maybe_zero parameters. Handle other cases besides a step of 1.
23299 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
23300 Add a path that uses a step of VF instead of 1, but disable it
23302 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
23303 and niters_no_overflow parameters. Update calls to
23304 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
23305 Create a new SSA name if the latter choses to use a ste other
23306 than zero, and return it via niters_vector_mult_vf_var.
23307 * tree-vect-loop.c (vect_transform_loop): Update calls to
23308 vect_do_peeling, vect_gen_vector_loop_niters and
23309 slpeel_make_loop_iterate_ntimes.
23310 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
23311 (vect_gen_vector_loop_niters): Update declarations after above changes.
23313 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
23315 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
23316 128-bit round to integer instructions.
23317 (ceil<mode>2): Likewise.
23318 (btrunc<mode>2): Likewise.
23319 (round<mode>2): Likewise.
23321 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23323 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
23324 unaligned VSX load/store on P8/P9.
23325 (expand_block_clear): Allow the use of unaligned VSX
23326 load/store on P8/P9.
23328 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23330 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
23332 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
23333 swap associated with both a load and a store.
23335 2018-01-02 Andrew Waterman <andrew@sifive.com>
23337 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
23338 * config/riscv/riscv.md (clear_cache): Use it.
23340 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
23342 * web.c: Remove out-of-date comment.
23344 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23346 * expr.c (fixup_args_size_notes): Check that any existing
23347 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
23348 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
23349 (emit_single_push_insn): ...here.
23351 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23353 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
23354 (const_vector_encoded_nelts): New function.
23355 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
23356 (const_vector_int_elt, const_vector_elt): Declare.
23357 * emit-rtl.c (const_vector_int_elt_1): New function.
23358 (const_vector_elt): Likewise.
23359 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
23360 of CONST_VECTOR_ELT.
23362 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23364 * expr.c: Include rtx-vector-builder.h.
23365 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
23366 directly on the tree encoding.
23367 (const_vector_from_tree): Likewise.
23368 * optabs.c: Include rtx-vector-builder.h.
23369 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
23370 sequence of "u" values.
23371 * vec-perm-indices.c: Include rtx-vector-builder.h.
23372 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
23373 directly on the vec_perm_indices encoding.
23375 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23377 * doc/rtl.texi (const_vector): Describe new encoding scheme.
23378 * Makefile.in (OBJS): Add rtx-vector-builder.o.
23379 * rtx-vector-builder.h: New file.
23380 * rtx-vector-builder.c: Likewise.
23381 * rtl.h (rtx_def::u2): Add a const_vector field.
23382 (CONST_VECTOR_NPATTERNS): New macro.
23383 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
23384 (CONST_VECTOR_DUPLICATE_P): Likewise.
23385 (CONST_VECTOR_STEPPED_P): Likewise.
23386 (CONST_VECTOR_ENCODED_ELT): Likewise.
23387 (const_vec_duplicate_p): Check for a duplicated vector encoding.
23388 (unwrap_const_vec_duplicate): Likewise.
23389 (const_vec_series_p): Check for a non-duplicated vector encoding.
23390 Say that the function only returns true for integer vectors.
23391 * emit-rtl.c: Include rtx-vector-builder.h.
23392 (gen_const_vec_duplicate_1): Delete.
23393 (gen_const_vector): Call gen_const_vec_duplicate instead of
23394 gen_const_vec_duplicate_1.
23395 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
23396 (gen_const_vec_duplicate): Use rtx_vector_builder.
23397 (gen_const_vec_series): Likewise.
23398 (gen_rtx_CONST_VECTOR): Likewise.
23399 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
23400 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
23401 Build a new vector rather than modifying a CONST_VECTOR in-place.
23402 (handle_special_swappables): Update call accordingly.
23403 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
23404 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
23405 Build a new vector rather than modifying a CONST_VECTOR in-place.
23406 (handle_special_swappables): Update call accordingly.
23408 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23410 * simplify-rtx.c (simplify_const_binary_operation): Use
23411 CONST_VECTOR_ELT instead of XVECEXP.
23413 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23415 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
23416 the selector elements to be different from the data elements
23417 if the selector is a VECTOR_CST.
