1 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
8 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
10 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
11 frequencies. Fix estimation of aggregate parameters.
13 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
15 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
18 2017-11-23 Tom de Vries <tom@codesourcery.com>
20 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
22 2017-11-23 Tom de Vries <tom@codesourcery.com>
24 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
27 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
30 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
31 sibcall_value_pcrel_fdpic): Use local variable instead of
33 (calli_tbr_rel): Add missing operand 2.
34 (call_valuei_tbr_rel): Add missing operand 3.
36 2017-11-23 Jakub Jelinek <jakub@redhat.com>
39 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
40 bitpos/bitsize covering the whole destination, use store_expr only if
41 the complex mode is the same. Otherwise, use expand_normal and if
42 it returns CONCAT, subreg each part separately instead of trying to
43 subreg the whole result.
45 2017-11-23 Richard Biener <rguenther@suse.de>
47 PR tree-optimization/23094
48 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
49 come here from walking over backedges in the first iteration.
50 (vn_reference_lookup_3): Skip clobbers that store the same value.
52 2017-11-23 Richard Biener <rguenther@suse.de>
54 PR tree-optimization/81403
55 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
56 a block we need a leader relative to.
57 (phi_translate_1): For nary processing require a leader from
58 get_representative_for given we run expression simplification
59 using match-and-simplify. Remove previous fix.
61 2017-11-22 Jeff Law <law@redhat.com>
63 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
64 Use new method allocate_value_range rather than accessing the
65 vrp_value_range_pool data member directly.
66 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
67 to use extract_range_from_stmt method to avoid need for
68 extract_range_from_assignment method.
69 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
70 method rather than setting values_propgated data member directly.
71 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
72 and values propagated data members and extract_range_from_assignment
73 method. Reorder private data members to conform to standards.
74 Add new methods set_lattice_propagation_complete and
77 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
79 PR rtl-optimization/83030
80 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
81 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
82 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
83 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
84 CROSSING_JUMP_P flag is already set before setting it.
85 * cfgrtl.c (fixup_partition_crossing): Likewise.
86 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
89 2017-11-22 Jakub Jelinek <jakub@redhat.com>
91 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
92 Handle the case where both arguments are using gen_const_vec_series.
94 2017-11-22 David Malcolm <dmalcolm@redhat.com>
97 * pretty-print.c (pp_format): Move quoting implementation to
98 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
99 to pass address of "quote" local.
100 (pp_begin_quote): New function.
101 (pp_end_quote): New function.
102 * pretty-print.h (printer_fn): Convert penultimate param from bool
104 (pp_begin_quote): New decl.
105 (pp_end_quote): New decl.
106 * tree-diagnostic.c (default_tree_printer): Convert penultimate
107 param from bool to bool *.
108 * tree-diagnostic.h (default_tree_printer): Likewise.
110 2017-11-22 Jeff Law <law@redhat.com>
112 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
113 Initialize vr_values.
114 (evrp_range_analyzer::try_find_new_range): Call methods attached to
115 vr_values via vr_values class instance rather than delegators.
116 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
117 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
118 (evrp_range_analyzer::push_value_range): Likewise.
119 (evrp_range_analyzer::pop_value_range): Likewise.
120 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
121 most delegators. Those remaining are exposed as public interfaces.
122 Make vr_values a pointer and private.
123 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
125 (evrp_range_analyzer::get_vr_value): New method.
126 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
127 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
128 (class evrp_dom_walker): Attach evrp_folder class, initialize
129 it in the ctor. Remove temporary delegators.
130 (evrp_dom_walker::before_dom_children): Call methods in attached
131 evrp_range_analyzer class via class instance pointer. Use
132 free value_range_constant_singleton to remove need for
133 op_with_constant_singleton_value delegator method. Do not
134 create a vrp_prop class instance for every call! Narrow
135 scope of a couple variables.
136 (evrp_dom_walker::cleanup): Call methods in attached
137 evrp_range_analyzer class via class instance pointer.
138 * vr-values.h (class vr_values): Privatize many methods and
141 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
143 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
144 * cfgexpand.c (expand_debug_expr): Remove handling for them.
145 * expr.c (expand_expr_real_2): Likewise.
146 * fold-const.c (const_unop): Likewise.
147 * optabs-tree.c (optab_for_tree_code): Likewise.
148 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
149 * tree-inline.c (estimate_operator_cost): Likewise.
150 * tree-pretty-print.c (dump_generic_node): Likewise.
151 (op_code_prio): Likewise.
152 (op_symbol_code): Likewise.
153 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
154 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
155 * internal-fn.c (direct_internal_fn_optab): New function.
156 (direct_internal_fn_array, direct_internal_fn_supported_p
157 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
158 * fold-const-call.c (fold_const_reduction): New function.
159 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
161 * tree-vect-loop.c: Include internal-fn.h.
162 (reduction_code_for_scalar_code): Rename to...
163 (reduction_fn_for_scalar_code): ...this and return an internal
165 (vect_model_reduction_cost): Take an internal_fn rather than
167 (vect_create_epilog_for_reduction): Likewise. Build calls rather
169 (vectorizable_reduction): Use internal functions rather than tree
170 codes for the reduction operation. Update calls to the functions
172 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
173 Use calls to internal functions rather than REDUC tree codes.
174 * config/aarch64/aarch64-simd.md: Update comment accordingly.
176 2017-11-22 Olivier Hainque <hainque@adacore.com>
178 * config/vxworks.c (vxworks_override_options): Pick default
179 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
180 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
181 DWARF_GNAT_ENCODINGS_DEFAULT.
182 * config/vxworksae.h: Likewise.
184 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
186 PR tree-optimization/83104
187 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
190 2017-11-22 Marek Polacek <polacek@redhat.com>
191 H.J. Lu <hongjiu.lu@intel.com>
192 Jason Merrill <jason@redhat.com>
197 * calls.c (initialize_argument_information): Call
198 warn_parameter_passing_abi target hook.
199 (store_one_arg): Use 0 for empty record size. Don't push 0 size
201 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
202 * common.opt: Update -fabi-version description.
203 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
204 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
206 (ix86_is_empty_record): New function.
207 (ix86_warn_parameter_passing_abi): New function.
208 (TARGET_EMPTY_RECORD_P): Redefine.
209 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
210 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
211 * doc/tm.texi: Regenerated.
212 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
213 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
214 * dwarf2out.c (get_ultimate_context): Move to tree.c.
215 * explow.c (hard_function_value): Call arg_int_size_in_bytes
216 instead of int_size_in_bytes.
217 * expr.c (copy_blkmode_to_reg): Likewise.
218 * function.c (aggregate_value_p): Return 0 for empty types.
219 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
220 (locate_and_pad_parm): Call arg size_in_bytes instead
222 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
223 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
224 * target.def (empty_record_p, warn_parameter_passing_abi): New target
226 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
227 (std_gimplify_va_arg_expr): Skip empty records. Call
228 arg_size_in_bytes instead size_in_bytes.
229 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
230 * tree-core.h (tree_type_common): Add empty_flag.
231 (tree_decl_common): Update comments.
232 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
234 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
235 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
237 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
238 * tree.c (default_is_empty_type): New function.
239 (default_is_empty_record): New function.
240 (arg_int_size_in_bytes): New function.
241 (arg_size_in_bytes): New function.
242 (get_ultimate_context): New function.
243 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
244 TRANSLATION_UNIT_WARN_EMPTY_P.
245 (default_is_empty_record, arg_int_size_in_bytes,
246 arg_size_in_bytes, get_ultimate_context): Declare.
248 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
250 * config/arm/arm.c (cmse_clear_registers): New function.
251 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
252 code by call to cmse_clear_registers.
253 (cmse_nonsecure_entry_clear_before_return): Likewise.
255 2017-11-22 Tamar Christina <tamar.christina@arm.com>
257 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
258 (vdot_s32, vdotq_s32): New.
259 (vdot_lane_u32, vdotq_lane_u32): New.
260 (vdot_lane_s32, vdotq_lane_s32): New.
263 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
266 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
267 for unsigned values with fewer HWIs than the precision.
268 (test_overflow): New function.
269 (wide_int_cc_tests): Call it.
271 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
273 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
274 has the same mode class as Pmode.
276 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
278 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
279 padding_bits_to_clear_ptr.
280 (cmse_nonsecure_entry_clear_before_return): Likewise.
282 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
284 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
285 auto_sbitap instead of integer bitfield to control register needing
288 2017-11-22 Jakub Jelinek <jakub@redhat.com>
290 PR tree-optimization/83044
291 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
292 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
293 that need the upper bound. Subtract offset from
294 get_addr_base_and_unit_offset only if positive and subtract it
295 before division by eltsize rather than after it.
298 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
299 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
300 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
304 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
305 expand_binop with *mul_widen_optab, make sure at least one of the
306 operands doesn't have VOIDmode.
309 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
311 PR rtl-optimization/82044
312 PR tree-optimization/82042
313 * dse.c (record_store): Check for overflow.
314 (check_mem_read_rtx): Properly check for overflow if width == -1, call
315 add_wild_read instead of clear_rhs_from_active_local_stores on
316 overflow and log it into dump_file.
318 2017-11-22 Richard Biener <rguenther@suse.de>
320 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
321 fake edges to exit when looking for a place to insert.
322 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
324 (insert_into_preds_of_block): Commit edge insertion immediately,
325 assert that doesn't require new BBs.
326 (fini_pre): Release expressions.
327 (pass_pre::execute): Shuffle things around a bit, if the fn
328 is too large do not compute AVAIL either as this is really the
331 2017-11-22 Richard Biener <rguenther@suse.de>
333 PR tree-optimization/83089
334 * tree-if-conv.c (pass_if_conversion::execute): If anything
335 changed reset SCEV and free the number of iteration estimates.
337 2017-11-21 Martin Sebor <msebor@redhat.com>
339 PR tree-optimization/82945
340 * calls.h (warn_nonstring_bound): Remove unused function.
342 2017-11-21 Martin Sebor <msebor@redhat.com>
344 PR tree-optimization/82945
345 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
346 * calls.h (maybe_warn_nonstring_arg): Declare new function.
347 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
349 (initialize_argument_information): Call maybe_warn_nonstring_arg.
350 * calls.h (get_attr_nonstring_decl): Declare new function.
351 * doc/extend.texi (attribute nonstring): Update.
352 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
353 get_attr_nonstring_decl and handle it.
354 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
355 detection of nul-termination.
356 (strlen_to_stridx): Change to a pointer.
357 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
358 (pass_strlen::execute): Same.
360 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
362 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
363 for the new option -mprefer-vector-width=[none|128|256|512].
364 * config/i386/i386.c (ix86_target_string): remove old style options
365 -mprefer-avx256 and make -mprefer-avx128 as alias.
366 (ix86_option_override_internal): Apply defaults for the
367 -mprefer-vector-width=[128|256] option.
368 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
369 Implement macros to work with -mprefer-vector-width=.
370 * config/i386/i386.opt: Implemented option
371 -mprefer-vector-width=[none|128|256|512].
372 * doc/invoke.texi: Documentation for
373 -mprefer-vector-width=[none|128|256|512].
375 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
377 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
379 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
381 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
383 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
385 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
387 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
388 * expr.c (expand_expr_real_2): Likewise.
389 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
390 fold_binary_loc): Likewise.
391 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
392 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
393 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
394 MINUS_EXPR transformations.
395 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
396 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
397 * tree-inline.c (estimate_operator_cost): Likewise.
398 * tree-pretty-print.c (dump_generic_node, op_code_prio,
399 op_symbol_code): Likewise.
400 * tree-vect-stmts.c (vectorizable_operation): Likewise.
401 * vr-values.c (extract_range_from_binary_expr): Likewise.
402 * varasm.c (initializer_constant_valid_p_1): Likewise.
403 * tree.def: New tree code POINTER_DIFF_EXPR.
405 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
407 * config/i386/i386.md (*bswap<mode>2_movbe): Add
408 integer suffix to movbe mnemonic.
409 (*bswaphi2_movbe): Ditto.
410 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
412 2017-11-21 Jakub Jelinek <jakub@redhat.com>
415 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
416 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
417 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
419 2017-11-21 Martin Liska <mliska@suse.cz>
421 * tree-inline.c (expand_call_inline): Remove not needed
424 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
425 Jakub Jelinek <jakub@redhat.com>
428 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
429 Remove static keyword from f variable.
431 2017-11-21 Jakub Jelinek <jakub@redhat.com>
433 PR tree-optimization/83086
434 * gimple-ssa-store-merging.c
435 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
436 rather than n.base_addr.
438 2017-11-21 Martin Liska <mliska@suse.cz>
440 PR rtl-optimization/82044
441 PR tree-optimization/82042
442 * dse.c (check_mem_read_rtx): Check for overflow.
444 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
446 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
449 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
451 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
452 and its corresponding call arg location note.
454 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
455 Andrew Burgess <andrew.burgess@embecosm.com>
457 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
459 (arc_return_slot_offset): Likewise.
460 (arc_eh_return_address_location): New declaration.
461 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
462 (MUST_SAVE_REGISTER): Add exception handler case.
463 (MUST_SAVE_RETURN_ADDR): Likewise.
464 (arc_frame_pointer_required): Likewise.
465 (arc_frame_pointer_needed): New function.
466 (arc_compute_frame_size): Changed.
467 (arc_expand_prologue): Likewise.
468 (arc_expand_epilogue): Likewise.
469 (arc_initial_elimination_offset): Likewise.
470 (arc_return_slot_offset): Delete.
471 (arc_eh_return_address_location): New function.
472 (arc_builtin_setjmp_frame_value): Likewise.
473 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
474 (EH_RETURN_STACKADJ_RTX): Define.
475 (EH_RETURN_HANDLER_RTX): Likewise.
476 * config/arc/arc.md (eh_return): Delete.
478 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
480 * print-tree.h (debug_vec_tree): Remove prototype.
481 * gdbinit.in (pvt): Remove macro.
483 2017-11-21 Jakub Jelinek <jakub@redhat.com>
485 PR tree-optimization/83047
486 * gimple-ssa-store-merging.c
487 (imm_store_chain_info::output_merged_store): If the loads with the
488 same vuse are in different basic blocks, for load_gsi pick a load
489 location that is dominated by the other loads.
492 * config/i386/i386.c (ix86_memmodel_check): Start
493 -Winvalid-memory-model diagnostics with lowercase letter.
496 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
497 set in_cold_section_p to the partition containing loc_list->first.
498 When seeing loc_list->last_before_switch node, update secname and
499 perform range_across_switch second partition handling only after that.
502 * run-rtl-passes.c: Include debug.h.
503 (run_rtl_passes): Call debug_hooks->assembly_start.
504 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
508 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
509 OPTAB_DIRECT in calls to expand_simple_binop.
511 2017-11-20 David Malcolm <dmalcolm@redhat.com>
514 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
516 2017-11-20 David Malcolm <dmalcolm@redhat.com>
519 * spellcheck.h (best_match::blithely_get_best_candidate): New
522 2017-11-20 Jakub Jelinek <jakub@redhat.com>
524 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
525 with lower case letter.
527 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
529 * config/i386/i386.md (bswaphi2): New expander.
530 (*bswaphi2_movbe): New insn pattern.
531 (bswaphi -> rorhi pepehole2): New peephole pattern.
533 2017-11-20 Jeff Law <law@redhat.com>
535 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
536 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
537 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
538 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
539 gimple-ssa-evrp-analyze.h.
541 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
543 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
545 2017-11-20 Bin Cheng <bin.cheng@arm.com>
547 * tree-predcom.c: Add general comment on Store-Store chains.
548 (split_data_refs_to_components): Postpone clearing eliminate_store_p
550 (get_chain_last_ref_at): Rename into...
551 (get_chain_last_write_at): ...this.
552 (get_chain_last_write_before_load): New function.
553 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
554 CT_STORE_STORE when write reference is added.
555 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
556 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
557 (initialize_root_vars_store_elim_1): Ditto.
558 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
559 definition is created.
560 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
561 chain by replacing it with dominant stored value.
563 2017-11-20 Bin Cheng <bin.cheng@arm.com>
565 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
567 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
569 * vr-values.c (extract_range_from_binary_expr): Use a full range
572 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
574 * config/arm/arm.md (R4_REGNUM): Define constant.
575 (nonsecure_call_internal): Remove r4 clobber.
576 (nonsecure_call_value_internal): Likewise.
577 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
578 clobber and resequence match_operands.
579 (nonsecure_call_value_reg_thumb1_v5): Likewise.
580 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
581 (nonsecure_call_value_reg_thumb2): Likewise.
583 2017-11-20 Jakub Jelinek <jakub@redhat.com>
585 PR tree-optimization/78821
586 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
587 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
588 to the address of the base rather than the base itself.
589 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
590 (find_bswap_or_nop_finalize): New function.
591 (find_bswap_or_nop): Use it.
592 (bswap_replace): Return a tree rather than bool, change first
593 argument from gimple * to gimple_stmt_iterator, allow inserting
594 into an empty sequence, allow ins_stmt to be NULL - then emit
595 all stmts into gsi. Fix up MEM_REF address gimplification.
596 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
597 (struct store_immediate_info): Add N and INS_STMT non-static
599 (store_immediate_info::store_immediate_info): Initialize them
600 from newly added ctor args.
601 (merged_store_group::apply_stores): Formatting fixes. Sort by
603 (stmts_may_clobber_ref_p): For stores call also
604 refs_anti_dependent_p.
605 (gather_bswap_load_refs): New function.
606 (imm_store_chain_info::try_coalesce_bswap): New method.
607 (imm_store_chain_info::coalesce_immediate_stores): Use it.
608 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
609 (imm_store_chain_info::output_merged_store): Fail if number of
610 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
611 and NOP_EXPR rhs_code.
612 (pass_store_merging::process_store): Compute n and ins_stmt, if
613 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
614 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
616 (pass_store_merging::execute): Calculate dominators.
618 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
619 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
620 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
621 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
622 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
623 class pass_optimize_bswap, bswap_replace,
624 pass_optimize_bswap::execute): Moved to ...
625 * gimple-ssa-store-merging.c: ... this file.
626 Include optabs-tree.h.
627 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
628 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
629 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
630 anonymous namespace, remove static keywords.
631 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
632 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
634 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
637 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
639 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
641 * vec.h (debug_helper): New function.
642 (DEFINE_DEBUG_VEC): New macro.
643 * hash-set.h (debug_helper): New function.
644 (DEFINE_DEBUG_HASH_SET): New macro.
645 * cfg.c (debug_slim (edge)): New function.
646 Call DEFINE_DEBUG_VEC for edges.
647 Call DEFINE_DEBUG_HASH_SET for edges.
648 * cfghooks.c (debug_slim (basic_block)): New function.
649 Call DEFINE_DEBUG_VEC for basic blocks.
650 Call DEFINE_DEBUG_HASH_SET for basic blocks.
651 * print-tree.c (debug_slim): New function to handle trees.
652 Call DEFINE_DEBUG_VEC for trees.
653 Call DEFINE_DEBUG_HASH_SET for trees.
654 (debug (vec<tree, va_gc>) &): Remove.
655 (debug (<vec<tree, va_gc>) *): Remove.
656 * print-rtl.c (debug_slim): New function to handle const_rtx.
657 Call DEFINE_DEBUG_VEC for rtx_def.
658 Call DEFINE_DEBUG_VEC for rtx_insn.
659 Call DEFINE_DEBUG_HASH_SET for rtx_def.
660 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
661 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
662 (debug (vec<rtx_insn *> *ptr): Remove.
663 (debug_insn_vector): Remove.
664 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
666 2017-11-20 Tom de Vries <tom@codesourcery.com>
668 PR rtl-optimization/82020
669 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
670 IF_THEN_ELSE condition.
672 2017-11-19 Jeff Law <law@redhat.com>
674 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
675 of degenerates resulting from ignoring an edge.
677 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
680 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
682 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
685 * profile-count.c (profile_count::to_sreal_scale): Fix return value
686 for uninitialied counts.
688 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
691 * tree-inline.c (estimate_num_insns): Set to 1 at least.
693 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
696 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
698 2017-11-19 Tom de Vries <tom@codesourcery.com>
700 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
703 2017-11-19 Tom de Vries <tom@codesourcery.com>
705 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
708 2017-11-19 Tom de Vries <tom@codesourcery.com>
710 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
713 2017-11-19 Tom de Vries <tom@codesourcery.com>
715 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
717 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
719 2017-11-19 Tom de Vries <tom@codesourcery.com>
721 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
723 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
724 semicolon after MCORE_EXPORT_NAME call.
726 2017-11-19 Tom de Vries <tom@codesourcery.com>
729 * vmsdbgout.c (vmsdbgout_early_finish): New function.
730 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
732 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
734 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
735 (cgraph_edge::clone): Cleanup updating of profile.
736 * ipa-cp.c (update_profiling_info): Likewise.
737 * ipa-inline-transform.c (inline_transform): Likewise.
738 * ipa-inline.c (inline_small_functions): Add missing space to dump.
739 * ipa-split.c (execute_split_functions): Do not split when function
741 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
742 * profile-count.c (profile_count::dump): Add global0.
743 (profile_count::to_cgraph_frequency): Do not ICE when entry is
745 (profile_count::to_sreal_scale): Likewise.
746 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
747 (profile_count::combine_with_ipa_count): New function.
748 * profile-count.h (profile_guessed_global0adjusted): New.
749 (profile_count::adjusted_zero): New.
750 (profile_count::global0adjusted): New.
751 (profile_count::combine_with_ipa_count): New.
752 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
753 correct profile of return block of split functions.
754 (copy_cfg_body): Remove unused profile_count.
755 (copy_body): Likewise.
756 (expand_call_inline): Update.
757 (tree_function_versioning): Update.
759 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
761 * hash-set.h (hash_set::empty): New.
762 * tree-ssa-threadbackward.h: Delete.
763 * tree-ssa-threadbackward.c (class thread_jumps): New.
764 Move max_threaded_paths into class.
765 (fsm_find_thread_path): Remove arguments that are now in class.
766 (profitable_jump_thread_path): Rename to...
767 (thread_jumps::profitable_jump_thread_path): ...this.
768 (convert_and_register_jump_thread_path): Rename to...
769 (thread_jumps::convert_and_register_current_path): ...this.
770 (check_subpath_and_update_thread_path): Rename to...
771 (thread_jumps::check_subpath_and_update_thread_path): ...this.
772 (register_jump_thread_path_if_profitable): Rename to...
773 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
774 (handle_phi): Rename to...
775 (thread_jumps::handle_phi): ...this.
776 (handle_assignment): Rename to...
777 (thread_jumps::handle_assignment): ...this.
778 (fsm_find_control_statement_thread_paths): Rename to...
779 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
780 (find_jump_threads_backwards): Rename to...