23418 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
23419 ssizetype for the selector.
23421 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23423 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
23424 before testing each element individually.
23425 * tree-vect-generic.c (lower_vec_perm): Likewise.
23427 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23429 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
23430 * selftest-run-tests.c (selftest::run_tests): Call it.
23431 * vector-builder.h (vector_builder::operator ==): New function.
23432 (vector_builder::operator !=): Likewise.
23433 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
23434 (vec_perm_indices::all_from_input_p): New function.
23435 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
23436 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
23437 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
23438 instead of reading the VECTOR_CST directly. Detect whether both
23439 vector inputs are the same before constructing the vec_perm_indices,
23440 and update the number of inputs argument accordingly. Use the
23441 utility functions added above. Only construct sel2 if we need to.
23443 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23445 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
23446 the broadcast of the low byte.
23447 (expand_mult_highpart): Use an explicit encoding for the permutes.
23448 * optabs-query.c (can_mult_highpart_p): Likewise.
23449 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23450 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
23451 (vectorizable_bswap): Likewise.
23452 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
23453 explicit encoding for the power-of-2 permutes.
23454 (vect_permute_store_chain): Likewise.
23455 (vect_grouped_load_supported): Likewise.
23456 (vect_permute_load_chain): Likewise.
23458 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23460 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
23461 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
23462 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
23463 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
23464 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
23465 (vect_gen_perm_mask_any): Likewise.
23467 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23469 * int-vector-builder.h: New file.
23470 * vec-perm-indices.h: Include int-vector-builder.h.
23471 (vec_perm_indices): Redefine as an int_vector_builder.
23472 (auto_vec_perm_indices): Delete.
23473 (vec_perm_builder): Redefine as a stand-alone class.
23474 (vec_perm_indices::vec_perm_indices): New function.
23475 (vec_perm_indices::clamp): Likewise.
23476 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
23477 (vec_perm_indices::new_vector): New function.
23478 (vec_perm_indices::new_expanded_vector): Update for new
23479 vec_perm_indices class.
23480 (vec_perm_indices::rotate_inputs): New function.
23481 (vec_perm_indices::all_in_range_p): Operate directly on the
23482 encoded form, without computing elided elements.
23483 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
23484 encoding. Update for new vec_perm_indices class.
23485 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
23486 the given vec_perm_builder.
23487 (expand_vec_perm_var): Update vec_perm_builder constructor.
23488 (expand_mult_highpart): Use vec_perm_builder instead of
23489 auto_vec_perm_indices.
23490 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
23491 vec_perm_indices instead of auto_vec_perm_indices. Use a single
23492 or double series encoding as appropriate.
23493 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
23494 vec_perm_indices instead of auto_vec_perm_indices.
23495 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
23496 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
23497 (vect_permute_store_chain): Likewise.
23498 (vect_grouped_load_supported): Likewise.
23499 (vect_permute_load_chain): Likewise.
23500 (vect_shift_permute_load_chain): Likewise.
23501 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
23502 (vect_transform_slp_perm_load): Likewise.
23503 (vect_schedule_slp_instance): Likewise.
23504 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
23505 (vectorizable_mask_load_store): Likewise.
23506 (vectorizable_bswap): Likewise.
23507 (vectorizable_store): Likewise.
23508 (vectorizable_load): Likewise.
23509 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
23510 vec_perm_indices instead of auto_vec_perm_indices. Use
23511 tree_to_vec_perm_builder to read the vector from a tree.
23512 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
23513 vec_perm_builder instead of a vec_perm_indices.
23514 (have_whole_vector_shift): Use vec_perm_builder and
23515 vec_perm_indices instead of auto_vec_perm_indices. Leave the
23516 truncation to calc_vec_perm_mask_for_shift.
23517 (vect_create_epilog_for_reduction): Likewise.
23518 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
23519 from auto_vec_perm_indices to vec_perm_indices.
23520 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
23521 instead of changing individual elements.
23522 (aarch64_vectorize_vec_perm_const): Use new_vector to install
23523 the vector in d.perm.
23524 * config/arm/arm.c (expand_vec_perm_d::perm): Change
23525 from auto_vec_perm_indices to vec_perm_indices.
23526 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
23527 instead of changing individual elements.
23528 (arm_vectorize_vec_perm_const): Use new_vector to install
23529 the vector in d.perm.
23530 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
23531 Update vec_perm_builder constructor.
23532 (rs6000_expand_interleave): Likewise.
23533 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
23534 (rs6000_expand_interleave): Likewise.
23536 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23538 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
23539 to qimode could truncate the indices.
23540 * optabs.c (expand_vec_perm_var): Likewise.
23542 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23544 * Makefile.in (OBJS): Add vec-perm-indices.o.
23545 * vec-perm-indices.h: New file.
23546 * vec-perm-indices.c: Likewise.
23547 * target.h (vec_perm_indices): Replace with a forward class
23549 (auto_vec_perm_indices): Move to vec-perm-indices.h.
23550 * optabs.h: Include vec-perm-indices.h.
23551 (expand_vec_perm): Delete.
23552 (selector_fits_mode_p, expand_vec_perm_var): Declare.
23553 (expand_vec_perm_const): Declare.
23554 * target.def (vec_perm_const_ok): Replace with...
23555 (vec_perm_const): ...this new hook.
23556 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
23557 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
23558 * doc/tm.texi: Regenerate.
23559 * optabs.def (vec_perm_const): Delete.
23560 * doc/md.texi (vec_perm_const): Likewise.
23561 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
23562 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
23563 expand_vec_perm for constant permutation vectors. Assert that
23564 the mode of variable permutation vectors is the integer equivalent
23565 of the mode that is being permuted.
23566 * optabs-query.h (selector_fits_mode_p): Declare.
23567 * optabs-query.c: Include vec-perm-indices.h.
23568 (selector_fits_mode_p): New function.
23569 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
23570 is defined, instead of checking whether the vec_perm_const_optab
23571 exists. Use targetm.vectorize.vec_perm_const instead of
23572 targetm.vectorize.vec_perm_const_ok. Check whether the indices
23573 fit in the vector mode before using a variable permute.
23574 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
23575 vec_perm_indices instead of an rtx.
23576 (expand_vec_perm): Replace with...
23577 (expand_vec_perm_const): ...this new function. Take the selector
23578 as a vec_perm_indices rather than an rtx. Also take the mode of
23579 the selector. Update call to shift_amt_for_vec_perm_mask.
23580 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
23581 Use vec_perm_indices::new_expanded_vector to expand the original
23582 selector into bytes. Check whether the indices fit in the vector
23583 mode before using a variable permute.
23584 (expand_vec_perm_var): Make global.
23585 (expand_mult_highpart): Use expand_vec_perm_const.
23586 * fold-const.c: Includes vec-perm-indices.h.
23587 * tree-ssa-forwprop.c: Likewise.
23588 * tree-vect-data-refs.c: Likewise.
23589 * tree-vect-generic.c: Likewise.
23590 * tree-vect-loop.c: Likewise.
23591 * tree-vect-slp.c: Likewise.
23592 * tree-vect-stmts.c: Likewise.
23593 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
23595 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
23596 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
23597 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
23598 (aarch64_vectorize_vec_perm_const): ...this new function.
23599 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
23600 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
23601 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
23602 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
23603 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
23604 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
23605 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
23607 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
23608 check for NEON modes.
23609 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
23610 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
23611 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
23612 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
23614 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
23615 the old VEC_PERM_CONST conditions.
23616 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
23617 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
23618 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
23619 (ia64_vectorize_vec_perm_const_ok): Merge into...
23620 (ia64_vectorize_vec_perm_const): ...this new function.
23621 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
23622 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
23623 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
23624 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
23625 * config/mips/mips.c (mips_expand_vec_perm_const)
23626 (mips_vectorize_vec_perm_const_ok): Merge into...
23627 (mips_vectorize_vec_perm_const): ...this new function.
23628 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
23629 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
23630 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
23631 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
23632 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
23633 (rs6000_expand_vec_perm_const): Delete.
23634 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
23636 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
23637 (altivec_expand_vec_perm_const_le): Take each operand individually.