781 (thread_jumps::find_jump_threads_backwards): ...this.
782 Initialize path local data.
783 (pass_thread_jumps::execute): Call find_jump_threads_backwards
784 from within thread_jumps class.
785 (pass_early_thread_jumps::execute): Same.
787 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
789 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
791 2017-11-17 Steve Ellcey <sellcey@cavium.com>
793 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
794 to canonical location.
796 2017-11-17 Steve Ellcey <sellcey@cavium.com>
799 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
801 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
803 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
805 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
806 rcrt1.o%s/grcrt1.o%s for -static-pie.
808 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
810 * i386.c (ix86_multiplication_cost, ix86_division_cost,
811 ix86_shift_rotate_cost): Break out from ...
812 (ix86_rtx_costs): ... here.
813 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
816 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
818 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
819 when entry block was promoted unlikely.
820 (estimate_bb_frequencies): Increase frequency scale.
821 * profile-count.h (profile_count): Export precision info.
823 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
825 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
826 disturbin profile of entry block.
828 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
830 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
833 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
835 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
836 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
838 2017-11-17 Jeff Law <law@redhat.com>
840 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
841 from evrp_dom_walker class. Various methods moved into new class.
842 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
843 (evrp_range_analyzer::enter): New method.
844 (evrp_range_analyzer::leave): New method.
845 (evrp_dom_walker): Remove delegators no longer needed by this class.
846 Replace vr_values data member with evrp_range_analyzer
848 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
849 method extracted from evrp_dom_walker::before_dom_children.
850 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
851 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
853 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
854 Add private copy constructor and move assignment operators.
855 Privatize methods and class data where trivially possible.
856 (evrp_dom_walker::cleanup): New function, extracted from
857 execute_early_vrp. Simplify access to class data.
859 * vr-values.h (get_output_for_vrp): Prototype.
860 * vr-values.c (get_output_for_vrp): New function extracted from
861 vrp_visit_assignment_or_call and extract_range_from_stmt.
862 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
864 2017-11-17 Luis Machado <luis.machado@linaro.org>
866 * config/aarch64/aarch64.c
867 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
868 (qdf24xx_tunings) <autoprefetcher_model>: Set to
869 tune_params::AUTOPREFETCHER_WEAK.
871 2017-11-17 Tamar Christina <tamar.christina@arm.com>
874 * config/arm/arm.c (arm_valid_target_attribute_rec):
875 Parse "arch=" and "+<ext>".
876 (arm_valid_target_attribute_tree): Re-init global options.
877 (arm_option_override): Make non-static.
878 (arm_options_perform_arch_sanity_checks): Make errors fatal.
879 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
880 (__ARM_FEATURE_CRC32): Support undef.
881 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
882 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
884 2017-11-17 David Malcolm <dmalcolm@redhat.com>
886 * gdbinit.in (break-on-diagnostic): New command.
888 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
890 * config/i386/i386.c (ix86_expand_epilogue): Change simple
891 return to indirect jump for EH return if control-flow
892 protection is enabled. Change explicit 'false' argument in
893 pro_epilogue_adjust_stack with a value of
895 * config/i386/i386.md (simple_return_indirect_internal):
896 Remove SImode restriction to support 64-bit.
898 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
900 * combine.c (added_notes_insn): New.
901 (try_combine): Handle added_notes_insn like added_links_insn.
902 Rewrite return value code.
903 (distribute_notes): Set added_notes_insn to the earliest insn we added
906 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
908 PR rtl-optimization/82621
909 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
910 dest of one of those SETs is unused.
912 2017-11-17 Richard Biener <rguenther@suse.de>
915 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
916 * tree-pretty-print.c (dump_generic_node): Handle
917 annot_expr_parallel_kind.
918 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
919 * gimplify.c (gimple_boolify): Likewise.
921 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
923 * config.gcc (extra_headers): Add cet.h for x86 targets.
924 * config/i386/cet.h: New file.
925 * doc/install.texi: Add --enable-cet/--disable-cet.
927 2017-11-17 Richard Biener <rguenther@suse.de>
929 PR tree-optimization/83017
930 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
931 (gen_parallel_loop): Properly count iterations.
932 (parallelize_loops): Handle loop->can_be_parallel independent
933 of flag_loop_parallelize_all. Make static profitability test match
935 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
936 * invoke.texi (parloops-min-per-thread): Document.
938 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
940 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
941 upstreaming review comments.
943 2017-11-17 Tamar Christina <tamar.christina@arm.com>
945 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
946 with fast unaligned access.
947 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
949 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
951 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
952 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
954 2017-11-17 Richard Biener <rguenther@suse.de>
956 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
957 folding of references.
959 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
962 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
963 of replacing call to strncmp with corresponding call to strcmp when
966 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
968 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
969 option prefer-avx256 for skylake-avx512 configuration.
970 * config/i386/i386.c (ix86_option_override_internal): Ditto.
971 (get_builtin_code_for_version): Ditto.
973 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
974 Monk Chiang <sh.chiang04@gmail.com>
976 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
977 (FIXED_REGISTERS): Reserve more register numbers.
978 (CALL_USED_REGISTERS): Likewise.
979 (REG_ALLOC_ORDER): Likewise.
980 (REG_CLASS_CONTENTS): Likewise.
981 (REGISTER_NAMES): Likewise.
983 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
984 Kito Cheng <kito.cheng@gmail.com>
986 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
988 * config/nds32/iterators.md: Add vector mode iterators and attributes.
990 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
992 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
993 parameter list for vec_splats.
995 2017-11-16 Joseph Myers <joseph@codesourcery.com>
997 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
999 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1001 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1004 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1005 with the required version of automake 1.11.6.
1007 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1009 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1011 (DU_C2_3_power9): Correct reservation combinations.
1012 (FP_DIV_power9, VEC_DIV_power9): New.
1013 (power9-alu): Split out rotate/shift...
1014 (power9-rot): ...to here, correct dispatch resource.
1015 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1017 (power9-fp): Correct latency.
1018 (power9-sdiv): Add div/sqrt resource.
1019 (power9-ddiv): Correct latency, add div/sqrt resource.
1020 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1021 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1023 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1025 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1027 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1028 switch statement mapping KF built-ins to TF built-ins if we don't
1029 have the proper ISA 3.0 assembler support.
1031 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1033 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1034 (lower_emutls_function_body): Do not compute it.
1036 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1038 * ipa-split.c (split_bb_info): Turn time to sreal.
1039 (split_point): Likewise.
1040 (dump_split_point): Likewise.
1041 (fine_split_points): Likewise.
1042 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1044 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1046 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1049 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1051 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1053 (expensive_function_p): Remove useless assert.
1054 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1056 2017-11-16 Martin Sebor <msebor@redhat.com>
1058 PR tree-optimization/82588
1059 PR tree-optimization/82583
1060 * tree-vrp.c (check_array_ref): Handle flexible array members,
1061 string literals, and inner indices.
1062 (search_for_addr_array): Add detail to diagnostics.
1064 2017-11-16 Nathan Sidwell <nathan@acm.org>
1068 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1069 (SET_DECL_ASSEMBLER_NAME): Forward to
1070 overwrite_decl_assembler_name.
1071 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1072 (overwrite_decl_assembler_name): Declare.
1073 * tree.c (overwrite_decl_assembler_name): New.
1074 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1075 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1076 (LANG_HOOKS_INITIALIZER): Add it.
1077 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1078 * langhooks.c (lhd_set_decl_assembler_name): Use
1079 SET_DECL_ASSEMBLER_NAME.
1080 (lhd_overwrite_decl_assembler_name): Default implementation.
1082 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1083 Jackson Woodruff <jackson.woodruff@arm.com>
1085 PR tree-optimization/71026
1086 * match.pd: Canonicalize constant multiplies in division.
1088 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1090 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1091 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1092 Remove OPT_fomit_frame_pointer entry.
1093 * common/config/arc/arc-common.c: Likewise.
1094 * common/config/arm/arm-common.c: Likewise.
1095 * common/config/avr/avr-common.c: Likewise.
1096 * common/config/c6x/c6x-common.c: Likewise.
1097 * common/config/cr16/cr16-common.c: Likewise.
1098 * common/config/cris/cris-common.c: Likewise.
1099 * common/config/epiphany/epiphany-common.c: Likewise.
1100 * common/config/fr30/fr30-common.c: Likewise.
1101 * common/config/frv/frv-common.c: Likewise.
1102 * common/config/ia64/ia64-common.c: Likewise.
1103 * common/config/iq2000/iq2000-common.c: Likewise.
1104 * common/config/lm32/lm32-common.c: Likewise.
1105 * common/config/m32r/m32r-common.c: Likewise.
1106 * common/config/mcore/mcore-common.c: Likewise.
1107 * common/config/microblaze/microblaze-common.c: Likewise.
1108 * common/config/mips/mips-common.c: Likewise.
1109 * common/config/mmix/mmix-common.c: Likewise.
1110 * common/config/mn10300/mn10300-common.c: Likewise.
1111 * common/config/nios2/nios2-common.c: Likewise.
1112 * common/config/pa/pa-common.c: Likewise.
1113 * common/config/pdp11/pdp11-common.c: Likewise.
1114 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1115 * common/config/riscv/riscv-common.c: Likewise.
1116 * common/config/rs6000/rs6000-common.c: Likewise.
1117 * common/config/rx/rx-common.c: Likewise.
1118 * common/config/s390/s390-common.c: Likewise.
1119 * common/config/sh/sh-common.c: Likewise.
1120 * common/config/sparc/sparc-common.c: Likewise.
1121 * common/config/tilegx/tilegx-common.c: Likewise.
1122 * common/config/tilepro/tilepro-common.c: Likewise.
1123 * common/config/v850/v850-common.c: Likewise.
1124 * common/config/visium/visium-common.c: Likewise.
1125 * common/config/xstormy16/xstormy16-common.c: Likewise.
1126 * common/config/xtensa/xtensa-common.c: Likewise.
1127 * invoke.texi (-fomit-frame-pointer): Update documentation.
1129 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1131 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1133 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1135 * cfg.c (scale_bbs_frequencies_int,
1136 cale_bbs_frequencies_gcov_type): Remove.
1137 * cfg.h (scale_bbs_frequencies_int,
1138 cale_bbs_frequencies_gcov_type): Remove.
1140 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1142 * tree-ssa-loop-manip.c
1143 (scale_dominated_blocks_in_loop): Update to profile counts.
1144 (tree_transform_and_unroll_loop): Likewise.
1146 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1148 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1149 scale_bbs_frequencies_int.
1151 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1153 * final.c (compute_alignments): Use counts rather than frequencies.
1155 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1157 * cfgloopanal.c: Include sreal.h
1158 (average_num_loop_insns): Use counts and sreal for accounting.
1160 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1162 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1165 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1166 Kito Cheng <kito.cheng@gmail.com>
1168 * config/nds32/constraints.md: Provide more constraints.
1169 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1170 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1171 support constraints usage.
1173 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1175 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1177 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1178 Kito Cheng <kito.cheng@gmail.com>
1180 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1181 * config/nds32/nds32.opt: Refine the layout.
1182 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1183 TARGET_EXT_STRING): Support new options.
1184 * config/nds32/nds32.h: Likewise.
1185 * config/nds32/nds32.md: Likewise.
1186 * config/nds32/nds32-predicates.c: Likewise.
1187 * config/nds32/constraints.md: Likewise.
1188 * common/config/nds32/nds32-common.c: Likewise.
1190 2017-11-16 Julia Koval <julia.koval@intel.com>
1193 * config/i386/gfniintrin.h: Add sse check.
1194 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1196 2017-11-16 Julia Koval <julia.koval@intel.com>
1198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1199 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1200 (ix86_handle_option): Handle -mavx512vbmi2.
1201 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1202 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1203 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1204 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1205 (ix86_valid_target_attribute_inner_p): Ditto.
1206 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1207 * config/i386/i386.opt (mavx512vbmi2): New option.
1208 * doc/invoke.texi: Add new option.
1210 [2017-11-16 Julia Koval <julia.koval@intel.com>
1212 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1213 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1214 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1215 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1216 _mm512_gf2p8mul_epi8): New intrinsics.
1217 * config/i386/i386-builtin-types.def
1218 (V64QI_FTYPE_V64QI_V64QI): New type.
1219 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1220 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1221 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1222 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1223 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1224 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1226 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1228 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1229 explicitly as a stream of bytes.
1231 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1233 * config/rs6000/altivec.h (vec_xst_be): New #define.
1234 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1235 and externalize from *altivec_vperm_<mode>_internal.
1236 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1238 (XL_BE_V8HI): Likewise.
1239 (XL_BE_V4SI): Likewise.
1240 (XL_BE_V4SI): Likewise.
1241 (XL_BE_V2DI): Likewise.
1242 (XL_BE_V4SF): Likewise.
1243 (XL_BE_V2DF): Likewise.
1244 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1245 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1246 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1247 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1248 VSX_BUILTIN_VEC_XST_BE.
1249 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1250 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1252 (altivec_init_builtins): Replace conditional calls to def_builtin
1253 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1254 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1255 with unconditional calls. Remove calls to def_builtin for
1256 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1257 __builtin_vec_xst_be.
1258 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1259 to define_expand, and add alternate RTL generation for P8.
1260 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1261 vsx_ld_elemrev_v8hi.
1262 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1263 add alternate RTL generation for P8.
1264 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1265 vsx_ld_elemrev_v16qi.
1266 (vsx_st_elemrev_v8hi): Convert define_insn
1267 to define_expand, and add alternate RTL generation for P8.
1268 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1269 vsx_st_elemrev_v8hi.
1270 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1271 add alternate RTL generation for P8.
1272 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1273 vsx_st_elemrev_v16qi.
1275 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1278 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1279 TARGET_AVX512ER check.
1280 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1281 neither -mzeroupper nor -mno-zeroupper is used and
1282 TARGET_EMIT_VZEROUPPER is set.
1283 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1284 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1286 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1288 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1289 folding of vector compares.
1290 (fold_build_vec_cmp): New helper function.
1291 (fold_compare_helper): New helper function.
1292 (builtin_function_type): Add compare builtins to the list of functions
1293 having unsigned arguments. Cosmetic updates to comment indentation.
1294 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1295 the not+eq combination.
1297 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1299 PR tree-optimization/82726
1300 PR tree-optimization/70754
1301 * tree-predcom.c (order_drefs_by_pos): New function.
1302 (combine_chains): Move code setting has_max_use_after to...
1303 (try_combine_chains): ...here. New parameter. Sort combined chains
1304 according to position information.
1305 (tree_predictive_commoning_loop): Update call to above function.
1306 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1308 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1310 PR tree-optimization/82726
1312 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1314 PR tree-optimization/70754
1315 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1316 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1317 combined stmt before it if not NULL.
1318 (combine_chains): Process refs reversely and compute dominance point
1322 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1324 PR tree-optimization/79663
1325 * tree-predcom.c (combine_chains): Process refs in reverse order
1326 only for ZERO length chains, and add explaining comment.
1328 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1330 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1331 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1332 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1334 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1336 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1338 2017-11-15 Martin Liska <mliska@suse.cz>
1340 * tree-cfg.c (pass_warn_function_return::execute):
1341 Compare warn_return_type for greater than zero.
1343 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1347 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1348 to return true on Xeon and not on Xeon Phi.
1349 (ix86_check_avx256_register): Changed to ...
1350 (ix86_check_avx_upper_register): ... this. Add extra check for
1351 VALID_AVX512F_REG_OR_XI_MODE.
1352 (ix86_avx_u128_mode_needed): Changed
1353 ix86_check_avx256_register to ix86_check_avx_upper_register.
1354 (ix86_check_avx256_stores): Changed to ...
1355 (ix86_check_avx_upper_stores): ... this. Changed
1356 ix86_check_avx256_register to ix86_check_avx_upper_register.
1357 (ix86_avx_u128_mode_after): Changed
1358 avx_reg256_found to avx_upper_reg_found. Changed
1359 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1360 (ix86_avx_u128_mode_entry): Changed
1361 ix86_check_avx256_register to ix86_check_avx_upper_register.
1362 (ix86_avx_u128_mode_exit): Ditto.
1363 * config/i386/i386.h: (host_detect_local_cpu): New define.
1365 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1367 * config/arm/xgene1.md (xgene1): Split into automatons
1368 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1369 (xgene1_f_load): Adjust reservations and/or types.
1370 (xgene1_f_store): Likewise.
1371 (xgene1_load_pair): Likewise.
1372 (xgene1_store_pair): Likewise.
1373 (xgene1_fp_load1): Likewise.
1374 (xgene1_load1): Likewise.
1375 (xgene1_store1): Likewise.
1376 (xgene1_move): Likewise.
1377 (xgene1_alu): Likewise.
1378 (xgene1_simd): Likewise.
1379 (xgene1_bfm): Likewise.
1380 (xgene1_neon_load1): Likewise.
1381 (xgene1_neon_store1): Likewise.
1382 (xgene1_neon_logic): Likewise.
1383 (xgene1_neon_st1): Likewise.
1384 (xgene1_neon_ld1r): Likewise.
1385 (xgene1_alu_cond): Added.
1386 (xgene1_shift_reg): Likwise.
1387 (xgene1_bfx): Likewise.
1388 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1390 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1393 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1395 (can_widen_mult_without_libcall): New function.
1396 (expand_mul_overflow): If only checking unsigned mul overflow,
1397 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1398 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1399 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1400 (expand_DIVMOD): Formatting fix.
1401 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1402 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1403 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1405 PR tree-optimization/82977
1406 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1407 constructed temporary to strlen_to_stridx.put.
1409 2017-11-15 Martin Liska <mliska@suse.cz>
1411 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1412 * configure: Regenerate.
1414 2017-11-15 Martin Liska <mliska@suse.cz>
1417 * config/sh/sh-mem.cc: Use proper probability for
1420 2017-11-14 Jeff Law <law@redhat.com>
1422 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1423 the red zone for stack_clash_protection_final_dynamic_probe targets
1424 when the total dynamic stack size is zero bytes.
1426 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1427 blocks is post order.
1429 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1431 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1432 * final.c (rest_of_clean_state): Set it for the
1433 -fcompare-debug dump.
1434 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1435 class when TDF_COMPARE_DEBUG is set.
1437 * dwarf2out.c (gen_producer_string): Discard
1440 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1443 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1444 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1445 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1446 (__TGMATH_REAL_2_3): Remove macros.
1447 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1448 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1450 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1452 (remquo): Define using __TGMATH_REAL_3.
1454 2017-11-14 Jeff Law <law@redhat.com>
1456 * vr-values.c: New file with contents extracted from tree-vrp.c.
1457 * Makefile.in (OBJS): Add vr-values.o
1458 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
1459 (set_value_range, set_and_canonicalize_value_range): Likewise.
1460 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
1461 (value_range_constant_singleton, symbolic_range_p): Likewise.
1462 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
1463 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
1464 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
1465 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
1466 (find_case_label_range, find_case_label_index): Likewise.
1467 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
1468 (range_int_cst_singleton_p, value_inside_range): Likewise.
1469 (get_single_symbol): Likewise.
1470 (switch_update): Move structure definition here.
1471 (to_remove_edges, to_update_switch_stmts): Provide externs.
1472 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
1473 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
1474 (vrp_val_is_min, set_value_range): Likewise.
1475 (set_and_canonicalize_value_range, copy_value_range): Likewise.
1476 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
1477 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
1478 (range_is_nonnull, range_int_cst_p): Likewwise.
1479 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
1480 (get_single_symbol, operand_less_p): Likewise
1481 (compare_values_warnv, compare_values): Likewise.
1482 (value_inside_range, value_range_constant_singleton): Likewise.
1483 (zero_nonzero_bitgs_from_vr): Likewise.
1484 (extract_range_from_binary_expr_1): Likewise.
1485 (overflow_comparison_p): Likewise.
1486 (to_remove_edges, to_update_switch_stmts): Likewise.
1487 (find_case_label-index, find_case_label_range): Likewise.
1488 (switch_update, set_value_range_to_nonnegative): Remove.
1489 (set_value_range_to_truthvalue): Likewise.
1490 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
1491 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1492 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1493 (find_case_label_ranges, test_for_singularity): Likewise.
1494 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1495 (x_vr_values): Move to its remaining use site.
1497 2017-11-10 Jeff Law <law@redhat.com>
1499 * vr-values.h (VR_INITIALIZER): Move #define here.
1500 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1501 * Makefile.in (OBJS): Add tree-evrp.o
1502 * tree-vrp.h (assert_info): Move structure definition here.
1503 (set_value_range_to_varying): Prototype.
1504 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1505 (infer_value_range, register_edge_assert_for): Likewise.
1506 (stmt_interesting_for_vrp): Likewise.
1507 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1508 (set_value_range_to_varying): No longer static.
1509 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1510 (infer_value_range, register_edge_assert_for): Likewise.
1512 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1514 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1515 generate the XXBRD instruction.
1517 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1518 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1519 as being compatible if -mabi=ieeelongdouble.
1520 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1521 to setup float128 built-ins with hardware support.
1522 (BU_FLOAT128_HW_2): Likewise.
1523 (BU_FLOAT128_HW_3): Likewise.
1524 (BU_FLOAT128_HW_VSX_1): Likewise.
1525 (BU_FLOAT128_HW_VSX_2): Likewise.
1526 (scalar_extract_expq): Change float128 built-in functions to
1527 accommodate having both KFmode and TFmode functions. Use the
1528 KFmode variant as the default.
1529 (scalar_extract_sigq): Likewise.
1530 (scalar_test_neg_qp): Likewise.
1531 (scalar_insert_exp_q): Likewise.
1532 (scalar_insert_exp_qp): Likewise.
1533 (scalar_test_data_class_qp): Likewise.
1534 (sqrtf128_round_to_odd): Delete processing the round to odd
1535 built-in functions as special built-in functions, and define them
1536 as float128 built-ins. Use the KFmode variant as the default.
1537 (truncf128_round_to_odd): Likewise.
1538 (addf128_round_to_odd): Likewise.
1539 (subf128_round_to_odd): Likewise.
1540 (mulf128_round_to_odd): Likewise.
1541 (divf128_round_to_odd): Likewise.
1542 (fmaf128_round_to_odd): Likewise.
1543 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1544 support for KFmode and TFmode xststdcqp calls.
1545 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1546 point, switch the built-in handlers for the get/set float128
1547 exponent, get float128 mantissa, float128 test built-ins, and the
1548 float128 round to odd built-in functions. Eliminate creating the
1549 float128 round to odd built-in functions as special built-ins.
1550 (rs6000_init_builtins): Eliminate special creation of the float128
1551 round to odd built-in functions.
1552 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1553 function insns to support both TFmode and KFmode variants.