23638 Operate on constant selectors rather than rtxes.
23639 (altivec_expand_vec_perm_const): Likewise. Update call to
23640 altivec_expand_vec_perm_const_le.
23641 (rs6000_expand_vec_perm_const): Delete.
23642 (rs6000_vectorize_vec_perm_const_ok): Delete.
23643 (rs6000_vectorize_vec_perm_const): New function.
23644 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
23645 an element count and rtx array.
23646 (rs6000_expand_extract_even): Update call accordingly.
23647 (rs6000_expand_interleave): Likewise.
23648 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
23649 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
23650 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
23651 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
23652 (rs6000_expand_vec_perm_const): Delete.
23653 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
23654 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
23655 (altivec_expand_vec_perm_const_le): Take each operand individually.
23656 Operate on constant selectors rather than rtxes.
23657 (altivec_expand_vec_perm_const): Likewise. Update call to
23658 altivec_expand_vec_perm_const_le.
23659 (rs6000_expand_vec_perm_const): Delete.
23660 (rs6000_vectorize_vec_perm_const_ok): Delete.
23661 (rs6000_vectorize_vec_perm_const): New function. Remove stray
23662 reference to the SPE evmerge intructions.
23663 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
23664 an element count and rtx array.
23665 (rs6000_expand_extract_even): Update call accordingly.
23666 (rs6000_expand_interleave): Likewise.
23667 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
23668 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
23670 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
23672 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23674 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
23675 vector mode and that that mode matches the mode of the data
23677 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
23678 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
23679 directly using expand_vec_perm_1 when forcing selectors into
23681 (expand_vec_perm_var): New function, split out from expand_vec_perm.
23683 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23685 * optabs-query.h (can_vec_perm_p): Delete.
23686 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
23687 * optabs-query.c (can_vec_perm_p): Split into...
23688 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
23689 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
23690 particular selector is valid.
23691 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
23692 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
23693 (vect_grouped_load_supported): Likewise.
23694 (vect_shift_permute_load_chain): Likewise.
23695 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
23696 (vect_transform_slp_perm_load): Likewise.
23697 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
23698 (vectorizable_bswap): Likewise.
23699 (vect_gen_perm_mask_checked): Likewise.
23700 * fold-const.c (fold_ternary_loc): Likewise. Don't take
23701 implementations of variable permutation vectors into account
23702 when deciding which selector to use.
23703 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
23704 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
23705 with a false third argument.
23706 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
23707 to test whether the constant selector is valid and can_vec_perm_var_p
23708 to test whether a variable selector is valid.
23710 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23712 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
23713 * optabs-query.c (can_vec_perm_p): Likewise.
23714 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
23715 instead of vec_perm_indices.
23716 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
23717 (vect_gen_perm_mask_checked): Likewise,
23718 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
23719 (vect_gen_perm_mask_checked): Likewise,
23721 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
23723 * optabs-query.h (qimode_for_vec_perm): Declare.
23724 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
23725 (qimode_for_vec_perm): ...this new function.
23726 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
23728 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23730 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
23731 does not have a conditional at the top.
23733 2018-01-02 Richard Biener <rguenther@suse.de>
23735 * ipa-inline.c (big_speedup_p): Fix expression.
23737 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
23740 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
23743 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
23747 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
23748 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
23749 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
23750 cond_taken_branch_cost 3->4.
23752 2018-01-01 Jakub Jelinek <jakub@redhat.com>
23754 PR tree-optimization/83581
23755 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
23756 TODO_cleanup_cfg if any changes have been made.
23758 PR middle-end/83608
23759 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
23760 convert_modes if target mode has the right side, but different mode
23763 PR middle-end/83609
23764 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
23765 last argument when extracting from CONCAT. If either from_real or
23766 from_imag is NULL, use expansion through memory. If result is not
23767 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
23768 the parts directly to inner mode, if even that fails, use expansion
23771 PR middle-end/83623
23772 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
23773 check for bswap in mode rather than HImode and use that in expand_unop
23776 Copyright (C) 2018 Free Software Foundation, Inc.
23778 Copying and distribution of this file, with or without modification,
23779 are permitted in any medium without royalty provided the copyright
23780 notice and this notice are preserved.