1554 (xsxsigqp_<mode>): Likewise.
1555 (xsiexpqpf_<mode>): Likewise.
1556 (xsiexpqp_<mode>): Likewise.
1557 (xststdcqp_<mode>): Likewise.
1558 (xststdcnegqp_<mode>): Likewise.
1559 (xststdcqp_<mode>): Likewise.
1561 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1563 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1564 unused path_in_freq_ptr parameter.
1565 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1567 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1569 * ipa-inline.c (edge_badness): Dump sreal frequency.
1570 (compute_inlined_call_time): Match natural implementaiton ...
1571 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1572 forgotten division by CGRAPH_FREQ_BASE.
1574 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1576 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1577 Solaris 11. Update comment.
1578 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1580 * config/sol2.h (STARTFILE_SPEC): Likewise.
1581 * configure: Regenerate.
1583 2017-11-14 Carl Love <cel@us.ibm.com>
1585 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1586 le_ and be_ prefixes to swap* variables. Remove
1587 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1590 2017-11-14 Jason Merrill <jason@redhat.com>
1592 Support GTY((cache)) on hash_map.
1593 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1594 (ggc_cache_remove): Override it instead of ggc_mx.
1595 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1596 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1597 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1598 (simple_cache_map_traits): Override maybe_mx.
1599 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1600 (hash_map): Friend gt_cleare_cache.
1601 (gt_cleare_cache): New.
1602 * tree.h (tree_cache_traits): New hash_map traits class.
1603 (tree_cache_map): New typedef.
1605 2017-11-14 Richard Biener <rguenther@suse.de>
1607 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1608 paramter and handling.
1609 (cleanup_control_flow_bb): Likewise.
1610 (cleanup_control_flow_pre): New helper performing a DFS walk
1611 to call cleanup_control_flow_bb in PRE order.
1612 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1613 via cleanup_control_flow_pre.
1615 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1617 * config/aarch64/aarch64-simd.md
1618 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1619 (*aarch64_simd_bsl<mode>_alt): Likewise.
1620 (aarch64_simd_bsldi_internal): New.
1621 (aarch64_simd_bsldi_alt): Likewise.
1623 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1625 * tracer.c (better_p): Do not compare frequencies.
1626 * reg-stack.c (better_edge): Likewise.
1627 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1630 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1632 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1633 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1634 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1635 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1636 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1637 * predict.c (maybe_hot_frequency_p): Inline to ...
1638 (maybe_hot_count_p): ... here; rewrite to counts.
1639 (counts_to_freqs): Rename to ...
1640 (update_max_bb_count): ... this one.
1641 (expensive_function_p): Use counts.
1642 (estimate_bb_frequencies): Update.
1643 (rebuild_frequencies): Update.
1644 * predict.h (counts_to_freqs): Rename to ...
1645 (update_max_bb_count): ... this one.
1646 * profile.c (compute_branch_probabilities): Add debug info
1647 * tree-inline.c (expand_call_inline): Update debug info.
1648 (optimize_inline_calls): Use update_max_bb_count..
1649 (tree_function_versioning): Use update_max_bb_count..
1650 * value-prof.c (gimple_value_profile_transformations):
1651 Do not use update_max_bb_count.
1653 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1655 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1656 always use frequencies.
1658 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1660 * bb-reorder.c: Remove frequencies from comments.
1661 (better_edge_p): Use profile counts.
1662 (find_traces): Dump profile counts.
1663 (rotate_loop): Use profile counts.
1664 (find_traces_1_round): Likewise.
1665 (connect_better_edge_p): Use counts instead of probabilities for
1667 (copy_bb_p): Drop early check for non-0 frequency.
1668 (sanitize_hot_paths): Update comments.
1670 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1672 * ipa-split.c (struct split_point): Add count.
1673 (consider_split): Do not compute incoming frequency; compute incoming
1674 count and store it to split_point.
1675 (split_function): Set count of the call to split part correctly.
1677 2017-11-13 Carl Love <cel@us.ibm.com>
1679 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1681 2017-11-13 Tom Tromey <tom@tromey.com>
1683 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1685 2017-11-13 Carl Love <cel@us.ibm.com>
1687 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1688 Add support for builtins:
1689 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1690 vector {un,}signed {char,int,short},
1691 vector {un,}signed {char,int,short}) arguments.
1692 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1693 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1694 Add BU_P9V_AV_2 expansions for the builtins.
1695 * config/rs6000/altivec.h (vec_first_match_index,
1696 vec_first_mismatch_index, vec_first_match_or_eos_index,
1697 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1698 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1699 new extern declaration.
1700 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1701 * config/rs6000/vsx.md (first_match_index_<mode>,
1702 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1703 first_mismatch_or_eos_index_<mode>): Add define expand.
1704 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1705 * doc/extend.texi: Update the built-in documenation file for the new
1708 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1710 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1711 into the min/max operations for _Float<N> and _Float<N>X types.
1713 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1716 * dwarf2out.c (do_eh_frame): New static variable.
1717 (dwarf2out_begin_prologue): Set it.
1718 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1720 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1722 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1724 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1726 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1729 * debug.h (dwarf2out_do_eh_frame): Declare.
1730 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1731 (dwarf2out_do_frame): Use it.
1732 (dwarf2out_do_cfi_asm): Likewise.
1733 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1734 (dwarf2out_assembly_start): Likewise.
1735 (dwarf2out_begin_prologue): Fix comment.
1736 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1737 if the target needs either debug or unwind DWARF2 info.
1738 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1739 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1740 -ftrapping-math, -ftrapv and -fwrapv.
1742 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1744 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1745 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1746 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1747 (estimate_edge_size_and_time): Likewise.
1748 (ipa_merge_fn_summary_after_inlining): Likewise.
1749 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1750 (compute_uninlined_call_time): Use sreal_frequency.
1751 (compute_inlined_call_time): Likewise.
1752 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1753 * profile-count.c: Include sreal.h.
1754 (profile_count::to_sreal_scale): New.
1755 * profile-count.h: Forward declare sreal.
1756 (profile_count::to_sreal_scale): Declare.
1758 2017-11-13 Nathan Sidwell <nathan@acm.org>
1760 * diagnostic.c (maybe_line_and_column): New.
1761 (diagnostic_get_location_text): Use it.
1762 (diagnostic_report_current_module): Likewise.
1763 (test_diagnostic_get_location_text): Add tests.
1765 2017-11-13 Luis Machado <luis.machado@linaro.org>
1767 * doc/md.texi (Specifying processor pipeline description): Fix
1768 incorrect latency for the div instruction example.
1770 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1772 PR tree-optimization/78821
1773 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1774 that bit_not_p is the same.
1775 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1776 (split_group): Count precisely bit_not_p bits in each statement.
1777 (invert_op): New function.
1778 (imm_store_chain_info::output_merged_store): Use invert_op to
1779 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1780 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1782 2017-11-13 Martin Liska <mliska@suse.cz>
1784 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1785 (struct source_info): Likewise.
1786 (add_branch_counts): Likewise.
1787 (add_line_counts): Likewise.
1788 (function_summary): Likewise.
1789 (output_intermediate_line): Likewise.
1790 (generate_results): Likewise.
1792 2017-11-13 Martin Liska <mliska@suse.cz>
1794 * gcov.c (struct block_info): Remove typedef for block_t.
1795 (struct line_info): Likewise.
1796 (line_info::has_block): Likewise.
1797 (EXIT_BLOCK): Likewise.
1798 (unblock): Likewise.
1799 (circuit): Likewise.
1800 (get_cycles_count): Likewise.
1801 (process_file): Likewise.
1802 (read_graph_file): Likewise.
1803 (solve_flow_graph): Likewise.
1804 (find_exception_blocks): Likewise.
1805 (add_line_counts): Likewise.
1806 (accumulate_line_info): Likewise.
1807 (output_line_details): Likewise.
1809 2017-11-13 Martin Liska <mliska@suse.cz>
1811 * gcov.c (struct arc_info): Remove typedef for arc_t.
1812 (struct line_info): Likewise.
1813 (add_branch_counts): Likewise.
1814 (output_branch_count): Likewise.
1815 (function_info::~function_info): Likewise.
1816 (circuit): Likewise.
1817 (output_intermediate_line): Likewise.
1818 (read_graph_file): Likewise.
1819 (solve_flow_graph): Likewise.
1820 (find_exception_blocks): Likewise.
1821 (add_line_counts): Likewise.
1822 (accumulate_line_info): Likewise.
1823 (output_line_details): Likewise.
1824 (output_function_details): Likewise.
1826 2017-11-13 Martin Liska <mliska@suse.cz>
1828 * gcov.c (struct function_info): Remove typedef for function_t.
1829 (struct source_info): Likewise.
1830 (source_info::get_functions_at_location): Likewise.
1831 (solve_flow_graph): Likewise.
1832 (find_exception_blocks): Likewise.
1833 (add_line_counts): Likewise.
1834 (output_intermediate_file): Likewise.
1835 (process_file): Likewise.
1836 (generate_results): Likewise.
1837 (release_structures): Likewise.
1838 (read_graph_file): Likewise.
1839 (read_count_file): Likewise.
1840 (accumulate_line_counts): Likewise.
1841 (output_lines): Likewise.
1843 2017-11-13 Martin Liska <mliska@suse.cz>
1845 * gcov.c (function_info::function_info): Remove num_counts
1846 and add vector<gcov_type>.
1847 (function_info::~function_info): Use the vector.
1848 (process_file): Likewise.
1849 (read_graph_file): Likewise.
1850 (read_count_file): Likewise.
1851 (solve_flow_graph): Likewise.
1853 2017-11-13 Martin Liska <mliska@suse.cz>
1855 * gcov.c (function_info::is_artificial): New function.
1856 (process_file): Erase all artificial early.
1857 (generate_results): Skip as all artificial are already
1860 2017-11-13 Martin Liska <mliska@suse.cz>
1862 * gcov.c (read_graph_file): Store to global vector of functions.
1863 (read_count_file): Iterate the vector.
1864 (process_file): Likewise.
1865 (generate_results): Likewise.
1866 (release_structures): Likewise.
1868 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1870 PR tree-optimization/82954
1871 * gimple-ssa-store-merging.c
1872 (imm_store_chain_info::coalesce_immediate_stores): If
1873 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1875 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1877 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1878 Upddate call to ENDIAN_LANE_N.
1879 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1880 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1881 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1882 and use aarch64_endian_lane_rtx.
1883 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1885 2017-11-12 Tom de Vries <tom@codesourcery.com>
1887 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1889 2017-11-12 Tom de Vries <tom@codesourcery.com>
1891 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1893 2017-11-12 Tom de Vries <tom@codesourcery.com>
1895 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1897 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1898 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1899 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1901 2017-11-11 Martin Sebor <msebor@redhat.com>
1904 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1907 * prefic.c (translate_name): Replace strncpy with memcpy to
1908 avoid -Wstringop-truncation.
1910 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1912 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1914 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1916 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1918 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1920 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1923 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1925 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1926 enable generating XXBRH if the value is in a vector register.
1927 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1928 value is in a vector register.
1929 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1930 register to register bswap64's instead of doing the GPR sequence
1931 used on previous machines.
1932 (bswapdi2_xxbrd): New insn.
1933 (bswapdi2_reg): Disallow on ISA 3.0.
1934 (register to register bswap64 splitter): Do not split the insn on
1935 ISA 3.0 systems that use XXBRD.
1937 2017-11-10 Martin Sebor <msebor@redhat.com>
1940 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1942 (find_subframework_file): Same.
1944 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1946 * auto-profile.c (afdo_indirect_call): Drop frequency.
1947 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1948 (cgraph_node::create_edge): Drop frequency argument.
1949 (cgraph_node::create_indirect_edge): Drop frequency argument.
1950 (cgraph_edge::make_speculative): Drop frequency arguments.
1951 (cgraph_edge::resolve_speculation): Do not update frequencies
1952 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1953 (cgraph_node::dump): Check consistency in IPA mode.
1954 (cgraph_edge::maybe_hot_p): Use IPA counter.
1955 (cgraph_edge::verify_count_and_frequency): Rename to ...
1956 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1957 (cgraph_node::verify_node): Update.
1958 * cgraph.h (struct cgraph_edge): Drop frequency.
1959 (cgraph_edge::frequency): New function.
1960 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1962 (cgraph_edge::rebuild_edges): Likewise.
1963 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1964 (duplicate_thunk_for_node): Do not pass frequency.
1965 (cgraph_node::create_clone): Scale only counts.
1966 (cgraph_node::create_virtual_clone): Do not pass frequency.
1967 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1968 (cgraph_node::create_version_clone): Do not pass frequency.
1969 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1970 (cgraph_node::expand_thunk): Do not pass frequency.
1971 (cgraph_node::create_wrapper): Do not pass frequency.
1972 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1974 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1975 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1976 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1977 (gather_caller_stats): Use frequency function.
1978 (ipcp_cloning_candidate_p): Use frequency function.
1979 (ipcp_propagate_stage): Use frequency function.
1980 (get_info_about_necessary_edges): Use frequency function.
1981 (update_profiling_info): Update only IPA profile.
1982 (update_specialized_profile): Use frequency functoin.
1983 (perhaps_add_new_callers): Update only IPA profile.
1984 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1985 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1986 (dump_ipa_call_summary): Use frequency function.
1987 (estimate_edge_size_and_time): Use frequency function.
1988 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1989 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1990 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1991 (update_noncloned_counts): ... ths one; scale counts only.
1992 (clone_inlined_nodes): Do not scale frequency.
1993 (inline_call): Do not pass frequency.
1994 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1995 (compute_inlined_call_time): Use IPA profile.
1996 (want_inline_small_function_p): Use IPA profile.
1997 (want_inline_self_recursive_call_p): Use IPA profile.
1998 (edge_badness): Use IPA profile.
1999 (lookup_recursive_calls): Use IPA profile.
2000 (recursive_inlining): Do not pass frequency.
2001 (resolve_noninline_speculation): Do not update frequency.
2002 (inline_small_functions): Collect max of IPA profile.
2003 (dump_overall_stats): Dump IPA porfile.
2004 (dump_inline_stats): Dump IPA porfile.
2005 (ipa_inline): Collect IPA stats.
2006 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2007 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2008 (ipa_propagate_frequency): Use frequency function.
2009 (ipa_profile): Cleanup.
2010 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2011 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2012 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2013 (input_node): Do not stream frequency.
2014 (input_edge): Do not stream frequency.
2015 (merge_profile_summaries): Scale only IPA profiles.
2016 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2017 * predict.c (drop_profile): Do not recompute frequency.
2018 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2019 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2020 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2021 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2022 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2023 * tree-inline.c (copy_bb): Do not scale frequency.
2024 (expand_call_inline): Do not scale frequency.
2025 (tree_function_versioning): Do not scale frequency.
2026 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2028 2017-11-10 Julia Koval <julia.koval@intel.com>
2030 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2031 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2032 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2033 (_mm256_maskz_gf2p8affine_epi64_epi8)
2034 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2035 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2036 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2037 (__builtin_ia32_vgf2p8affineqb_v32qi)
2038 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2039 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2041 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2045 (arm_option_override): Refactor.
2046 (arm_option_reconfigure_globals): New.
2047 (arm_options_perform_arch_sanity_checks): New.
2048 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2050 (arm_options_perform_arch_sanity_checks): Likewise
2052 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2054 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2055 (power9-qpmul): New.
2056 * rs6000/rs6000.md ("type" attr): Add qmul.
2057 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2058 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2059 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2061 2017-11-10 Martin Sebor <msebor@redhat.com>
2064 * builtins.c (compute_objsize): Handle arrays that
2065 compute_builtin_object_size likes to fail for. Make extern.
2066 * builtins.h (compute_objsize): Declare.
2067 (check_strncpy_sizes): New function.
2068 (expand_builtin_strncpy): Call check_strncpy_sizes.
2069 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2070 -Wstringop-truncation.
2071 (gimple_fold_builtin_strncat): Same.
2072 * gimple.c (gimple_build_call_from_tree): Set call location.
2073 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2074 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2075 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2076 (handle_builtin_strlen): Use strlen_to_stridx.
2077 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2079 Use strlen_to_stridx.
2080 (pass_strlen::execute): Release strlen_to_stridx.
2081 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2082 (-Wstringop-truncation): Document new option.
2084 2017-11-10 Martin Liska <mliska@suse.cz>
2086 PR gcov-profile/82702
2087 * gcov.c (main): Handle intermediate files in a different
2089 (get_gcov_intermediate_filename): New function.
2090 (output_gcov_file): Remove support of intermediate files.
2091 (generate_results): Allocate intermediate file.
2092 (release_structures): Clean-up properly fn_end.
2093 (output_intermediate_file): Start iterating with line 1.
2095 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2097 PR tree-optimization/82929
2098 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2099 ops_swapped_p non-static data member.
2100 (store_immediate_info::store_immediate_info): Clear it.
2101 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2102 ops set ops_swapped_p.
2103 (count_multiple_uses): Handle ops_swapped_p.
2105 2017-11-10 Martin Liska <mliska@suse.cz>
2107 * coverage.c (coverage_init): Stream information about
2108 support of has_unexecuted_blocks.
2109 * doc/gcov.texi: Document that.
2110 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2111 * gcov.c (read_graph_file): Likewise.
2112 (output_line_beginning): Fix a small issue with
2115 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2117 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2118 reference of trivial component.
2120 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2123 * gimple-ssa-store-merging.c
2124 (pass_store_merging::terminate_all_aliasing_chains): For
2125 gimple_store_p stmts also call refs_output_dependent_p.
2127 PR rtl-optimization/82913
2128 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2131 2017-11-09 Jeff Law <law@redhat.com>
2133 * vr-values.h: New file with vr_values class.
2134 * tree-vrp.c: Include vr-values.h
2135 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2136 data objects into the vr_values class.
2137 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2138 (get_value_range): Make it a member function within vr_values class.
2139 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2140 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2141 (op_with_constant_singleton_value_range): Likewise.
2142 (extract_range_for_var_from_comparison_expr): Likewise.
2143 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2144 (extract_range_from_binary_expr): Likewise.
2145 (extract_range_from_unary_expr): Likewise.
2146 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2147 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2148 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2149 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2150 (compare_name_with_value, compare_names): Likewise.
2151 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2152 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2153 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2154 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2155 (extract_range_from_phi_node): Likewise.
2156 (simplify_truth_ops_using_ranges): Likewise.
2157 (simplify_div_or_mod_using_ranges): Likewise.
2158 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2159 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2160 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2161 (simplify_float_conversion_using_ranges): Likewise.
2162 (simplify_internal_call_using_ranges): Likewise.
2163 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2164 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2165 poitner via x_vr_values for calls into gimple folder.
2166 (vrp_initialize_lattice): Make this the vr_values ctor.
2167 (vrp_free_lattice): Make this the vr_values dtor.
2168 (set_vr_value): New function.
2169 (class vrp_prop): Add vr_values data member. Add various member
2170 functions as well as member functions that delegate to vr_values.
2171 (check_array_ref): Make a member function within vrp_prop class.
2172 (search_for_addr_array, vrp_initialize): Likewise.
2173 (vrp_finalize): Likewise. Revamp to avoid direct access to
2174 vr_value, values_propagated, etc.
2175 (check_array_bounds): Extract vrp_prop class instance pointer from
2176 walk info structure. Use it to call member functions.
2177 (check_all_array_refs): Make a member function within vrp_prop class.
2178 Smuggle class instance pointer via walk info structure.
2179 (x_vr_values): New local static.
2180 (vrp_valueize): Use x_vr_values to get class instance.
2181 (vr_valueize_1): Likewise.
2182 (class vrp_folder): Add vr_values data member. Add various member
2183 functions as well as member functions that delegate to vr_values.
2184 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2185 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2186 class instance from vr_values. Use it to call member functions.
2187 (vrp_dom_walker): Add vr_values data member.
2188 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2189 instance via x_vr_values.
2190 (identify_jump_threads): Accept vr_values as argument. Store
2191 it into the walker structure.
2192 (evrp_dom_walker): Add vr_values class data member. Add various
2194 (evrp_dom_walker::try_find_new_range): Use vr_values data
2195 member to access the memory allocator.
2196 (evrp_dom_walker::before_dom_children): Store vr_values class
2197 instance into the vrp_folder class.
2198 (evrp_dom_walker::push_value_range): Rework to avoid direct
2199 access to num_vr_values and vr_value.
2200 (evrp_dom_walker::pop_value_range): Likewise.
2201 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2202 Use vr_values to get to dump_all_value_ranges member function.
2203 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2204 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2205 Pass vr_values class instance down to identify_jump_threads.
2206 Remove call to vrp_free_lattice.
2207 (debug_all_value_ranges): Remove.
2209 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2210 (vrp_folder): Likewise.
2212 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2213 Get it from the existing bitmap instead.
2214 (vrp_intersect_ranges_1): Likewise.
2216 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2218 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2220 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2221 set bit_not_p to it.
2222 (imm_store_chain_info::coalesce_immediate_stores): Break group
2223 if bit_not_p is different.
2224 (count_multiple_uses, split_group,
2225 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2226 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2227 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2228 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2229 to store_immediate_info ctor.
2231 2017-11-09 Jim Wilson <jimw@sifive.com>
2233 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2234 (scan_prog_file): Likewise.
2236 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2238 * bb-reorder.c (max_entry_frequency): Remove.
2239 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2240 connect_traces, push_to_next_round_p): Remove prototypes.
2241 (find_traces_1_round): Use counts only.
2242 (push_to_next_round_p): Likewise.
2243 (find_traces): Likewise.
2244 (rotate_loop): Likewise.
2245 (find_traces_1_round): Likewise.
2246 (connect_traces): Likewise.
2247 (edge_order): Likewise.
2249 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2251 * config/arm/arm.c (output_return_instruction): Add comments to
2252 indicate requirement for cmse_nonsecure_entry return to account
2253 for the size of clearing instruction output here.
2254 (thumb_exit): Likewise.
2255 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2256 return in hardfloat mode.
2258 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2260 * config/rs6000/rs6000.c (machine_function): Add a bool,
2261 "toc_is_wrapped_separately".
2262 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2263 if it wasn't explicitly set or unset, we are optimizing for speed, and
2264 doing separate shrink-wrapping.
2265 (rs6000_get_separate_components): Enable the TOC component if
2266 saving the TOC register in the prologue.
2267 (rs6000_components_for_bb): Handle the TOC component.
2268 (rs6000_emit_prologue_components): Store the TOC register where needed.
2269 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2270 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2272 2017-11-09 Martin Jambor <mjambor@suse.cz>
2274 * ipa-param-manipulation.c: New file.
2275 * ipa-param-manipulation.h: Likewise.
2276 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2277 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2278 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2279 (ipa_parm_adjustment): Likewise.
2280 (ipa_parm_adjustment_vec): Likewise.
2281 (ipa_get_vector_of_formal_parms): Moved declaration to
2282 ipa-param-manipulation.h.
2283 (ipa_get_vector_of_formal_parm_types): Likewise.
2284 (ipa_modify_formal_parameters): Likewise.
2285 (ipa_modify_call_arguments): Likewise.
2286 (ipa_combine_adjustments): Likewise.
2287 (ipa_dump_param_adjustments): Likewise.
2288 (ipa_modify_expr): Likewise.
2289 (ipa_get_adjustment_candidate): Likewise.
2290 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2291 ipa-param-manipulation.c.
2292 (ipa_get_vector_of_formal_parm_types): Likewise.
2293 (ipa_modify_formal_parameters): Likewise.
2294 (ipa_modify_call_arguments): Likewise.
2295 (ipa_modify_expr): Likewise.
2296 (get_ssa_base_param): Likewise.
2297 (ipa_get_adjustment_candidate): Likewise.
2298 (index_in_adjustments_multiple_times_p): Likewise.
2299 (ipa_combine_adjustments): Likewise.
2300 (ipa_dump_param_adjustments): Likewise.
2301 * tree-sra.c: Also include ipa-param-manipulation.h
2302 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2305 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2306 Alan Hayward <alan.hayward@arm.com>
2307 David Sherwood <david.sherwood@arm.com>
2309 * doc/sourcebuild.texi (vect_masked_store): Document.
2311 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2312 Alan Hayward <alan.hayward@arm.com>
2313 David Sherwood <david.sherwood@arm.com>
2315 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2317 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2318 Alan Hayward <alan.hayward@arm.com>
2319 David Sherwood <david.sherwood@arm.com>
2321 * doc/sourcebuild.texi (vect_variable_length): Document.
2323 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2324 Alan Hayward <alan.hayward@arm.com>
2325 David Sherwood <david.sherwood@arm.com>
2327 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2329 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2330 Alan Hayward <alan.hayward@arm.com>
2331 David Sherwood <david.sherwood@arm.com>
2333 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2335 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2336 Alan Hayward <alan.hayward@arm.com>
2337 David Sherwood <david.sherwood@arm.com>
2339 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2340 previously undocumented selectors.
2341 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2343 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2345 * doc/rtl.texi (const_vector): Say that elements can be
2346 const_wide_ints too.
2347 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2348 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2349 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2350 * optabs.c (expand_vector_broadcast): Likewise.
2352 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2353 Alan Hayward <alan.hayward@arm.com>
2354 David Sherwood <david.sherwood@arm.com>
2356 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2357 scaled index even if the unscaled address was invalid.
2358 Don't increase the complexity of using a scale in that case.
2360 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2361 Alan Hayward <alan.hayward@arm.com>
2362 David Sherwood <david.sherwood@arm.com>
2364 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2365 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2366 UNITS_PER_WORD bytes.
2367 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2368 based on REGMODE_NATURAL_SIZE of the inner mode.
2369 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2370 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2371 * expmed.c (lowpart_bit_field_p): Divide the value up into
2372 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2373 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2374 whether something is likely to occupy more than one register.
2376 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2379 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2380 profile_count::adjust_for_ipa_scaling.
2381 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2382 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2384 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2386 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2388 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2389 (split_group): Add total_orig and total_new arguments, estimate the
2390 number of statements related to the store group without store merging
2391 and with store merging.
2392 (imm_store_chain_info::output_merged_store): Adjust split_group
2393 callers, punt if estimated number of statements with store merging
2394 is not smaller than estimated number of statements without it.
2396 (handled_load): Remove has_single_use checks.
2397 (pass_store_merging::process_store): Likewise.
2399 2017-11-09 Richard Biener <rguenther@suse.de>
2401 PR tree-optimization/82902
2402 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2404 2017-11-09 Martin Liska <mliska@suse.cz>
2407 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2410 2017-11-09 Martin Liska <mliska@suse.cz>
2412 PR tree-optimization/82669
2413 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2415 2017-11-09 Martin Liska <mliska@suse.cz>
2417 PR gcov-profile/48463
2418 * coverage.c (coverage_begin_function): Output also end locus
2419 of a function and information whether the function is
2421 * gcov-dump.c (tag_function): Parse and print the information.
2422 * gcov.c (INCLUDE_MAP): Add include.
2423 (INCLUDE_SET): Likewise.
2424 (struct line_info): Move earlier in the source file because
2425 of vector<line_info> in function_info structure.
2426 (line_info::line_info): Likewise.
2427 (line_info::has_block): Likewise.
2428 (struct source_info): Add new member index.
2429 (source_info::get_functions_at_location): New function.
2430 (function_info::group_line_p): New function.
2431 (output_intermediate_line): New function.
2432 (output_intermediate_file): Use the mentioned function.
2433 (struct function_start): New.
2434 (struct function_start_pair_hash): Likewise.
2435 (process_file): Add code that identifies group functions.
2436 Assign lines either to global or function scope.
2437 (generate_results): Skip artificial functions.
2438 (find_source): Assign index for each source file.
2439 (read_graph_file): Read new flag artificial and end_line.
2440 (add_line_counts): Assign it either to global of function scope.
2441 (accumulate_line_counts): Isolate core of the function to
2442 accumulate_line_info and call it for both function and global
2444 (accumulate_line_info): New function.
2445 (output_line_beginning): Fix GNU coding style.
2446 (print_source_line): New function.
2447 (output_line_details): Likewise.
2448 (output_function_details): Likewise.
2449 (output_lines): Iterate both source (global) scope and function
2451 (struct function_line_start_cmp): New class.
2452 * doc/gcov.texi: Reflect changes in documentation.
2454 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2457 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
2458 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
2459 and similarly for not instead of neg.
2461 2017-11-08 Andi Kleen <ak@linux.intel.com>
2463 * config/i386/i386.opt: Add -mforce-indirect-call.
2464 * config/i386/predicates.md: Check for flag_force_indirect_call.
2465 * doc/invoke.texi: Document -mforce-indirect-call
2467 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
2469 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
2471 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
2472 config/riscv/riscv.c (predict.h): New include.
2473 (riscv_slow_unaligned_access_p): No longer static.
2474 (riscv_block_move_straight): Add require.
2475 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2477 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2480 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
2481 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
2482 nonimmediate_operand predicate for operand 1 instead of
2485 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2487 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
2489 * config/aarch64/constraints.md (Uml): New constraint.
2490 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2493 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2495 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2496 of two vec_duplicates into a vec_concat.
2498 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2500 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2501 Simplify vec_merge of vec_duplicate and vec_concat.
2502 * config/aarch64/constraints.md (Utq): New constraint.
2503 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2506 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2508 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2509 Simplify vec_merge of vec_duplicate and const_vector.
2510 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2512 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2513 mode iterator. Update predicate on operand 1 to
2514 handle non-const_vec constants. Delete constraints.
2515 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2517 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2519 PR tree-optimization/78821
2520 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2522 (store_operand_info::store_operand_info): Initialize it to false.
2523 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2524 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2525 store in the group, and if chain_info is non-NULL, to ignore altogether
2527 (compatible_load_p): Fail if bit_not_p does not match.
2528 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2529 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2530 (pass_store_merging::process_store): Adjust
2531 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2532 call terminate_all_aliasing_chains newly when adding a store into
2533 a chain with non-NULL chain_info.
2535 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2537 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2539 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2541 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2543 (aarch64_layout_frame): Initialise emit_frame_chain.
2544 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2545 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2547 2017-11-08 Martin Liska <mliska@suse.cz>
2549 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2550 of gimple_call_internal_p.
2552 2017-11-07 Tom de Vries <tom@codesourcery.com>
2554 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2556 2017-11-07 Tom de Vries <tom@codesourcery.com>
2558 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2561 2017-11-08 Martin Liska <mliska@suse.cz>
2564 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2566 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2568 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2570 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2573 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2574 (*cmp<mode>_ccz_1): New insn with $k alternative.
2577 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2578 RTX_COMM_COMPARE as commutative as well.
2579 (ix86_binary_operator_ok): Formatting fix.
2580 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2581 *<code><mode>3<mask_name><round_saeonly_name>,
2582 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2583 *<s>mul<mode>3_highpart<mask_name>,
2584 *vec_widen_umult_even_v16si<mask_name>,
2585 *vec_widen_umult_even_v8si<mask_name>,
2586 *vec_widen_umult_even_v4si<mask_name>,
2587 *vec_widen_smult_even_v16si<mask_name>,
2588 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2589 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2590 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2591 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2592 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2593 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2594 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2595 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2596 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2597 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2598 ix86_binary_operator_ok. Formatting fixes.
2599 (*<plusminus_insn><mode>3<mask_name><round_name>,
2600 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2603 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2605 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2606 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2607 eq and ne if TARGET_ISEL.
2608 (cmp): New code_iterator.
2609 (UNS, UNSU_, UNSIK): New code_attrs.
2610 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2611 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2613 ("eq<mode>3"): ... this.
2614 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2616 ("ne<mode>3"): ... this.
2618 2017-11-07 Julia Koval <julia.koval@intel.com>
2621 * common/config/i386/i386-common.c
2622 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2623 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2624 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2625 * config/i386/i386.opt: Ditto.
2626 * config/i386/i386.c (ix86_target_string): Ditto.
2627 (ix86_option_override_internal): Ditto.
2628 (ix86_init_mpx_builtins): Move MPX to args2.
2629 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2630 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2631 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2632 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2633 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2634 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2635 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2637 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2640 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2641 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2642 (zero-extendsidi peephole2): Remove peephole.
2644 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2647 * stor-layout.c: Include attribs.h.
2648 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2649 explicit lookup of "aligned" attribute.
2651 2017-11-07 Andrew Waterman <andrew@sifive.com>
2653 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2654 (riscv_expand_block_move): Likewise.
2655 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2657 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2658 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2659 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2660 (riscv_adjust_block_mem): Likewise.
2661 (riscv_block_move_loop): Likewise.
2662 (riscv_expand_block_move): Likewise.
2663 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2665 2017-11-07 Michael Clark <michaeljclark@mac.com>
2667 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2668 (MUSL_DYNAMIC_LINKER): Likewise.
2670 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2672 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2673 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2676 2017-11-07 Richard Biener <rguenther@suse.de>
2678 * match.pd: Fix build.
2680 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2681 Jackson Woodruff <jackson.woodruff@arm.com>
2683 PR tree-optimization/71026
2684 * match.pd: Canonicalize negate in division.
2686 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2689 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2691 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2693 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2694 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2696 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2698 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2699 non-scalar integral types.
2700 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2701 (-(A-B), -(~A)): New transformations.
2703 2017-11-07 Tom de Vries <tom@codesourcery.com>
2705 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2706 semicolon after "do {} while (0)".
2707 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2708 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2709 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2710 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2711 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2713 2017-11-07 Tom de Vries <tom@codesourcery.com>
2715 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2716 after "do {} while (0)".
2717 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2718 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2719 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2720 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2721 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2723 2017-11-07 Tom de Vries <tom@codesourcery.com>
2726 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2728 (arm_rtx_costs_internal): Add missing semicolon after
2729 HANDLE_NARROW_SHIFT_ARITH call.
2731 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2733 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2734 disable isel if it was not set explicitly.
2736 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2738 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2739 (add_type_attribute) likewise.
2741 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2743 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2744 of struct ix86_frame.
2745 (ix86_initial_elimination_offset): Likewise.
2746 (ix86_expand_split_stack_prologue): Likewise.
2748 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2750 * tree-vrp.h (enum value_range_type): Update stale comment.
2752 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2753 Alan Hayward <alan.hayward@arm.com>
2754 David Sherwood <david.sherwood@arm.com>
2756 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2757 (aarch64_expand_vec_perm_const): Take the number of units too.
2758 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2759 (aarch64_expand_vec_perm_const): Likewise.
2760 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2761 (vec_perm<mode>): Update accordingly.
2763 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2764 Alan Hayward <alan.hayward@arm.com>
2765 David Sherwood <david.sherwood@arm.com>
2767 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2768 Take the number of units too.
2769 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2770 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2771 but check for a vector mode before rather than after the call.
2772 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2773 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2774 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2775 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2776 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2777 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2778 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2779 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2780 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2781 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2782 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2783 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2784 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2785 (aarch64_sqdmull2_n<mode>): Update accordingly.
2787 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2788 Alan Hayward <alan.hayward@arm.com>
2789 David Sherwood <david.sherwood@arm.com>
2791 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2792 the number of units too.
2793 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2794 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2795 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2796 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2797 (vec_store_lanesxi<mode>): Update accordingly.
2799 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2800 Alan Hayward <alan.hayward@arm.com>
2801 David Sherwood <david.sherwood@arm.com>
2803 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2804 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2805 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2806 of units rather than the mode.
2807 * config/aarch64/iterators.md (nunits): New mode attribute.
2808 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2809 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2810 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2811 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2812 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2813 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2814 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2815 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2816 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2817 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2818 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2819 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2820 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2821 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2822 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2823 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2824 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2825 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2826 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2827 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2828 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2829 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2830 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2831 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2832 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2833 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2834 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2835 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2836 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2837 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2838 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2839 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2840 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2841 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2842 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2843 (aarch64_simd_vec_setv2di): Likewise.
2845 2017-11-06 Carl Love <cel@us.ibm.com>
2847 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2849 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2850 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2852 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2854 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2855 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2856 Add power 8 macro expansions.
2857 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2858 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2859 power 8 instructions. (VSX_XXBR): Add iterator.
2861 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2863 * config/arm/arm.md (predicable_short_it): Change default to "no",
2864 improve documentation, remove uses that are identical to the default.
2865 (enabled_for_depr_it): Rename to enabled_for_short_it.
2866 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2867 Remove default uses.
2868 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2869 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2870 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2871 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2873 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2876 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2877 float128 helper macros, which are no longer used after deleting
2878 the old 'q' built-in functions, and moving the round to odd
2879 built-in functions to being special built-in functions.
2880 (BU_FLOAT128_2): Likewise.
2881 (BU_FLOAT128_1_HW): Likewise.
2882 (BU_FLOAT128_2_HW): Likewise.
2883 (BU_FLOAT128_3_HW): Likewise.
2884 (FABSQ): Delete old 'q' built-in functions.
2885 (COPYSIGNQ): Likewise.
2886 (SQRTF128_ODD): Move round to odd built-in functions to be
2887 special built-in functions, so that we can handle
2888 -mabi=ieeelongdouble.
2889 (TRUNCF128_ODD): Likewise.
2890 (ADDF128_ODD): Likewise.
2891 (SUBF128_ODD): Likewise.
2892 (MULF128_ODD): Likewise.
2893 (DIVF128_ODD): Likewise.
2894 (FMAF128_ODD): Likewise.
2895 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2896 built-in names to 'f128'.
2897 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2898 old 'q' built-in functions, as the machine independent code for
2899 'f128' built-in functions handles this.
2900 (rs6000_expand_builtin): Add expansion for float128 round to odd
2901 functions, keying off on -mabi=ieeelongdouble of whether to use
2902 the KFmode or TFmode variant.
2903 (rs6000_init_builtins): Initialize the _Float128 round to odd
2905 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2906 _Float128 'q' built-in functions are now mapped into the new
2907 'f128' built-in functions.
2909 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2911 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2912 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2914 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2917 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2918 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2919 * toplev.c: Include "ipa-fnsummary.h".
2920 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2922 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2924 PR tree-optimization/82838
2925 * gimple-ssa-store-merging.c
2926 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2927 on a separate gimple_seq which is then appended to seq.
2929 2017-11-06 Jeff Law <law@redhat.com>
2932 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2933 (get_probe_interval): New functions.
2934 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2935 (ix86_adjust_stack_and_probe): Likewise.
2936 (output_adjust_stack_and_probe): Likewise.
2937 (ix86_emit_probe_stack_range): Likewise.
2938 (ix86_expand_prologue): Likewise.
2940 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2942 PR tree-optimization/82816
2943 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2944 if the modes of the two types are the same.
2945 (convert_plusminus_to_widen): Likewise.
2947 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2949 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2951 (usadv16qi): New define_expand.
2952 (usadv8hi): New define_expand.
2954 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2957 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2959 (inline_transform): Likewise.
2960 * predict.c (counts_to_freqs): Remove useless conditional.
2961 * profile-count.h (profile_count::apply_scale): Move sanity check.
2962 * tree-inline.c (copy_bb): Always scale.
2963 (copy_cfg_body): Likewise.
2965 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2968 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2970 (*cmp_ite0): Add enabled_for_depr_it attribute.
2971 (*cmp_ite1): Likewise.
2973 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2975 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2978 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2980 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2981 return the value by pointer.
2982 (extract_range_from_multiplicative_op_1): Update accordingly.
2983 Return as soon as an operation fails.
2985 2017-11-05 Tom de Vries <tom@codesourcery.com>
2988 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2989 (DEF_SANITIZER_BUILTIN): ... here.
2990 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2991 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2993 2017-11-05 Tom de Vries <tom@codesourcery.com>
2996 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2998 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2999 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3000 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3002 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3003 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3004 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3006 2017-11-05 Tom de Vries <tom@codesourcery.com>
3009 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3012 2017-11-04 Michael Clark <michaeljclark@mac.com>
3014 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3015 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3017 (*addsi3_extended): Likewise.
3018 (*addsi3_extended2): Likewise.
3019 (<optab>si3): Likewise.
3020 (<optab>di3): Likewise.
3021 (<optab><mode>3): Likewise.
3022 (<*optabe>si3_internal): Likewise.
3023 (zero_extendqi<SUPERQI:mode>2): Likewise.
3024 (*add<mode>hi3): Likewise.
3025 (*xor<mode>hi3): Likewise.
3026 (<optab>di3): Likewise.
3027 (*<optab>si3_extend): Likewise.
3028 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3029 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3030 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3032 2017-11-04 Andrew Waterman <andrew@sifive.com>
3034 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3035 TARGET_STRICT_ALIGN based upon -mtune argument.
3037 2017-11-04 Andrew Waterman <andrew@sifive.com>
3039 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3041 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3043 config/i386/i386.c (choose_basereg): Use optional scratch
3044 register and add assertion.
3045 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3046 needed, and don't allocate stack.
3047 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3048 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3049 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3051 2017-11-03 Jeff Law <law@redhat.com>
3053 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3054 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3055 to probe at the start of a noreturn function.
3057 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3059 PR tree-optimization/78821
3060 * gimple-ssa-store-merging.c: Update the file comment.
3061 (MAX_STORE_ALIAS_CHECKS): Define.
3062 (struct store_operand_info): New type.
3063 (store_operand_info::store_operand_info): New constructor.
3064 (struct store_immediate_info): Add rhs_code and ops data members.
3065 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3066 and op1r arguments to the ctor, initialize corresponding data members.
3067 (struct merged_store_group): Add load_align_base and load_align
3069 (merged_store_group::merged_store_group): Initialize them.
3070 (merged_store_group::do_merge): Update them.
3071 (merged_store_group::apply_stores): Pick the constant for
3072 encode_tree_to_bitpos from one of the two operands, or skip
3073 encode_tree_to_bitpos if neither operand is a constant.
3074 (class pass_store_merging): Add process_store method decl. Remove
3075 bool argument from terminate_all_aliasing_chains method decl.
3076 (pass_store_merging::terminate_all_aliasing_chains): Remove
3077 var_offset_p argument and corresponding handling.
3078 (stmts_may_clobber_ref_p): New function.
3079 (compatible_load_p): New function.
3080 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3081 if there is overlap and rhs_code is not INTEGER_CST. For
3082 non-overlapping stores terminate group if rhs is not mergeable.
3083 (get_alias_type_for_stmts): Change first argument from
3084 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3085 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3086 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3088 (get_location_for_stmts): Change first argument from
3089 auto_vec<gimple *> & to vec<gimple *> &.
3090 (struct split_store): Remove orig_stmts data member, add orig_stores.
3091 (split_store::split_store): Create orig_stores rather than orig_stmts.
3092 (find_constituent_stmts): Renamed to ...
3093 (find_constituent_stores): ... this. Change second argument from
3094 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3095 to info structures rather than the statements.
3096 (split_group): Rename ALLOW_UNALIGNED argument to
3097 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3098 it. Adjust find_constituent_stores caller.
3099 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3100 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3101 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3102 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3103 (mem_valid_for_store_merging): New function.
3104 (handled_load): New function.
3105 (pass_store_merging::process_store): New method.
3106 (pass_store_merging::execute): Use process_store method. Adjust
3107 terminate_all_aliasing_chains caller.
3109 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3111 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3112 Return true for more constants, symbols and label references.
3113 (aarch64_valid_floating_const): Remove unused function.
3115 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3118 * config/aarch64/aarch64.c (aarch64_layout_frame):
3119 Undo forcing of LR at bottom of frame.
3121 2017-11-03 Jeff Law <law@redhat.com>
3124 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3125 for int_registers_saved.
3127 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3128 extracted from tree-ssa-dom.c.
3129 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3130 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3131 (record_equivalences_from_incoming_edge): Add additional argument
3132 to single_pred_edge_ignoring_loop_edges call.
3133 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3134 (uncprop_dom_walker::before_dom_children): Add additional argument
3135 to single_pred_edge_ignoring_loop_edges call.
3136 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3137 single_pred_edge_ignoring_loop_edges rather than open coding.
3138 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3140 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3142 * match.pd (-(-A)): Rewrite.
3144 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3146 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3147 (rs6000_emit_int_cmove): New declaration.
3148 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3149 (rs6000_emit_sISEL): Delete.
3150 (rs6000_emit_int_cmove): Make non-static.
3151 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3152 instead of rs6000_emit_sISEL.
3154 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3156 * asan.c (create_cond_insert_point): Maintain profile.
3157 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3159 * basic-block.h (struct basic_block_def): Remove frequency.
3160 (EDGE_FREQUENCY): Use to_frequency
3161 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3163 (find_traces): Update to use to_frequency.
3164 (find_traces_1_round): Likewise; use only IPA counts.
3165 (bb_to_key): Likewise.
3166 (connect_traces): Use IPA counts only.
3167 (copy_bb_p): Update to use to_frequency.
3168 (fix_up_crossing_landing_pad): Likewise.
3169 (sanitize_hot_paths): Likewise.
3170 * bt-load.c (basic_block_freq): Likewise.
3171 * cfg.c (init_flow): Set count_max to uninitialized.
3172 (check_bb_profile): Remove frequencies; check counts.
3173 (dump_bb_info): Do not dump frequencies.
3174 (update_bb_profile_for_threading): Update counts only.
3175 (scale_bbs_frequencies_int): Likewise.
3176 (MAX_SAFE_MULTIPLIER): Remove.
3177 (scale_bbs_frequencies_gcov_type): Update counts only.
3178 (scale_bbs_frequencies_profile_count): Update counts only.
3179 (scale_bbs_frequencies): Update counts only.
3180 * cfg.h (struct control_flow_graph): Add count-max.
3181 (update_bb_profile_for_threading): Update prototype.
3182 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3183 (find_many_sub_basic_blocks): Likewise.
3184 * cfgcleanup.c (try_forward_edges): Likewise.
3185 (try_crossjump_to_edge): Likewise.
3186 * cfgexpand.c (expand_gimple_cond): Likewise.
3187 (expand_gimple_tailcall): Likewise.
3188 (construct_init_block): Likewise.
3189 (construct_exit_block): Likewise.
3190 * cfghooks.c (verify_flow_info): Check consistency of counts.
3191 (dump_bb_for_graph): Do not dump frequencies.
3192 (split_block_1): Do not update frequencies.
3193 (split_edge): Do not update frequencies.
3194 (make_forwarder_block): Do not update frequencies.
3195 (duplicate_block): Do not update frequencies.
3196 (account_profile_record): Do not update frequencies.
3197 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3198 for global heuristics.
3199 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3200 (expected_loop_iterations_unbounded): Use counts only.
3201 * cfgloopmanip.c (scale_loop_profile): Simplify.
3202 (create_empty_loop_on_edge): Simplify
3204 (duplicate_loop_to_header_edge): Simplify
3205 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3206 (update_br_prob_note): Take care of removing note when profile
3208 (relink_block_chain): Do not dump frequency.
3209 (rtl_account_profile_record): Use to_frequency.
3210 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3211 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3212 (cgraph_update_edges_for_call_stmt_node): Likewise.
3213 (cgraph_edge::verify_count_and_frequency): Update.
3214 (cgraph_node::verify_node): Temporarily disable frequency verification.
3215 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3216 to_cgraph_frequency.
3217 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3218 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3220 (cgraph_node::expand_thunk): Update profile.
3221 * except.c (dw2_build_landing_pads): Do not update frequency.
3222 * final.c (compute_alignments): Use to_frequency.
3223 (dump_basic_block_info): Do not dump frequency.
3224 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3225 (dump_gimple_bb_header): Do not dump frequency.
3226 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3228 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3229 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3230 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3231 (init_before_recovery): Do not update frequency.
3232 (sched_create_recovery_edges): Do not update frequency.
3233 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3234 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3235 (ipa_cp_c_finalize): Set max_count to uninitialized.
3236 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3237 (param_change_prob): Use counts.
3238 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3240 * ipa-split.c (consider_split): Use to_frequency.
3241 (split_function): Use to_frequency.
3242 * ira-build.c (loop_compare_func): Likewise.
3243 (mark_loops_for_removal): Likewise.
3244 (mark_all_loops_for_removal): Likewise.
3245 * loop-doloop.c (doloop_modify): Do not update frequency.
3246 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3248 * lto-streamer-in.c (input_function): Update count_max.
3249 * omp-expand.c (expand_omp_taskreg): Update count_max.
3250 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3251 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3252 (maybe_hot_count_p): Use ipa counts only.
3253 (maybe_hot_bb_p): Simplify.
3254 (maybe_hot_edge_p): Simplify.
3255 (probably_never_executed): Do not take frequency argument.
3256 (probably_never_executed_bb_p): Do not pass frequency.
3257 (probably_never_executed_edge_p): Likewise.
3258 (combine_predictions_for_bb): Check that profile is nonzero.
3259 (propagate_freq): Do not set frequency.
3260 (drop_profile): Simplify.
3261 (counts_to_freqs): Simplify.
3262 (expensive_function_p): Use to_frequency.
3263 (propagate_unlikely_bbs_forward): Simplify.
3264 (determine_unlikely_bbs): Simplify.
3265 (estimate_bb_frequencies): Add hack to silence graphite issues.
3266 (compute_function_frequency): Use ipa counts.
3267 (pass_profile::execute): Update.
3268 (rebuild_frequencies): Use counts only.
3269 (force_edge_cold): Use counts only.
3270 * profile-count.c (profile_count::dump): Dump new count types.
3271 (profile_count::differs_from_p): Check compatiblity.
3272 (profile_count::to_frequency): New function.
3273 (profile_count::to_cgraph_frequency): New function.
3274 * profile-count.h (struct function): Declare.
3275 (enum profile_quality): Add profile_guessed_local and
3276 profile_guessed_global0.
3277 (class profile_proability): Decrease number of bits to 29;
3278 update from_reg_br_prob_note and to_reg_br_prob_note.
3279 (class profile_count: Update comment; decrease number of bits
3280 to 61. Check compatibility.
3281 (profile_count::compatible_p): New private member function.
3282 (profile_count::ipa_p): New member function.
3283 (profile_count::operator<): Handle global zero correctly.
3284 (profile_count::operator>): Handle global zero correctly.
3285 (profile_count::operator<=): Handle global zero correctly.
3286 (profile_count::operator>=): Handle global zero correctly.
3287 (profile_count::nonzero_p): New member function.
3288 (profile_count::force_nonzero): New member function.
3289 (profile_count::max): New member function.
3290 (profile_count::apply_scale): Handle IPA scalling.
3291 (profile_count::guessed_local): New member function.
3292 (profile_count::global0): New member function.
3293 (profile_count::ipa): New member function.
3294 (profile_count::to_frequency): Declare.
3295 (profile_count::to_cgraph_frequency): Declare.
3296 * profile.c (OVERLAP_BASE): Delete.
3297 (compute_frequency_overlap): Delete.
3298 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3299 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3300 * sched-ebb.c (rank): Use counts only.
3301 * shrink-wrap.c (handle_simple_exit): Use counts only.
3302 (try_shrink_wrapping): Use counts only.
3303 (place_prologue_for_one_component): Use counts only.
3304 * tracer.c (find_best_predecessor): Use to_frequency.
3305 (find_trace): Use to_frequency.
3306 (tail_duplicate): Use to_frequency.
3307 * trans-mem.c (expand_transaction): Do not update frequency.
3308 * tree-call-cdce.c: Do not update frequency.
3309 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3310 (gimple_merge_blocks): Likewise.
3311 (gimple_split_edge): Likewise.
3312 (gimple_duplicate_sese_region): Likewise.
3313 (gimple_duplicate_sese_tail): Likewise.
3314 (move_sese_region_to_fn): Likewise.
3315 (gimple_account_profile_record): Likewise.
3316 (insert_cond_bb): Likewise.
3317 * tree-complex.c (expand_complex_div_wide): Likewise.
3318 * tree-eh.c (lower_resx): Update profile.
3319 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3321 (initialize_cfun): Do not initialize frequencies
3322 (freqs_to_counts): Delete.
3323 (copy_cfg_body): Ignore count parameter.
3324 (copy_body): Update.
3325 (expand_call_inline): Update count_max.
3326 (optimize_inline_calls): Update count_max.
3327 (tree_function_versioning): Update count_max.
3328 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3329 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3331 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3332 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3333 (try_peel_loop): Likewise.
3334 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3336 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3337 (tree_transform_and_unroll_loop): Do not use frequencies
3338 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3339 Use reliable prediction only.
3340 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3341 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3342 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3343 probability scaling.
3344 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3345 not update frequency
3346 (any_remaining_duplicated_blocks): Likewise.
3347 (update_profile): Likewise.
3348 (estimated_freqs_path): Delete.
3349 (freqs_to_counts_path): Delete.
3350 (clear_counts_path): Delete.
3351 (ssa_fix_duplicate_block_edges): Likewise.
3352 (duplicate_thread_path): Likewise.
3353 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3354 * tree-tailcall.c (decrease_profile): Do not update frequency.
3355 (eliminate_tail_call): Likewise.
3356 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3357 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3358 (optimize_mask_stores): Likewise.
3359 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3360 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3361 (ubsan_expand_ptr_ifn): Update profile.
3362 * value-prof.c (gimple_ic): Simplify.
3363 * value-prof.h (gimple_ic): Update prototype.
3364 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3365 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3367 (want_inline_self_recursive_call_p): Likewise.
3368 (resolve_noninline_speculation): Only cummulate defined counts.
3369 (inline_small_functions): Use nonzero_p.
3370 (ipa_inline): Do not access freed node.
3372 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3374 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3375 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3377 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3379 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3380 non-legitimate address.
3382 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3384 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3385 (*lt0_<mode>di, *lt0_<mode>si): New.
3387 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3389 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3390 TARGET_PAIRED_FLOAT.
3392 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3393 Jim Wilson <jim.wilson@linaro.org>
3395 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3396 * config/aarch64/aarch64-tune.md: Regenerated.
3397 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3398 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3400 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3402 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3403 (arc_expand_prologue): Restore blink for millicode.
3404 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3406 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3409 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3410 gen_vec_duplicate after forcing the scalar into a register.
3412 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3414 * combine (try_combine): Print the insns input to try_combine to the
3417 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3420 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3421 Remove second argument from aarch64_process_target_attr call.
3422 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3424 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3426 (aarch64_handle_attr_arch): Remove second argument.
3427 (aarch64_handle_attr_cpu): Ditto.
3428 (aarch64_handle_attr_tune): Ditto.
3429 (aarch64_handle_attr_isa_flags): Ditto.
3430 (aarch64_process_one_target_attr): Ditto.
3431 (aarch64_process_target_attr): Ditto.
3432 (aarch64_option_valid_attribute_p): Remove second argument.
3433 on aarch64_process_target_attr call.
3435 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3437 * diagnostic.c: Include "selftest-diagnostic.h".
3438 (selftest::assert_location_text): New function.
3439 (selftest::test_diagnostic_get_location_text): New function.
3440 (selftest::diagnostic_c_tests): Call it.
3442 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3444 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3445 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3446 (class selftest::test_diagnostic_context): Move to...
3447 * selftest-diagnostic.c: New file.
3448 * selftest-diagnostic.h: New file.
3450 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3452 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3453 offset range for FT32B.
3454 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3455 * config/ft32/ft32.md: Add TARGET_NOPM.
3456 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3457 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
3459 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
3461 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
3463 2017-11-02 Jeff Law <law@redhat.com>
3465 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
3466 virtual keyword on FINAL OVERRIDE members.
3468 * tree-ssa-propagate.h (ssa_propagation_engine): Group
3469 virtuals together. Add virtual destructor.
3470 (substitute_and_fold_engine): Similarly.
3472 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3474 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
3476 2017-11-02 Richard Biener <rguenther@suse.de>
3478 PR tree-optimization/82795
3479 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
3481 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3483 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
3485 (gcc_GAS_CHECK_FEATURE): Remove.
3486 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
3487 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
3488 * configure: Regenerate.
3490 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3492 * config/arc/arc.c (hwloop_optimize): Account for empty
3495 2017-11-02 Richard Biener <rguenther@suse.de>
3498 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3499 Truncate ARRAY_REF index and element size.
3501 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3503 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3505 2017-11-01 Jeff Law <law@redhat.com>
3507 * tree-ssa-ccp.c (ccp_folder): New class derived from
3508 substitute_and_fold_engine.
3509 (ccp_folder::get_value): New member function.
3510 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3511 (ccp_fold_stmt): Remove prototype.
3512 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3513 * tree-ssa-copy.c (copy_folder): New class derived from
3514 substitute_and_fold_engine.
3515 (copy_folder::get_value): Renamed from get_value.
3516 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3517 * tree-vrp.c (vrp_folder): New class derived from
3518 substitute_and_fold_engine.
3519 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3520 (vrp_folder::get_value): New member function.
3521 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3522 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3523 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3524 provide a class interface to folder/substitute routines.
3525 (ssa_prop_fold_stmt_fn): Remove typedef.
3526 (ssa_prop_get_value_fn): Likewise.
3527 (subsitute_and_fold): Remove prototype.
3528 (replace_uses_in): Likewise.
3529 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3530 Renamed from replace_uses_in. Call the virtual member function
3531 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3532 (substitute_and_fold_dom_walker): Remove initialization of
3533 data member entries for calbacks. Add substitute_and_fold_engine
3534 member and initialize it.
3535 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3536 member functions for get_value, replace_phi_args_in c
3537 replace_uses_in, and fold_stmt calls.
3538 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3539 substitute_and_fold. Remove assert. Update ctor call.
3541 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3542 (ssa_prop_visit_phi_fn): Likewise.
3543 (class ssa_propagation_engine): New class to provide an interface
3545 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3547 (ssa_prop_visit_phi): Likewise.
3548 (ssa_propagation_engine::simulate_stmt): Moved into class.
3549 Call visit_phi/visit_stmt from the class rather than via
3550 file scoped static variables.
3551 (ssa_propagation_engine::simulate_block): Moved into class.
3552 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3553 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3554 set file scoped statics for the visit_stmt/visit_phi callbacks.
3555 * tree-complex.c (complex_propagate): New class derived from
3556 ssa_propagation_engine.
3557 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3558 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3559 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3561 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3562 ssa_propagation_engine.
3563 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3564 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3565 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3566 * tree-ssa-copy.c (copy_prop): New class derived from
3567 ssa_propagation_engine.
3568 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3569 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3570 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3571 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3572 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3573 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3574 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3576 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3578 PR rtl-optimization/82778
3579 PR rtl-optimization/82597
3580 * compare-elim.c (struct comparison): Add in_a_setter field.
3581 (find_comparison_dom_walker::before_dom_children): Remove killed
3582 bitmap and df_simulate_find_defs call, instead walk the defs.
3583 Compute last_setter and initialize in_a_setter. Merge definitions
3584 with first initialization for a few variables.
3585 (try_validate_parallel): Use insn_invalid_p instead of
3586 recog_memoized. Return insn rather than just the pattern.
3587 (try_merge_compare): Fix up comment. Don't uselessly test if
3588 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3590 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3593 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3594 Alan Hayward <alan.hayward@arm.com>
3595 David Sherwood <david.sherwood@arm.com>
3597 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3598 aarch64_hard_regno_nregs to get the number of registers
3601 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3602 Alan Hayward <alan.hayward@arm.com>
3603 David Sherwood <david.sherwood@arm.com>
3605 * config/aarch64/constraints.md (Upl): Rename to...
3607 * config/aarch64/aarch64.md
3608 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3611 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3612 Alan Hayward <alan.hayward@arm.com>
3613 David Sherwood <david.sherwood@arm.com>
3615 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3616 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3619 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3620 Alan Hayward <alan.hayward@arm.com>
3621 David Sherwood <david.sherwood@arm.com>
3623 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3624 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3625 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3627 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3628 of a single element.
3629 * config/aarch64/iterators.md: Add a comment above the permute
3630 unspecs to say that they are generated directly by
3631 aarch64_expand_vec_perm_const.
3632 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3634 2017-11-01 Nathan Sidwell <nathan@acm.org>
3636 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3638 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3640 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3641 and medany code models, and describe what they do.
3643 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3645 Revert accidental duplicate:
3647 * combine.c (can_change_dest_mode): Reject changes in
3648 REGMODE_NATURAL_SIZE.
3650 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3652 PR rtl-optimization/64682
3653 PR rtl-optimization/69567
3654 PR rtl-optimization/69737
3655 PR rtl-optimization/82683
3656 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3657 register mentioned in the note, drop the note, unless it came from I3,
3658 in which case it should go to I3 again.
3660 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3662 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3663 and return false if not.
3664 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3668 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3669 range is known to be empty.
3671 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3672 Alan Hayward <alan.hayward@arm.com>
3673 David Sherwood <david.sherwood@arm.com>
3675 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3676 and CONST_VECTOR_NUNITS instead of computing the number of units from
3677 the byte sizes of the vector and element.
3678 (simplify_binary_operation_1): Likewise.
3679 (simplify_const_binary_operation): Likewise.
3680 (simplify_ternary_operation): Likewise.
3682 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3683 Alan Hayward <alan.hayward@arm.com>
3684 David Sherwood <david.sherwood@arm.com>
3686 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3687 (int_mem_offset): ...this new function.
3688 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3689 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3690 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3693 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3694 Alan Hayward <alan.hayward@arm.com>
3695 David Sherwood <david.sherwood@arm.com>
3697 * lower-subreg.c (interesting_mode_p): New function.
3698 (compute_costs, find_decomposable_subregs, decompose_register)
3699 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3700 (resolve_clobber, dump_choices): Use it.
3702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3703 Alan Hayward <alan.hayward@arm.com>
3704 David Sherwood <david.sherwood@arm.com>
3706 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3709 Alan Hayward <alan.hayward@arm.com>
3710 David Sherwood <david.sherwood@arm.com>
3712 * alias.c (find_base_value, find_base_term): Only process integer
3713 truncations. Check the precision rather than the size.
3715 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3716 Alan Hayward <alan.hayward@arm.com>
3717 David Sherwood <david.sherwood@arm.com>
3719 * machmode.h (is_narrower_int_mode): New function
3720 * optabs.c (expand_float, expand_fix): Use it.
3721 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3723 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3724 Alan Hayward <alan.hayward@arm.com>
3725 David Sherwood <david.sherwood@arm.com>
3727 * rtl.h (narrower_subreg_mode): New function.
3728 * ira-color.c (update_costs_from_allocno): Use it.
3730 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3731 Alan Hayward <alan.hayward@arm.com>
3732 David Sherwood <david.sherwood@arm.com>
3734 * optabs-query.h (convert_optab_p): New function, split out from...
3735 (convert_optab_handler): ...here.
3736 (widening_optab_handler): Delete.
3737 (find_widening_optab_handler): Remove permit_non_widening parameter.
3738 (find_widening_optab_handler_and_mode): Likewise. Provide an
3739 override that operates on mode class wrappers.
3740 * optabs-query.c (widening_optab_handler): Delete.
3741 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3742 parameter. Assert that the two modes are the same class and that
3743 the "from" mode is narrower than the "to" mode. Use
3744 convert_optab_handler instead of widening_optab_handler.
3745 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3746 instead of widening_optab_handler.
3747 * expr.c (expand_expr_real_2): Update calls to
3748 find_widening_optab_handler.
3749 * optabs.c (expand_widen_pattern_expr): Likewise.
3750 (expand_binop_directly): Take the insn_code as a parameter.
3751 (expand_binop): Only call find_widening_optab_handler for
3752 conversion optabs; use optab_handler otherwise. Update calls
3753 to find_widening_optab_handler and expand_binop_directly.
3754 Use convert_optab_handler instead of widening_optab_handler.
3755 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3756 find_widening_optab_handler and use scalar_mode rather than
3758 (convert_plusminus_to_widen): Likewise.
3760 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3761 Alan Hayward <alan.hayward@arm.com>
3762 David Sherwood <david.sherwood@arm.com>
3764 * machmode.h (fixed_size_mode): New class.
3765 * rtl.h (get_pool_mode): Return fixed_size_mode.
3766 * gengtype.c (main): Add fixed_size_mode.
3767 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3768 (get_raw_arg_mode): Likewise.
3769 * doc/tm.texi: Regenerate.
3770 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3771 * targhooks.c (default_get_reg_raw_mode): Likewise.
3772 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3773 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3774 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3775 (msp430_get_raw_result_mode): Likewise.
3776 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3777 * dbxout.c (dbxout_parms): Require fixed-size modes.
3778 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3779 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3780 * omp-low.c (lower_oacc_reductions): Likewise.
3781 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3782 (simplify_subreg): Update accordingly.
3783 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3784 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3785 that aren't fixed-size.
3786 (get_pool_mode): Return a fixed_size_mode.
3787 (output_constant_pool_2): Take a fixed_size_mode.
3789 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3790 Alan Hayward <alan.hayward@arm.com>
3791 David Sherwood <david.sherwood@arm.com>
3793 * doc/rtl.texi (vec_series): Document.
3794 (const): Say that the operand can be a vec_series.
3795 * rtl.def (VEC_SERIES): New rtx code.
3796 * rtl.h (const_vec_series_p_1): Declare.
3797 (const_vec_series_p): New function.
3798 * emit-rtl.h (gen_const_vec_series): Declare.
3799 (gen_vec_series): Likewise.
3800 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3801 (gen_vec_series): Likewise.
3802 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3803 * simplify-rtx.c (simplify_unary_operation): Handle negations
3805 (simplify_binary_operation_series): New function.
3806 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3807 (test_vector_ops_series): New function.
3808 (test_vector_ops): Call it.
3809 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3810 gen_const_vec_series.
3811 (altivec_lvsr): Likewise.
3812 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3814 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3815 Alan Hayward <alan.hayward@arm.com>
3816 David Sherwood <david.sherwood@arm.com>
3818 * doc/rtl.texi (const): Update description of address constants.
3819 Say that vector constants are allowed too.
3820 * common.md (E, F): Use CONSTANT_P instead of checking for
3822 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3823 checking for CONST_VECTOR.
3824 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3826 * expr.c (expand_expr_real_2): Check for vector modes instead
3827 of checking for CONST_VECTOR.
3828 * rtl.h (const_vec_p): New function.
3829 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3830 (unwrap_const_vec_duplicate): Handle them here too.
3832 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3833 David Malcolm <dmalcolm@redhat.com>
3834 Alan Hayward <alan.hayward@arm.com>
3835 David Sherwood <david.sherwood@arm.com>
3837 * rtl.h (vec_duplicate_p): New function.
3838 * selftest-rtl.c (assert_rtx_eq_at): New function.
3839 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3840 (assert_rtx_eq_at): Declare.
3841 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3842 * selftest-run-tests.c (selftest::run_tests): Call it.
3843 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3844 (simplify_unary_operation_1): Recursively handle vector duplicates.
3845 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3847 (simplify_subreg): Handle subregs of vector duplicates.
3848 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3849 (selftest::simplify_rtx_c_tests): New functions.
3851 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3852 Alan Hayward <alan.hayward@arm.com>
3853 David Sherwood <david.sherwood@arm.com>
3855 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3856 (gen_vec_duplicate): Likewise.
3857 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3859 (gen_const_vector): ...here.
3860 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3861 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3862 whose elements are all equal.
3863 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3864 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3865 (simplify_relational_operation): Likewise.
3866 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3868 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3869 (aarch64_expand_vector_init): Likewise.
3870 * config/arm/arm.c (neon_vdup_constant): Likewise.
3871 (neon_expand_vector_init): Likewise.
3872 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3873 (arm_block_set_unaligned_vect): Likewise.
3874 (arm_block_set_aligned_vect): Likewise.
3875 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3876 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3877 (expand_vec_perm_even_odd_pack): Likewise.
3878 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3879 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3880 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3881 gen_const_vec_duplicate.
3882 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3883 * config/mips/mips.c (mips_gen_const_int_vector): Use
3884 gen_const_vec_duplicate.
3885 (mips_expand_vector_init): Use CONST0_RTX.
3886 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3887 (define_split): Use gen_const_vec_duplicate.
3888 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3889 (define_split): Use gen_const_vec_duplicate.
3890 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3891 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3892 * config/spu/spu.c (spu_const): Likewise.
3894 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3895 Alan Hayward <alan.hayward@arm.com>
3896 David Sherwood <david.sherwood@arm.com>
3898 * combine.c (can_change_dest_mode): Reject changes in
3899 REGMODE_NATURAL_SIZE.
3901 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3903 * configure.ac (--enable-libssp): New.
3904 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3905 trying to determine target-specific default. Adjust indentation.
3906 * configure: Regenerated.
3907 * doc/install.texi (Configuration): Expand --disable-libssp
3910 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3912 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3915 2017-10-31 Martin Jambor <mjambor@suse.cz>
3918 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3920 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3922 * auto-profile.c (autofdo_source_profile::read): Use
3923 UNKNOWN_LOCATION rather than 0.
3924 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3925 (warning_at): ...this overload.
3926 (warning_at_rich_loc_n): Rename to...
3927 (warning_n): ...this overload.
3928 (error_at_rich_loc): Rename to...
3929 (error_at): ...this overload.
3930 (pedwarn_at_rich_loc): Rename to...
3931 (pedwarn): ...this overload.
3932 (permerror_at_rich_loc): Rename to...
3933 (permerror): ...this overload.
3934 (inform_at_rich_loc): Rename to...
3935 (inform): ...this overload.
3936 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3937 (diagnostic_n_impl_richloc): Rename to...
3938 (diagnostic_n_impl): ...this rich_location *-based decl.
3939 (inform_at_rich_loc): Rename to...
3940 (inform): ...this, and add an assertion.
3941 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3942 (warning_at_rich_loc): Rename to...
3943 (warning_at): ...this, and add an assertion.
3944 (warning_at_rich_loc_n): Rename to...
3945 (warning_n): ...this, and add an assertion.
3946 (warning_n): Update location_t-based implementation for removal of
3947 location_t-based diagnostic_n_impl.
3948 (pedwarn_at_rich_loc): Rename to...
3949 (pedwarn): ...this, and add an assertion.
3950 (permerror_at_rich_loc): Rename to...
3951 (permerror): ...this, and add an assertion.
3952 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3953 (error_at_rich_loc): Rename to...
3954 (error_at): ...this, and add an assertion.
3955 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3956 (driver::do_spec_on_infiles): Likewise.
3957 * substring-locations.c (format_warning_va): Update for renaming
3958 of inform_at_rich_loc.
3960 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3962 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3963 _Float<N>X built-in functions so that the variant without the
3964 "__builtin_" prefix is only enabled for the GNU C and Objective C
3965 languages when they are in non-strict ANSI/ISO mode.
3966 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3967 * target.def (floatn_builtin_p): Add a target hook to control
3968 whether _Float<N> and _Float<N>X built-in functions without the
3969 "__builtin_" prefix are enabled, and return true for C and
3970 Objective C in the default hook. Include langhooks.h in
3972 * targhooks.h (default_floatn_builtin_p): Likewise.
3973 * targhooks.c (default_floatn_builtin_p): Likewise.
3974 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3975 floatn_builtin_p target hook.
3976 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3978 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3979 Eric Botcazou <ebotcazou@adacore.com>
3981 PR rtl-optimization/81803
3982 * lra-constraints.c (curr_insn_transform): Also reload the whole
3983 register for a strict subreg no wider than a word if this is for
3984 a WORD_REGISTER_OPERATIONS target.
3986 2017-10-31 Jason Merrill <jason@redhat.com>
3988 * gdbinit.in: Skip over inlines from timevar.h.
3990 2017-10-31 Martin Liska <mliska@suse.cz>
3992 * doc/gcov.texi: Document new option.
3993 * gcov.c (print_usage): Likewise print it.
3994 (process_args): Support the argument.
3995 (format_count): New function.
3996 (format_gcov): Use the function.
3998 2017-10-31 Martin Liska <mliska@suse.cz>
4000 * gcov.c (struct name_map): do not use typedef.
4001 Define operator== and operator<.
4002 (name_search): Remove.
4003 (name_sort): Remove.
4004 (main): Do not allocate names.
4005 (process_file): Add vertical space.
4006 (generate_results): Use std::find.
4007 (release_structures): Do not release memory.
4008 (find_source): Use std::find.
4010 2017-10-31 Martin Liska <mliska@suse.cz>
4012 * gcov.c (struct line_info): Remove it's typedef.
4013 (line_info::line_info): Add proper ctor.
4014 (line_info::has_block): Do not use a typedef.
4015 (struct source_info): Do not use typedef.
4016 (circuit): Likewise.
4017 (get_cycles_count): Likewise.
4018 (output_intermediate_file): Iterate via vector iterator.
4019 (add_line_counts): Use std::vector methods.
4020 (accumulate_line_counts): Likewise.
4021 (output_lines): Likewise.
4023 2017-10-31 Martin Liska <mliska@suse.cz>
4025 * gcov.c (struct source_info): Remove typedef.
4026 (source_info::source_info): Add proper ctor.
4027 (accumulate_line_counts): Use struct, not it's typedef.
4028 (output_gcov_file): Likewise.
4029 (output_lines): Likewise.
4030 (main): Do not allocate an array.
4031 (output_intermediate_file): Use size of vector container.
4032 (process_file): Resize the vector.
4033 (generate_results): Do not preallocate, use newly added vector
4035 (release_structures): Do not release sources.
4036 (find_source): Use vector methods.
4037 (add_line_counts): Do not use typedef.
4039 2017-10-31 Martin Liska <mliska@suse.cz>
4041 * doc/gcov.texi: Document that.
4042 * gcov.c (add_line_counts): Mark lines with a non-executed
4044 (output_line_beginning): Handle such lines.
4045 (output_lines): Pass new argument.
4046 (output_intermediate_file): Print it in intermediate format.
4048 2017-10-31 Martin Liska <mliska@suse.cz>
4050 * color-macros.h: New file.
4051 * diagnostic-color.c: Factor out color related to macros to
4053 * doc/gcov.texi: Document -k option.
4054 * gcov.c (INCLUDE_STRING): Include string.h.
4055 (print_usage): Add -k option.
4056 (process_args): Parse it.
4057 (pad_count_string): New function.
4058 (output_line_beginning): Likewise.
4059 (DEFAULT_LINE_START): New macro.
4060 (output_lines): Support color output.
4062 2017-10-31 Martin Liska <mliska@suse.cz>
4064 PR gcov-profile/82633
4065 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4066 their interaction with GCOV infrastructure.
4067 * configure.ac: Add -fkeep-{inline,static}-functions to
4069 * configure: Regenerate.
4071 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4074 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4076 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4079 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4080 into a register if it does not fit into an immediate offset field.
4082 2017-10-31 Olivier Hainque <hainque@adacore.com>
4084 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4086 2017-10-31 Julia Koval <julia.koval@intel.com>
4088 * config.gcc: Add gfniintrin.h.
4089 * config/i386/gfniintrin.h: New.
4090 * config/i386/i386-builtin-types.def
4091 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4092 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4093 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4094 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4095 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4096 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4097 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4098 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4099 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4100 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4101 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4102 * config/i386/immintrin.h: Include gfniintrin.h.
4103 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4105 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4107 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4109 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4111 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4112 (arm_ashldi3_1bit): Remove pattern.
4113 (ashrdi3): Remove shift by 1 expansion.
4114 (arm_ashrdi3_1bit): Remove pattern.
4115 (lshrdi3): Remove shift by 1 expansion.
4116 (arm_lshrdi3_1bit): Remove pattern.
4117 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4118 cost of ashldi3 by 1.
4119 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4120 (<shift>di3_neon): Likewise.
4122 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4124 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4125 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4126 and (*aarch64_simd_mov<VQ:mode>).
4127 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4128 pattern alternative.
4129 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4130 attributes to match pattern alternative.
4132 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4134 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4135 * config/rs6000/emmintrin.h: New file.
4136 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4138 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4140 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4141 * (movdi_vfp_cortexa8): Remove pattern.
4143 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4145 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4147 (Specific, alpha*-dec-osf5.1): Remove.
4148 (Specific, mips-sgi-irix5): Remove.
4149 (Specific, mips-sgi-irix6): Remove.
4151 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4154 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4155 arguments to clear_bit_region_be.
4157 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4159 * gimplify.c: Include memmodel.h.
4161 2017-10-30 Martin Jambor <mjambor@suse.cz>
4163 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4164 condition whether loop should be executed at all.
4166 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4169 gimple folding of vec_madd() intrinsics.
4170 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4171 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4172 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4174 2017-10-30 Richard Biener <rguenther@suse.de>
4176 PR tree-optimization/82762
4178 2017-10-23 Richard Biener <rguenther@suse.de>
4180 PR tree-optimization/82129
4182 2017-08-01 Richard Biener <rguenther@suse.de>
4184 PR tree-optimization/81181
4185 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4186 (compute_antic): ... end of iteration here.
4188 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4190 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4191 -std=iso9899:2017 and -std=gnu17.
4192 * doc/standards.texi (C Language): Document C17 support.
4193 * doc/cpp.texi (Overview): Mention -std=c17.
4194 (Standard Predefined Macros): Document C11 and C17 values of
4195 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4196 * doc/extend.texi (Inline): Do not list individual options for
4197 standards newer than C99.
4198 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4200 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4203 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4205 * asan.c (asan_finish_file): Align asan globals array by shadow
4208 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4211 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4212 (struct store_immediate_info): Add bitregion_start and bitregion_end
4214 (store_immediate_info::store_immediate_info): Add brs and bre
4215 arguments and initialize bitregion_{start,end} from those.
4216 (struct merged_store_group): Add bitregion_start, bitregion_end,
4217 align_base and mask fields. Drop unnecessary struct keyword from
4218 struct store_immediate_info. Add do_merge method.
4219 (clear_bit_region_be): Use memset instead of loop storing zeros.
4220 (merged_store_group::do_merge): New method.
4221 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4222 stores as long as the surrounding bitregions have no gaps.
4223 (merged_store_group::merge_overlapping): Use do_merge.
4224 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4225 is byte aligned, rather than requiring that start and width are
4226 byte aligned. Drop unnecessary struct keyword from
4227 struct store_immediate_info. Allocate and populate also mask array.
4228 Make start of the arrays relative to bitregion_start rather than
4229 start and size them according to bitregion_{end,start} difference.
4230 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4231 struct store_immediate_info.
4232 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4233 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4234 struct keyword from struct store_immediate_info.
4235 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4236 between stores as long as the surrounding bitregions have no gaps.
4238 (struct split_store): Add orig non-static data member.
4239 (split_store::split_store): Initialize orig to false.
4240 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4241 if there is exactly a single original stmt. Change stmts argument
4242 to pointer from reference, if NULL, don't push anything to it. Add
4243 first argument, use it to optimize skipping over orig stmts that
4244 are known to be before bitpos already. Simplify.
4245 (split_group): Return unsigned int count how many stores are or
4246 would be needed rather than a bool. Add allow_unaligned argument.
4247 Change split_stores argument from reference to pointer, if NULL,
4248 only do a dry run computing how many stores would be produced.
4249 Rewritten algorithm to use both alignment and misalign if
4250 !allow_unaligned and handle bitfield stores with gaps.
4251 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4252 from bitregion_start instead of start. Compute allow_unaligned
4253 here, if true, do 2 split_group dry runs to compute which one
4254 produces fewer stores and prefer aligned if equal. Punt if
4255 new count is bigger or equal than original before emitting any
4256 statements, rather than during that. Remove no longer needed
4257 new_ssa_names tracking. Replace num_stmts with
4258 split_stores.length (). Use 32-bit stack allocated entries
4259 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4260 if possible. Handle bitfields with gaps.
4261 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4262 Compute bitregion_{start,end} for the stores and construct
4263 store_immediate_info with that. Formatting fixes.
4265 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4268 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4269 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4271 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4273 * gimplify.c: Include tm_p.h.
4275 * common.opt (gcoff): Re-add as ignored option.
4276 (gcoff1, gcoff2, gcoff3): Likewise.
4278 * Makefile.in (OBJS): Delete sdbout.o.
4279 (GTFILES): Delete $(srcdir)/sdbout.c.
4280 * debug.h: Delete sdb_debug_hooks.
4281 * final.c: Delete sdbout.h include.
4282 (final_scan_insn): Delete SDB_DEBUG check.
4283 (rest_of_clean_state): Likewise.
4284 * output.h: Delete sdb_begin_function_line.
4287 * toplev.c: Delete sdbout.h include.
4288 (process_options): Delete SDB_DEBUG check.
4289 * tree-core.h (tree_type_common): Delete pointer field of
4291 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4292 TYPE_SYMTAB_POINTER.
4293 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4294 (TYPE_SYMTAB_IS_POINTER): Delete.
4295 (TYPE_SYMTAB_IS_DIE): Renumber.
4296 * xcoffout.c: Refer to former sdbout.c file.
4297 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4299 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4300 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4301 (Debugging Options): Delete -gcoff.
4302 (-gstabs): Delete SDB reference.
4304 (-gcoff@var{level}): Delete.
4305 * doc/passes.texi (Debugging information output): Delete SDB and
4306 sdbout.c references.
4307 * doc/tm.texi: Regenerate.
4308 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4309 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4311 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4312 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4313 (SDB_DEBUGGING_INFO): Delete.
4314 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4315 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4316 * target.def (output_source_filename): Delete COFF reference.
4318 * common.opt (gcoff): Delete.
4319 (gxcoff+): Update Negative chain.
4320 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4322 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4323 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4324 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4326 (expand_function_start): Change sdb reference to past tense.
4327 (expand_function_end): Change sdb reference to past tense.
4328 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4329 * opts.c (debug_type_names): Delete coff entry.
4330 (common_handle_option): Delete OPT_gcoff case.
4331 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4333 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4334 * config/cris/cris.h: Delete SDB reference in comment.
4335 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4336 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4337 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4338 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4340 (ix86_expand_prologue): Likewise.
4341 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4342 * config/ia64/ia64.h: Likewise.
4343 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4344 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4346 * config/mmix/mmix.h: Likewise.
4347 * config/nds32/nds32.c: Likewise.
4348 * config/stormy/storym16.h: Likewise.
4349 * config/visium/visium.h: Likewise.
4350 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4352 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4354 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4355 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4356 FRAME_POINTER_REGNUM point at high end of local var area.
4358 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4360 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4361 Move comment around. Do not reset best_edge for a copiable
4362 destination if the copy would cause a partition change.
4363 (better_edge_p): Remove redundant check.
4365 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4367 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4369 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4371 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4372 for math functions that have _Float<N> and _Float<N>X variants.
4373 (mathfn_built_in_2): Add support for math functions that have
4374 _Float<N> and _Float<N>X variants.
4375 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4376 (expand_builtin_mathfn_ternary): Add support for fma with
4377 _Float<N> and _Float<N>X variants.
4378 (expand_builtin): Likewise.
4379 (fold_builtin_3): Likewise.
4380 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4381 create math function _Float<N> and _Float<N>X variants as external
4383 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4384 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4385 the __builtin_ prefix and if not strict ansi, without the prefix.
4386 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4387 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4388 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4389 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4390 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4391 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4392 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4393 function signatures for fma _Float<N> and _Float<N>X variants.
4394 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4395 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4396 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4397 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4398 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4399 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4400 * gencfn-macros.c (print_case_cfn): Add support for math functions
4401 that have _Float<N> and _Float<N>X variants.
4402 (print_define_operator_list): Likewise.
4403 (fltfn_suffixes): Likewise.
4405 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4406 for math functions that have _Float<N> and _Float<N>X variants.
4407 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4408 and _Float<N>X variants.
4409 (COPYSIGN): Likewise.
4412 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4413 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4415 (integer_valued_read_call_p): Likewise.
4416 * fold-const-call.c (fold_const_call_ss): Likewise.
4417 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4418 _Float<N> and _Float<N>X variants.
4419 (fold_const_call_ssss): Add support for fma _Float<N> and
4420 _Float<N>X variants.
4421 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4422 support for copysign and fma _Float<N> and _Float<N>X variants.
4423 (backprop::process_builtin_call_use): Likewise.
4424 * tree-call-cdce.c (can_test_argument_range); Add support for
4425 sqrt _Float<N> and _Float<N>X variants.
4426 (edom_only_function): Likewise.
4427 (get_no_error_domain): Likewise.
4428 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4429 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4430 copysign _Float<N> and _Float<N>X variants.
4431 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4432 handled by machine independent code.
4433 (FMAF128): Likewise.
4434 * doc/cpp.texi (Common Predefined Macros): Document defining
4435 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4436 fma _Float<N> and _Float<N>X variants.
4438 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4441 * config/i386/i386-modes.def (CCFPU): Remove definition.
4442 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4443 (ix86_cc_modes_compatible): Ditto.
4444 (ix86_expand_carry_flag_compare): Ditto.
4445 (ix86_expand_int_movcc): Ditto.
4446 (ix86_expand_int_addcc): Ditto.
4447 (ix86_reverse_condition): Ditto.
4448 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4449 Return true/false for unordered/ordered fp comparisons.
4450 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4451 (ix86_prepare_fp_compare_args): Update for rename.
4452 (ix86_expand_fp_compare): Update for rename. Generate unordered
4453 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4454 (ix86_expand_sse_compare_and_jump): Ditto.
4455 * config/i386/predicates.md (fcmov_comparison_operator):
4456 Remove CCFPU mode handling.
4457 (ix86_comparison_operator): Ditto.
4458 (ix86_carry_flag_operator): Ditto.
4459 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
4460 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
4461 (*cmpu<mode>_cc_i387): Ditto.
4462 (FPCMP): Remove mode iterator.
4463 (unord): Remove mode attribute.
4464 (unord_subst): New define_subst transformation
4465 (unord): New define_subst attribute.
4467 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
4468 (*cmpi<unord>xf_i387): Ditto.
4469 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
4470 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
4471 using unord_subst transformation.
4472 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
4473 (round_saeonly): Also handle CCFP mode.
4474 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
4475 Remove UNSPEC_SAHF unspec handling.
4477 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4479 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
4481 2017-10-27 Jeff Law <law@redhat.com>
4483 * gimple-ssa-sprintf.c: Include domwalk.h.
4484 (class sprintf_dom_walker): New class, derived from dom_walker.
4485 (sprintf_dom_walker::before_dom_children): New function.
4486 (struct call_info): Moved into sprintf_dom_walker class
4487 (compute_formath_length, handle_gimple_call): Likewise.
4488 (sprintf_length::execute): Call the dominator walker rather
4489 than walking the statements.
4491 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4492 gimple statement locations.
4493 (check_array_bounds): Corresponding changes. Get the statement's
4494 location directly from wi->stmt.
4496 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4499 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4501 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4503 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4504 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4506 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4509 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4510 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4511 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4512 maybe_get_pool_constant.
4513 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4516 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4518 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4519 2.26 caveat. Update gas and gld versions.
4520 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4523 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4525 * cgraph.h (set_malloc_flag): Declare.
4526 * cgraph.c (set_malloc_flag_1): New function.
4527 (set_malloc_flag): Likewise.
4528 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4529 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4531 (read_ipa_call_summary): Add support for reading is_return_callee.
4532 (write_ipa_call_summary): Stream is_return_callee.
4533 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4534 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4535 ipa-prop.h, ipa-fnsummary.h.
4536 (pure_const_names): Change to static.
4537 (malloc_state_e): Define.
4538 (malloc_state_names): Define.
4539 (funct_state_d): Add field malloc_state.
4540 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4541 (check_retval_uses): New function.
4542 (malloc_candidate_p): Likewise.
4543 (analyze_function): Add support for malloc attribute.
4544 (pure_const_write_summary): Stream malloc_state.
4545 (pure_const_read_summary): Add support for reading malloc_state.
4546 (dump_malloc_lattice): New function.
4547 (propagate_malloc): New function.
4548 (warn_function_malloc): New function.
4549 (ipa_pure_const::execute): Call propagate_malloc and
4550 ipa_free_fn_summary.
4551 (pass_local_pure_const::execute): Add support for malloc attribute.
4552 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4553 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4555 2017-10-27 Martin Liska <mliska@suse.cz>
4557 PR gcov-profile/82457
4558 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4559 for fork-like functions to be properly instrumented.
4561 2017-10-27 Richard Biener <rguenther@suse.de>
4564 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4565 info when we redirected EH.
4567 2017-10-26 Michael Collison <michael.collison@arm.com>
4569 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4571 (<optab>_trunchf<GPI:mode>2: New pattern.
4572 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4573 * config/aarch64/iterators.md (wv): New mode attribute.
4574 (vf, VF): New mode attributes.
4575 (vgp, VGP): New mode attributes.
4576 (s): Update attribute with SImode and DImode prefixes.
4578 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4580 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4581 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4582 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4583 (nios2_option_overide): Initialize it. Don't allow R0-relative
4584 addressing with PIC.
4585 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4586 (nios2_symbolic_constant_p): Likewise.
4587 (nios2_legitimate_address_p): Likewise.
4588 (nios2_r0rel_section_name_p): New.
4589 (nios2_symbol_ref_in_r0rel_data_p): New.
4590 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4591 (r0rel_constant_p): New.
4592 (nios2_print_operand_address): Handle r0rel_constant_p.
4593 (nios2_cdx_narrow_form_p): Likewise.
4594 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4595 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4596 (Nios II Options): Document -mr0rel-sec.
4598 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4600 * config/nios2/nios2.c: Include xregex.h.
4601 (nios2_gprel_sec_regex): New.
4602 (nios2_option_overide): Initialize it. Don't allow GP-relative
4603 addressing with PIC.
4604 (nios2_small_section_name_p): Check for regex match.
4605 * config/nios2/nios2.opt (mgprel-sec=): New option.
4606 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4607 (Nios II Options): Document -mgprel-sec.
4609 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4611 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4613 2017-10-26 Tom de Vries <tom@codesourcery.com>
4615 PR tree-optimization/82707
4616 * gimple.c (gimple_copy): Fix unsharing of
4617 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4619 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4621 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4622 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4623 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4624 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4625 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4626 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4627 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4628 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4629 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4630 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4631 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4632 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4633 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4634 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4635 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4636 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4637 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4639 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4641 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4643 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4644 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4645 warning to rs6000.c. Remove the Undocumented flag, since it has
4647 (-mabi=ibmlongdouble): Likewise.
4648 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4649 already set, set the default format for long double.
4650 (rs6000_debug_reg_global): Print whether long double is IBM or
4652 (rs6000_option_override_internal): Rework setting long double
4653 format. Only warn if the user is changing the long double default
4654 and they did not use -Wno-psabi.
4655 * doc/invoke.texi (PowerPC options): Update the documentation for
4656 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4658 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4659 Alan Hayward <alan.hayward@arm.com>
4660 David Sherwood <david.sherwood@arm.com>
4662 * rtl.h (wider_subreg_mode): New function.
4663 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4664 rather than an unsigned int *.
4665 * ira-color.c (regno_max_ref_width): Replace with...
4666 (regno_max_ref_mode): ...this new variable.
4667 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4668 Use wider_subreg_mode.
4669 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4670 rather than an unsigned int *.
4671 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4672 (process_alt_operands): Likewise.
4673 (invariant_p): Likewise.
4674 * lra-spills.c (assign_mem_slot): Likewise.
4675 (add_pseudo_to_slot): Likewise.
4676 * lra.c (collect_non_operand_hard_regs): Likewise.
4677 (add_regs_to_insn_regno_info): Likewise.
4678 * reload1.c (regno_max_ref_width): Replace with...
4679 (regno_max_ref_mode): ...this new variable.
4680 (reload): Update accordingly. Update call to
4681 ira_sort_regnos_for_alter_reg.
4682 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4683 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4684 (scan_paradoxical_subregs): Likewise.
4686 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4688 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4689 (aarch64_frame): Add emit_frame_chain boolean.
4690 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4691 Move eh_return case to aarch64_layout_frame.
4692 (aarch64_layout_frame): Initialize emit_frame_chain.
4693 (aarch64_expand_prologue): Use emit_frame_chain.
4695 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4697 * config/aarch64/aarch64.c (aarch64_layout_frame):
4698 Ensure LR is always stored at the bottom of the callee-saves.
4699 Remove rarely used frame layout which saves callee-saves at top of
4700 frame, so the store of LR can be used as a valid probe in all cases.
4702 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4704 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4705 Improve unaligned TImode/TFmode base/offset split.
4707 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4708 Alan Hayward <alan.hayward@arm.com>
4709 David Sherwood <david.sherwood@arm.com>
4711 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4712 * combine.c (find_single_use_1): Likewise.
4713 (expand_field_assignment): Likewise.
4714 (move_deaths): Likewise.
4715 * lra-constraints.c (simplify_operand_subreg): Likewise.
4716 (curr_insn_transform): Likewise.
4717 * lra.c (collect_non_operand_hard_regs): Likewise.
4718 (add_regs_to_insn_regno_info): Likewise.
4719 * rtlanal.c (reg_referenced_p): Likewise.
4720 (covers_regno_no_parallel_p): Likewise.
4722 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4724 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4725 Don't print any bits outside the precision of the value.
4726 * wide-int.cc (test_printing): Add some new tests.
4728 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4730 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4731 supports -xbrace_comment option.
4732 * configure: Regenerate.
4733 * config.in: Regenerate.
4734 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4735 (ASM_CPU_SPEC): Use it.
4737 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4739 * target.def (static_rtx_alignment): New hook.
4740 * targhooks.h (default_static_rtx_alignment): Declare.
4741 * targhooks.c (default_static_rtx_alignment): New function.
4742 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4743 * doc/tm.texi: Regenerate.
4744 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4745 instead of targetm.constant_alignment. Remove call to
4747 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4748 (cris_preferred_mininum_alignment): New function, split out from...
4749 (cris_constant_alignment): ...here.
4750 (cris_static_rtx_alignment): New function.
4751 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4753 (ix86_constant_alignment): ...here.
4754 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4755 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4756 (mmix_static_rtx_alignment): New function.
4757 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4758 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4760 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4763 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4764 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4766 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4768 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4769 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4770 mark as a sign-extending load.
4771 (local_pic_load_u): Define.
4773 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4776 * fold-const.c (operand_equal_for_comparison_p): Also return true
4777 if ARG0 is a simple variant of ARG1 with narrower precision.
4778 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4780 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4782 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4784 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4785 scatter_static, scatter_per_elt.
4786 * x86-tune-costs.h: Add new cost entries.
4788 2017-10-25 Richard Biener <rguenther@suse.de>
4790 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4791 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4792 class pass_fre): Move to ...
4793 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4794 class pass_fre): ... here and adjust for statistics.
4796 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4799 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4800 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4801 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4802 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4805 2017-10-25 Richard Biener <rguenther@suse.de>
4807 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4808 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4809 eliminate_push_avail, eliminate_insert): Move inside...
4810 (class eliminate_dom_walker): ... this class in preparation
4812 (fini_eliminate): Remove by merging with ...
4813 (eliminate): ... this function. Adjust for class changes.
4814 (pass_pre::execute): Remove fini_eliminate call.
4815 (pass_fre::execute): Likewise.
4817 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4820 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4821 (VPERMI2, VPERMI2I): New mode iterators.
4822 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4823 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4825 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4826 mode iterator. Remove 3 old define_insn patterns.
4827 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4828 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4829 VPERMI2 mode iterator, remove the other two expanders.
4830 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4831 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4832 instructions, remove the other two patterns.
4833 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4834 mode iterator, remove the other two patterns.
4835 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4836 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4837 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4838 and adjust argument order accordingly.
4839 (ix86_expand_vec_perm): Adjust caller.
4840 (expand_vec_perm_1): Likewise.
4841 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4842 (expand_vec_perm_vpermt2_vpshub2): ... this.
4843 (ix86_expand_vec_perm_const_1): Adjust caller.
4844 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4847 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4848 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4849 (vec_shl_<mode>): Remove unused expander.
4850 (avx512bw_<shift_insn><mode>3): New define_insn.
4851 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4852 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4854 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4857 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4860 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4862 PR rtl-optimization/82396
4863 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4864 (autopref_multipass_init): Simplify initialization.
4865 (autopref_rank_data): Simplify sort order.
4866 * gcc/sched-int.h (autopref_multipass_data_): Remove
4867 multi_mem_insn_p, min_offset and max_offset.
4869 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4872 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4873 Check special value of flag_omit_frame_pointer.
4874 (aarch64_can_eliminate): Likewise.
4875 (aarch64_override_options_after_change_1): Simplify handling of
4876 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4878 2017-10-24 Richard Biener <rguenther@suse.de>
4880 PR tree-optimization/82697
4881 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4882 zero for conditional load and unconditional store.
4884 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4886 * doc/install.texi: Document bootstrap-cet.
4888 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4891 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4892 ENDBR instruction at function entrance if function is only
4895 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4898 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4899 patterns to better describe from which operation the CF is computed.
4900 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4901 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4902 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4903 is 0, use _0 suffixed expanders instead of emitting a comparison
4906 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4908 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4909 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4911 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4914 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4915 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4916 * loop-iv.c (iv_get_reaching_def): Likewise.
4917 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4918 variable is promoted and the partition contains undefined values.
4920 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4922 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4924 (nios2_address_cost): Define.
4925 (nios2_legitimize_address): Recognize (exp + constant) directly.
4926 (TARGET_ADDRESS_COST): Define.
4928 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4930 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4931 (nios2_symbolic_memory_operand_p): Declare.
4932 (nios2_split_large_constant): Declare.
4933 (nios2_split_symbolic_memory_operand): Declare.
4934 * config/nios2/nios2.c: Adjust includes.
4935 (nios2_symbolic_constant_allowed): New.
4936 (nios2_symbolic_constant_p): New.
4937 (nios2_plus_symbolic_constant_p): New.
4938 (nios2_valid_addr_expr_p): Recognize addresses involving
4940 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4941 (nios2_symbolic_memory_operand_p): New.
4942 (nios2_large_constant_p): New.
4943 (nios2_split_large_constant): New.
4944 (nios2_split_plus_large_constant): New.
4945 (nios2_split_symbolic_memory_operand): New.
4946 (nios2_legitimize_address): Code refactoring. Handle addresses
4947 involving symbolic constants.
4948 (nios2_emit_move_sequence): Likewise.
4949 (nios2_print_operand): Improve error output.
4950 (nios2_print_operand_address): Handle LO_SUM.
4951 (nios2_cdx_narrow_form_p): Likewise.
4952 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4953 operands involving symbolic constants.
4954 (movhi_internal, movsi_internal): Likewise.
4955 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4956 (extendhisi2, extendqi<mode>2): Likewise.
4958 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4960 * tree-pass.h (PROP_rtl_split_insns): Define.
4961 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4963 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4965 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4967 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4970 * target.def (const_not_ok_for_debug_p): Default to
4971 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4972 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4973 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4974 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4975 which targetm.const_not_ok_for_debug_p returned true.
4976 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4978 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4980 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4981 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4982 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4984 (ix86_const_not_ok_for_debug_p): New function.
4985 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4986 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4988 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4991 * system.h: Conditionally include "unique-ptr.h" if
4992 INCLUDE_UNIQUE_PTR is defined.
4993 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4994 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4996 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4998 * config/rl78/rl78.md: New define_expand "subdi3".
5000 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5003 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5004 DF_REF_INSN if DF_REF_INSN_INFO is false.
5006 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5008 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5009 xmm_move instead of sse_move.
5010 (sse_store_index): New function.
5011 (ix86_register_move_cost): Be more sensible about mismatch stall;
5012 model AVX moves correctly; make difference between sse->integer and
5014 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5015 moves; make difference between SSE and AVX.
5016 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5017 and zmm_move. Increase size of sse load and store tables;
5018 add unaligned load and store tables; add ssemmx_to_integer.
5019 * x86-tune-costs.h: Update all entries according to real
5020 move latencies from Agner Fog's manual and chip documentation.
5022 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5025 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5026 * config/i386/constraints.md (Wf): New constraint.
5027 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5028 (cmp<dwi>_doubleword): Removed.
5029 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5030 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5031 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5032 expand with cmp<dwi>_doubleword. For LTU and GEU use
5033 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5035 * common.opt (gcolumn-info): Enable by default.
5036 * doc/invoke.texi (gcolumn-info): Document new default.
5038 2017-10-23 Richard Biener <rguenther@suse.de>
5040 PR tree-optimization/82672
5041 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5042 Fold the stmt if we propagated into it.
5044 2017-10-23 Richard Biener <rguenther@suse.de>
5046 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5047 (bitmap_remove_expr_from_set): ... this. All callers call this
5048 for non-constant values.
5049 (bitmap_set_subtract): Rename to...
5050 (bitmap_set_subtract_expressions): ... this. Adjust and
5052 (bitmap_set_contains_value): Remove superfluous check.
5053 (bitmap_set_replace_value): Inline into single caller ...
5054 (bitmap_value_replace_in_set): ... here and simplify.
5055 (dependent_clean): Merge into ...
5056 (clean): ... this using an overload. Adjust.
5057 (prune_clobbered_mems): Adjust.
5058 (compute_antic_aux): Likewise.
5059 (compute_partial_antic_aux): Likewise.
5061 2017-10-23 Richard Biener <rguenther@suse.de>
5063 PR tree-optimization/82129
5065 2017-08-01 Richard Biener <rguenther@suse.de>
5067 PR tree-optimization/81181
5068 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5069 (compute_antic): ... end of iteration here.
5071 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5073 * target.def (starting_frame_offset): New hook.
5074 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5075 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5076 * doc/tm.texi.in: Regenerate.
5077 * hooks.h (hook_hwi_void_0): Declare.
5078 * hooks.c (hook_hwi_void_0): New function.
5079 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5080 STARTING_FRAME_OFFSET.
5081 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5082 * reload1.c (reload): Likewise.
5083 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5084 instead of STARTING_FRAME_OFFSET.
5085 * function.c (try_fit_stack_local): Likewise.
5086 (assign_stack_local_1): Likewise
5087 (instantiate_virtual_regs): Likewise.
5088 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5089 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5090 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5091 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5092 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5093 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5094 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5095 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5096 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5097 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5098 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5099 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5100 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5101 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5102 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5103 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5104 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5105 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5106 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5107 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5108 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5109 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5110 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5111 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5112 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5113 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5114 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5115 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5116 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5117 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5118 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5119 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5120 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5121 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5122 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5123 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5124 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5125 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5126 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5127 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5128 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5129 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5130 return a HOST_WIDE_INT.
5131 (avr_builtin_setjmp_frame_value): Use it instead of
5132 STARTING_FRAME_OFFSET.
5133 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5134 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5135 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5137 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5138 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5139 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5140 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5141 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5142 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5143 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5144 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5145 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5146 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5147 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5148 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5150 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5151 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5152 * config/mips/mips.c (mips_compute_frame_info): Refer to
5153 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5154 (mips_starting_frame_offset): New function.
5155 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5156 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5157 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5158 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5159 and return a HOST_WIDE_INT.
5160 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5161 (mmix_initial_elimination_offset): Refer to
5162 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5163 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5164 * config/pa/pa.c (pa_starting_frame_offset): New function.
5165 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5166 (pa_expand_prologue): Likewise.
5167 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5168 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5169 !FRAME_GROWS_DOWNWARD handling to...
5170 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5171 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5172 !FRAME_GROWS_DOWNWARD handling to...
5173 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5174 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5175 !FRAME_GROWS_DOWNWARD handling to...
5176 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5177 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5179 (rs6000_starting_frame_offset): New function.
5180 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5181 !FRAME_GROWS_DOWNWARD handling to...
5182 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5183 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5184 !FRAME_GROWS_DOWNWARD handling to...
5185 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5186 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5187 !FRAME_GROWS_DOWNWARD handling to...
5188 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5189 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5190 (rs6000_starting_frame_offset): New function.
5191 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5192 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5193 * config/vax/vax.c (vax_starting_frame_offset): New function.
5194 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5195 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5196 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5197 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5198 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5199 * system.h (STARTING_FRAME_OFFSET): Poison.
5201 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5203 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5204 SCALAR_TYPE_MODE instead of TYPE_MODE.
5206 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5207 Alan Hayward <alan.hayward@arm.com>
5208 David Sherwood <david.sherwood@arm.com>
5210 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5212 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5213 Alan Hayward <alan.hayward@arm.com>
5214 David Sherwood <david.sherwood@arm.com>
5216 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5218 2017-10-23 Richard Biener <rguenther@suse.de>
5220 PR tree-optimization/82129
5221 * tree-ssa-pre.c (bitmap_set_and): Remove.
5222 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5223 canonicalizing expressions in the set to those with lowest
5224 ID rather than taking that from the first edge.
5226 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5228 * combine.c (rtx_equal_for_field_assignment_p): Use
5229 byte_lowpart_offset.
5231 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5236 to a SUBREG_PROMOTED_VAR.
5238 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5239 Alan Hayward <alan.hayward@arm.com>
5240 David Sherwood <david.sherwood@arm.com>
5242 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5243 (expand_debug_source_expr): Likewise.
5244 * combine.c (combine_simplify_rtx): Likewise.
5245 * cse.c (fold_rtx): Likewise.
5246 * optabs.c (expand_float): Likewise.
5247 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5248 (simplify_binary_operation_1): Likewise.
5250 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5251 Alan Hayward <alan.hayward@arm.com>
5252 David Sherwood <david.sherwood@arm.com>
5254 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5255 (record_promoted_value): Likewise.
5256 * expr.c (expand_expr_real_2): Likewise.
5257 * ree.c (update_reg_equal_equiv_notes): Likewise.
5258 (combine_set_extension): Likewise.
5259 * rtlanal.c (low_bitmask_len): Likewise.
5260 * simplify-rtx.c (neg_const_int): Likewise.
5261 (simplify_binary_operation_1): Likewise.
5263 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5264 Alan Hayward <alan.hayward@arm.com>
5265 David Sherwood <david.sherwood@arm.com>
5267 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5268 * regcprop.c (maybe_mode_change): Likewise.
5269 * reload1.c (alter_reg): Likewise.
5271 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5273 * inchash.h (inchash::hash::add_wide_int): New function.
5274 * lto-streamer-out.c (hash_tree): Use it.
5276 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5278 * inchash.h (inchash::hash::add_wide_int): Rename to...
5279 (inchash::hash::add_hwi): ...this.
5280 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5281 (polymorphic_call_target_hasher::hash): Likewise.
5282 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5283 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5284 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5285 * lto-streamer-out.c (hash_tree): Likewise.
5286 * optc-save-gen.awk: Likewise.
5287 * tree.c (add_expr): Likewise.
5289 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5292 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5293 for ordered inequality comparisons even with TARGET_IEEE_FP.
5295 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5298 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5299 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5300 Expand with cmp<dwi>_doubleword.
5302 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5304 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5305 List CET intrinsics.
5306 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5307 specific to -fcf-protection option.
5309 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5311 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5312 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5313 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5314 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5315 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5316 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5317 (extra_objs): Add cet.o for Linux/x86 targets.
5318 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5319 * config/i386/cet.c: New file.
5320 * config/i386/cetintrin.h: Likewise.
5321 * config/i386/t-cet: Likewise.
5322 * config/i386/cpuid.h (bit_SHSTK): New.
5323 (bit_IBT): Likewise.
5324 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5325 pass IBT and SHSTK bits.
5326 * config/i386/i386-builtin-types.def
5327 (VOID_FTYPE_UNSIGNED_PVOID): New.
5328 (VOID_FTYPE_UINT64_PVOID): Likewise.
5329 * config/i386/i386-builtin.def: Add CET intrinsics.
5330 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5331 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5332 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5333 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5335 * config/i386/i386.c (rest_of_insert_endbranch): New.
5336 (pass_data_insert_endbranch): Likewise.
5337 (pass_insert_endbranch): Likewise.
5338 (make_pass_insert_endbranch): Likewise.
5339 (ix86_notrack_prefixed_insn_p): Likewise.
5340 (ix86_target_string): Add -mibt, -mshstk flags.
5341 (ix86_option_override_internal): Add flag_cf_protection
5343 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5344 (ix86_print_operand): Add 'notrack' prefix output.
5345 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5346 (ix86_expand_builtin): Expand CET intrinsics.
5347 (x86_output_mi_thunk): Add 'endbranch' instruction.
5348 * config/i386/i386.h (TARGET_IBT): New.
5349 (TARGET_IBT_P): Likewise.
5350 (TARGET_SHSTK): Likewise.
5351 (TARGET_SHSTK_P): Likewise.
5352 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5353 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5354 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5355 (builtin_setjmp_setup): New pattern.
5356 (builtin_longjmp): Likewise.
5357 (rdssp<mode>): Likewise.
5358 (incssp<mode>): Likewise.
5359 (saveprevssp): Likewise.
5360 (rstorssp): Likewise.
5361 (wrss<mode>): Likewise.
5362 (wruss<mode>): Likewise.
5363 (setssbsy): Likewise.
5364 (clrssbsy): Likewise.
5365 (nop_endbr): Likewise.
5366 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5368 * config/i386/immintrin.h: Include <cetintrin.h>.
5369 * config/i386/linux-common.h
5370 (file_end_indicate_exec_stack_and_cet): New prototype.
5371 (TARGET_ASM_FILE_END): New.
5373 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5375 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5376 latencies instead of having separate table; make difference between
5377 integer and float costs.
5378 * i386.h (processor_costs): Remove scalar_stmt_cost,
5379 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5380 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5382 * x86-tune-costs.h: Remove entries which has been removed in
5383 procesor_costs from all tables; make cond_taken_branch_cost
5384 and cond_not_taken_branch_cost COST_N_INSNS based.
5386 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5388 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5390 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5392 * config/i386/i386.md (isa): Remove fma_avx512f.
5393 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5394 <avx512>_fmadd_<mode>_mask3<round_name>,
5395 <avx512>_fmsub_<mode>_mask<round_name>,
5396 <avx512>_fmsub_<mode>_mask3<round_name>,
5397 <avx512>_fnmadd_<mode>_mask<round_name>,
5398 <avx512>_fnmadd_<mode>_mask3<round_name>,
5399 <avx512>_fnmsub_<mode>_mask<round_name>,
5400 <avx512>_fnmsub_<mode>_mask3<round_name>,
5401 <avx512>_fmaddsub_<mode>_mask<round_name>,
5402 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5403 <avx512>_fmsubadd_<mode>_mask<round_name>,
5404 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5405 (*vec_widen_umult_even_v16si<mask_name>,
5406 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5407 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5409 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5411 * extend.texi: Add 'nocf_check' documentation.
5412 * gimple.texi: Add second parameter to
5413 gimple_build_call_from_tree.
5414 * invoke.texi: Add -fcf-protection documentation.
5415 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5417 2017-10-20 Richard Biener <rguenther@suse.de>
5419 PR tree-optimization/82473
5420 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5421 the largest input type.
5423 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5425 * c-attribs.c (handle_nocf_check_attribute): New function.
5426 (c_common_attribute_table): Add 'nocf_check' handling.
5427 * gimple-parser.c: Add second argument NULL to
5428 gimple_build_call_from_tree.
5429 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5430 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5432 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5433 * common.opt: Add fcf-protection flag.
5434 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5435 * flag-types.h: Add enum cf_protection_level.
5436 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5437 Add 'nocf_check' attribute propagation to gimple call.
5438 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5439 (gimple_build_call_from_tree): Update prototype.
5440 (gimple_call_nocf_check_p): New function.
5441 (gimple_call_set_nocf_check): Likewise.
5442 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5443 * ipa-icf.c: Add nocf_check attribute in statement hash.
5444 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5445 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5446 * toplev.c (process_options): Add flag_cf_protection handling.
5448 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5450 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5452 2017-10-20 Richard Biener <rguenther@suse.de>
5454 PR tree-optimization/82603
5455 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5456 remove false predicated stores.
5458 2017-10-20 Richard Biener <rguenther@suse.de>
5460 * graphite-isl-ast-to-gimple.c
5461 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
5462 Remove return value and simplify, dump copied stmt after lhs
5464 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
5465 Reduce dump verbosity.
5466 (gsi_insert_earliest): Likewise.
5467 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
5468 * graphite.c (print_global_statistics): Adjust dumping.
5469 (print_graphite_scop_statistics): Likewise.
5470 (print_graphite_statistics): Do not dump loops here.
5471 (graphite_transform_loops): But here.
5473 2017-10-20 Nicolas Roche <roche@adacore.com>
5475 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
5476 * configure: Regenerate.
5478 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5481 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
5482 functions when optimizing replace GIMPLE_RETURN stmts with
5483 calls to __builtin_unreachable ().
5486 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5487 for -fsanitize=thread link of executables.
5488 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
5489 link of executables.
5492 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5494 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5495 define_insns for logical vector shifts to use VI248_AVX512BW
5496 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5497 condition, useless isa and prefix attributes. Change the first
5498 2 of these define_insns to ...
5499 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5500 define_insn for avx512vl.
5501 (<shift_insn><mode>3): ... and this, new define_insn without
5502 masking for non-avx512vl.
5505 * config/i386/sse.md (*andnot<mode>3,
5506 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5507 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5508 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5509 not applied use empty suffix even for TARGET_AVX512VL.
5510 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5511 is applied, supply evex,evex or evex,evex,evex instead of just
5514 2017-10-20 Julia Koval <julia.koval@intel.com>
5516 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5517 (OPTION_MASK_ISA_GFNI_UNSET): New.
5518 (ix86_handle_option): Handle OPT_mgfni.
5519 * config/i386/cpuid.h (bit_GFNI): New.
5520 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5521 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5522 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5523 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5524 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5525 * config/i386/i386.opt: Add mgfni.
5527 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5529 * config/msp430/msp430.c (msp430_option_override): Disable
5530 -fdelete-null-pointer-checks.
5531 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5533 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5535 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5536 of x87 and SSE instructions.
5538 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5540 * asan.c (create_cond_insert_point): Do not update edge count.
5541 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5542 (afdo_propagate_circuit): Likewise.
5543 (afdo_calculate_branch_prob): Likewise.
5544 (afdo_annotate_cfg): Likewise.
5545 * basic-block.h (struct edge_def): Remove count.
5546 (edge_def::count): New accessor.
5547 * bb-reorder.c (rotate_loop): Update.
5548 (find_traces_1_round): Update.
5549 (connect_traces): Update.
5550 (sanitize_hot_paths): Update.
5551 * cfg.c (unchecked_make_edge): Update.
5552 (make_single_succ_edge): Update.
5553 (check_bb_profile): Update.
5554 (dump_edge_info): Update.
5555 (update_bb_profile_for_threading): Update.
5556 (scale_bbs_frequencies_int): Update.
5557 (scale_bbs_frequencies_gcov_type): Update.
5558 (scale_bbs_frequencies_profile_count): Update.
5559 (scale_bbs_frequencies): Update.
5560 * cfganal.c (connect_infinite_loops_to_exit): Update.
5561 * cfgbuild.c (compute_outgoing_frequencies): Update.
5562 (find_many_sub_basic_blocks): Update.
5563 * cfgcleanup.c (try_forward_edges): Update.
5564 (try_crossjump_to_edge): Update
5565 * cfgexpand.c (expand_gimple_cond): Update
5566 (expand_gimple_tailcall): Update
5567 (construct_exit_block): Update
5568 * cfghooks.c (verify_flow_info): Update
5569 (redirect_edge_succ_nodup): Update
5570 (split_edge): Update
5571 (make_forwarder_block): Update
5572 (duplicate_block): Update
5573 (account_profile_record): Update
5574 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5575 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5576 * cfgloopmanip.c (scale_loop_profile): Update.
5578 (lv_adjust_loop_entry_edge): Update.
5579 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5580 (force_nonfallthru_and_redirect): Update.
5581 (purge_dead_edges): Update.
5582 (rtl_flow_call_edges_add): Update.
5583 * cgraphunit.c (init_lowered_empty_function): Update.
5584 (cgraph_node::expand_thunk): Update.
5585 * gimple-pretty-print.c (dump_probability): Update.
5586 (dump_edge_probability): Update.
5587 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5588 * haifa-sched.c (sched_create_recovery_edges): Update.
5589 * hsa-gen.c (convert_switch_statements): Update.
5590 * ifcvt.c (dead_or_predicable): Update.
5591 * ipa-inline-transform.c (inline_transform): Update.
5592 * ipa-split.c (split_function): Update.
5593 * ipa-utils.c (ipa_merge_profiles): Update.
5594 * loop-doloop.c (add_test): Update.
5595 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5596 * lto-streamer-in.c (input_cfg): Update.
5597 (input_function): Update.
5598 * lto-streamer-out.c (output_cfg): Update.
5599 * modulo-sched.c (sms_schedule): Update.
5600 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5601 * predict.c (maybe_hot_edge_p): Update.
5602 (unlikely_executed_edge_p): Update.
5603 (probably_never_executed_edge_p): Update.
5604 (dump_prediction): Update.
5605 (drop_profile): Update.
5606 (propagate_unlikely_bbs_forward): Update.
5607 (determine_unlikely_bbs): Update.
5608 (force_edge_cold): Update.
5609 * profile.c (compute_branch_probabilities): Update.
5610 * reg-stack.c (better_edge): Update.
5611 * shrink-wrap.c (handle_simple_exit): Update.
5612 * tracer.c (better_p): Update.
5613 * trans-mem.c (expand_transaction): Update.
5614 (split_bb_make_tm_edge): Update.
5615 * tree-call-cdce.c: Update.
5616 * tree-cfg.c (gimple_find_sub_bbs): Update.
5617 (gimple_split_edge): Update.
5618 (gimple_duplicate_sese_region): Update.
5619 (gimple_duplicate_sese_tail): Update.
5620 (gimple_flow_call_edges_add): Update.
5621 (insert_cond_bb): Update.
5622 (execute_fixup_cfg): Update.
5623 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5624 * tree-complex.c (expand_complex_div_wide): Update.
5625 * tree-eh.c (lower_resx): Update.
5626 (unsplit_eh): Update.
5627 (cleanup_empty_eh_move_lp): Update.
5628 * tree-inline.c (copy_edges_for_bb): Update.
5629 (freqs_to_counts): Update.
5630 (copy_cfg_body): Update.
5631 * tree-ssa-dce.c (remove_dead_stmt): Update.
5632 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5633 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5634 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5635 (unloop_loops): Update.
5636 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5637 * tree-ssa-loop-split.c (connect_loops): Update.
5638 (split_loop): Update.
5639 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5640 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5641 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5642 * tree-ssa-reassoc.c (branch_fixup): Update.
5643 * tree-ssa-tail-merge.c (replace_block_by): Update.
5644 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5645 (compute_path_counts): Update.
5646 (update_profile): Update.
5647 (recompute_probabilities): Update.
5648 (update_joiner_offpath_counts): Update.
5649 (estimated_freqs_path): Update.
5650 (freqs_to_counts_path): Update.
5651 (clear_counts_path): Update.
5652 (ssa_fix_duplicate_block_edges): Update.
5653 (duplicate_thread_path): Update.
5654 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5655 (case_bit_test_cmp): Update.
5656 (collect_switch_conv_info): Update.
5657 (gen_inbound_check): Update.
5658 (do_jump_if_equal): Update.
5659 (emit_cmp_and_jump_insns): Update.
5660 * tree-tailcall.c (decrease_profile): Update.
5661 (eliminate_tail_call): Update.
5662 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5663 (vect_do_peeling): Update.
5664 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5665 * ubsan.c (ubsan_expand_null_ifn): Update.
5666 (ubsan_expand_ptr_ifn): Update.
5667 * value-prof.c (gimple_divmod_fixed_value): Update.
5668 (gimple_mod_pow2): Update.
5669 (gimple_mod_subtract): Update.
5670 (gimple_ic): Update.
5671 (gimple_stringop_fixed_value): Update.
5673 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5676 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5678 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5680 PR rtl-optimization/82395
5681 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5682 based on non_spilled_static_chain_regno_p.
5684 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5686 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5687 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5688 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5690 2017-10-19 Martin Sebor <msebor@redhat.com>
5692 PR tree-optimization/82596
5693 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5695 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5697 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5698 (get_mem_refs_of_builtin_call): Likewise.
5699 * builtins.c (expand_builtin_apply): Adjust call to
5700 allocate_dynamic_stack_space.
5701 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5702 the third argument to allocate_dynamic_stack_space, otherwise -1.
5703 (expand_builtin): Deal with all alloca variants.
5704 (is_inexpensive_builtin): Likewise.
5705 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5706 * calls.c (special_function_p): Deal with all alloca variants.
5707 (initialize_argument_information): Adjust call to
5708 allocate_dynamic_stack_space.
5709 (expand_call): Likewise.
5710 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5711 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5712 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5713 use it for the stack usage computation.
5714 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5715 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5716 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5717 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5718 (in_loop_p): Remove first argument and useless check.
5719 (pass_walloca::execute): Remove useless test and adjust call to above.
5720 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5721 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5722 (gimplify_call_expr): Deal with all alloca variants.
5723 * hsa-gen.c (gen_hsa_alloca): Likewise.
5724 (gen_hsa_insns_for_call): Likewise.
5725 * ipa-pure-const.c (special_builtin_state): Likewise.
5726 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5727 * tree-object-size.c (alloc_object_size): Likewise.
5728 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5729 (call_may_clobber_ref_p_1): Likewise.
5730 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5731 (ccp_fold_stmt): Likewise.
5732 (optimize_stack_restore): Likewise.
5733 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5734 (mark_all_reaching_defs_necessary_1): Likewise.
5735 (propagate_necessity): Likewise.
5736 (eliminate_unnecessary_stmts): Likewise.
5737 * tree.c (build_common_builtin_nodes): Build
5738 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5739 (build_alloca_call_expr): New function.
5740 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5741 (CASE_BUILT_IN_ALLOCA): Likewise.
5742 (build_alloca_call_expr): Declare.
5743 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5745 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5748 * dwarf2out.c (new_die_raw): New static inline function.
5749 (new_die): Use it to create the DIE.
5750 (add_AT_external_die_ref): Likewise.
5751 (clone_die): Likewise.
5752 (clone_as_declaration): Likewise.
5753 (dwarf2out_vms_debug_main_pointer): Likewise.
5754 (base_type_die): Likewise. Remove early return for corner cases.
5755 Do not call add_pubtype on the DIE here.
5756 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5757 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5758 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5759 native order exists for base types, attach the DIE manually and call
5760 add_pubtype on it. Do not equate a reverse order DIE to the type.
5762 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5764 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5765 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5767 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5769 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5771 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5772 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5773 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5774 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5775 * opts.c (sanitizer_opts): Add builtin.
5776 * ubsan.c (instrument_builtin): New function.
5777 (pass_ubsan::execute): Call it.
5778 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5779 * doc/invoke.texi: Document -fsanitize=builtin.
5781 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5782 builtins, store max (log2 (align), 0) into uchar field instead of
5783 align into uptr field.
5784 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5785 store uchar 0 field instead of uptr 0 field.
5786 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5787 instead of passing one address of struct with 2 locations pass
5788 two addresses of structs with 1 location each.
5789 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5790 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5791 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5792 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5793 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5794 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5795 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5796 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5798 2017-10-19 Martin Liska <mliska@suse.cz>
5801 * file-find.c (remove_prefix): Remove.
5802 * file-find.h (remove_prefix): Likewise.
5803 * gcc-ar.c: Remove smartness of lookup.
5805 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5807 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5808 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5809 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5811 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5814 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5815 (setcc + and to xor + setcc): New peephole2.
5817 2017-10-19 Tom de Vries <tom@codesourcery.com>
5819 * doc/sourcebuild.texi (Test Directives, Variants of
5820 dg-require-support): Add dg-require-stack-size.
5822 2017-10-19 Martin Liska <mliska@suse.cz>
5825 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5826 that have a large alignment.
5827 (gimplify_target_expr): Likewise.
5829 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5831 PR rtl-optimization/82602
5832 * ira.c (rtx_moveable_p): Return false for volatile asm.
5834 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5837 * config/i386/i386-modes.def (CCGZ): New CC mode.
5838 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5839 * config/i386/predicates.md (ix86_comparison_operator):
5841 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5842 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5843 with double-word subtraction.
5844 (put_condition_code): Handle CCGZmode.
5846 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5848 * wide-int.cc (debug (const wide_int &)): New.
5849 (debug (const wide_int *)): New.
5850 (debug (const widest_int &)): New.
5851 (debug (const widest_int *)): New.
5853 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5856 * lra-constraints.c (curr_insn_transform): Use non-input operand
5857 instead of output one for matched reload.
5859 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5861 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5862 (tree-ssa-loop-ivopts.h): New header file.
5863 (struct builtin_info): New fields.
5864 (classify_builtin_1): Compute and record base and offset parts for
5865 memset builtin partition by calling strip_offset.
5866 (offset_cmp, fuse_memset_builtins): New functions.
5867 (finalize_partitions): Fuse adjacent memset partitions by calling
5869 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5870 Expose the interface.
5871 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5873 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5875 PR tree-optimization/82574
5876 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5877 that data reference must be executed exactly once per iteration
5878 against the outermost loop in nest.
5879 (classify_partition): Update call to above function.
5881 2017-10-18 Richard Biener <rguenther@suse.de>
5883 PR tree-optimization/82591
5884 * graphite.c (graphite_transform_loops): Move code gen message
5886 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5887 Here. Handle scop_to_isl_ast failing.
5888 (scop_to_isl_ast): Limit the number of ISL operations.
5890 2017-10-18 Richard Biener <rguenther@suse.de>
5892 * graphite-isl-ast-to-gimple.c
5893 (translate_isl_ast_to_gimple::set_rename): Simplify.
5894 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5895 (graphite_copy_stmts_from_block): ... here.
5896 (copy_bb_and_scalar_dependences): Simplify.
5897 (add_parameters_to_ivs_params): Canonicalize.
5898 (generate_entry_out_of_ssa_copies): Simplify.
5899 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5900 by passing in ISL dimension.
5901 (parameter_index_in_region_1): Rename to ...
5902 (parameter_index_in_region): ... this.
5903 (extract_affine): Adjust assert, pass down parameter index.
5904 (add_param_constraints): Use range-info when available.
5905 (build_scop_context): Adjust.
5906 * sese.c (new_sese_info): Adjust.
5907 (free_sese_info): Likewise.
5908 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5909 Remove unused typedefs.
5910 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5912 2017-10-18 Martin Liska <mliska@suse.cz>
5914 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5916 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5918 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5919 (s390_sched_init): Do not reset s390_sched_state if we entered the
5920 current basic block via a fallthru edge and all others are unlikely.
5922 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5924 * config/s390/s390.c (NUM_SIDES): New variable.
5925 (LONGRUNNING_THRESHOLD): New variable.
5926 (LATENCY_FACTOR): New variable.
5927 (s390_sched_score): Decrease score for long-running instructions on
5929 (s390_sched_variable_issue): Perform bookkeeping for long-running
5932 2017-10-18 Richard Biener <rguenther@suse.de>
5934 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5935 Simplify with removal of the parameter rename map.
5936 (set_rename): Likewise.
5937 (should_copy_to_new_region): Likewise.
5938 (graphite_copy_stmts_from_block): Likewise.
5939 (copy_bb_and_scalar_dependences): Remove initialization of
5940 unused copied_bb_map.
5942 (copy_internal_parameters): Likewise.
5943 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5944 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5945 Use INTEGRAL_TYPE_P.
5946 (parameter_index_in_region_1): Rename to ...
5947 (assign_parameter_index_in_region): ... this. Assert we have
5948 a parameter we handle.
5949 (scan_tree_for_params): Adjust.
5950 * sese.h (parameter_rename_map_t): Remove.
5951 (struct sese_info_t): Remove unused parameter_rename_map and
5952 copied_bb_map members.
5953 * sese.c (new_sese_info): Adjust.
5954 (free_sese_info): Likewise.
5956 2017-10-18 Martin Liska <mliska@suse.cz>
5959 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5960 on an abnormal edge.
5962 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5964 * doc/invoke.texi (ffunction-sections and fdata-sections):
5967 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5969 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5970 the use statement can throw internally.
5972 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5974 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5975 any RTX present on the RHS of a SET.
5976 * compare-elim.c (try_eliminate_compare): Restore comment.
5978 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5980 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5981 may be also called on tcc_type nodes.
5982 * langhooks.c (lhd_tree_size): Likewise.
5984 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5986 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5987 format_warning_at_substring.
5988 (maybe_warn): Convert source_range * param to a location_t. Pass
5989 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5990 (format_directive): Remove code to extract source_ranges and
5991 source_range * in favor of just a location_t.
5992 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5994 * substring-locations.c (format_warning_va): Convert
5995 source_range * param to a location_t.
5996 (format_warning_at_substring): Likewise.
5997 * substring-locations.h (format_warning_va): Likewise.
5998 (format_warning_at_substring): Likewise.
6000 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6002 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6004 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6005 and scatter/gather ops.
6007 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6008 vec_gather_load and vec_scatter_store.
6009 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6010 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6011 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6012 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6013 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6014 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6016 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6018 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6019 Detect FTST instruction and handle its register pops. Only pop
6020 second operand if can_pop_second_op is true.
6021 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6022 set can_pop_second_op to false in the compare_for_stack_reg call.
6024 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6025 output_fp_compare for stack register operands.
6026 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6027 instructions here. Do not emit stack register pops here. Assert
6028 that FCOMPP pops next to top stack register. Rewrite function.
6030 2017-10-17 Nathan Sidwell <nathan@acm.org>
6033 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6034 use DECL_ASSEMBLER_NAME_RAW.
6037 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6040 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6041 Wilco Dijkstra <wilco.dijkstra@arm.com>
6043 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6044 converstion to Pmode from the buf_addr.
6046 2017-10-17 Richard Biener <rguenther@suse.de>
6048 * graphite-dependences.c (scop_get_reads_and_writes): Change
6049 output parameters to references.
6051 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6053 PR 71026/tree-optimization
6054 * fold-const.c (distribute_real_division): Removed.
6055 (fold_binary_loc): Remove calls to distribute_real_divison.
6057 2017-10-17 Richard Biener <rguenther@suse.de>
6059 * graphite-scop-detection.c
6060 (scop_detection::stmt_has_simple_data_refs_p): Always use
6061 the full nest as region.
6062 (try_generate_gimple_bb): Likewise.
6063 * sese.c (scalar_evolution_in_region): Simplify now that
6064 SCEV can handle instantiation in regions.
6065 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6066 in the non-loop part of a function if requested.
6068 2017-10-17 Richard Biener <rguenther@suse.de>
6070 PR tree-optimization/82563
6071 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6073 (graphite_regenerate_ast_isl): Call it.
6074 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6076 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6078 PR tree-optimization/82549
6079 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6080 Formatting fixes. Instead of calling make_bit_field_ref with negative
6083 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6085 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6086 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6087 _mm_maskz_reduce_ss): New.
6088 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6089 __builtin_ia32_reducess_mask): Ditto..
6090 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6091 * config/i386/sse.md (reduces<mode>): Renamed to ...
6092 (reduces<mode><mask_scalar_name>): ... this.
6093 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6095 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6096 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6098 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6100 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6101 * selftest-run-tests.c (selftest::run_tests): Call
6102 selftest::unique_ptr_tests_cc_tests.
6103 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6104 * unique-ptr-tests.cc: New file.
6106 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6109 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6111 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6112 (make_hard_regno_born, make_hard_regno_dead): Update
6113 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6115 2017-10-16 Jeff Law <law@redhat.com>
6117 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6119 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6121 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6123 2017-10-16 Olivier Hainque <hainque@adacore.com>
6125 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6126 with_cpu if we were configured for an e500v2 target cpu name.
6128 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6130 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6131 * doc/invoke.texi: Document +nodsp as a valid extension for
6134 2017-10-16 Martin Liska <mliska@suse.cz>
6136 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6137 (test_set_range): Likewise.
6138 (test_range_functions): Rename to ...
6139 (test_bit_in_range): ... this.
6140 (sbitmap_c_tests): Add new test.
6142 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6144 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6146 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6147 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6149 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6151 * config/aarch64/aarch64-builtins.c
6152 (aarch64_types_quadopu_lane_qualifiers): New.
6153 (TYPES_QUADOPU_LANE): New.
6154 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6155 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6156 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6157 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6158 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6159 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6160 (Vdottype, DOTPROD): New.
6161 (sur): Add SDOT and UDOT.
6163 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6165 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6166 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6167 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6169 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6170 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6171 Enable TARGET_DOTPROD.
6172 (cortex-a75.cortex-a55): Likewise.
6173 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6175 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6177 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6178 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6180 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6182 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6183 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6184 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6185 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6186 * config/arm/types.md (neon_dot, neon_dot_q): New.
6187 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6189 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6191 * config/arm/arm.h (TARGET_DOTPROD): New.
6192 * config/arm/arm.c (arm_arch_dotprod): New.
6193 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6194 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6195 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6196 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6197 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6198 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6199 * doc/invoke.texi (armv8.2-a): Document dotprod
6201 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6203 * i386.c (ix86_vec_cost): New function.
6204 (ix86_rtx_costs): Handle vector operations better.
6205 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6206 * x86-tune-costs.h: Add new costs to all tables.
6208 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6210 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6212 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6213 divsd, sqrtss and sqrtsd
6214 * x86-tune-costs.h: Add new entries to all costs.
6215 (znver1_cost): Fix to match real instruction latencies.
6217 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6218 Michael Collison <michael.collison@arm.com>
6220 * compare-elim.c: Include emit-rtl.h.
6221 (can_merge_compare_into_arith): New function.
6222 (try_validate_parallel): Likewise.
6223 (try_merge_compare): Likewise.
6224 (try_eliminate_compare): Call the above when no previous clobber
6226 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6229 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6233 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6234 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6238 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6239 to be any operand_equal_p operands. For & (B - 1) require
6240 B to be power of 2. Recognize
6241 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6243 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6246 * optabs.c (expand_memory_blockage): Fix call of
6247 targetm.have_memory_blockage.
6249 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6252 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6253 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6254 objects to extra_objs instead of overwriting it.
6256 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6258 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6259 Use any_fp_register_operand as operand[3] predicate. Simplify
6260 equality test for operands[2] and operands[4] memory location.
6261 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6262 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6263 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6264 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6265 any_fp_register_operand as operand[1] predicate. Simplify
6266 equality test for operands[0] and operands[3] memory location.
6267 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6268 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6269 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6271 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6273 * target-insns.def: Add memory_blockage.
6274 * optabs.c (expand_memory_blockage): New function.
6275 (expand_asm_memory_barrier): Rename ...
6276 (expand_asm_memory_blockage): ... to this.
6277 (expand_mem_thread_fence): Call expand_memory_blockage
6278 instead of expand_asm_memory_barrier.
6279 (expand_mem_singnal_fence): Ditto.
6280 (expand_atomic_load): Ditto.
6281 (expand_atomic_store): Ditto.
6282 * doc/md.texi (Standard Pattern Names For Generation):
6283 Document memory_blockage instruction pattern.
6285 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6287 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6288 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6289 * config/rl78/rl78.md: New define_expand "adddi3".
6291 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6293 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6296 2017-10-13 Jeff Law <law@redhat.com>
6298 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6300 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6303 * internal-fn.c (expand_mul_overflow): If both operands have
6304 the same highpart of -1 or 0 and the topmost bit of lowpart
6305 is different, overflow is if res <= 0 rather than res < 0.
6307 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6309 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6310 TARGET_P9_VECTOR code for unaligned_load case.
6312 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6314 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6316 2017-10-13 Nathan Sidwell <nathan@acm.org>
6318 * tree-core.h (tree_contains_struct): Make bool.
6319 * tree.c (tree_contains_struct): Likewise.
6320 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6321 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6322 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6323 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6324 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6326 2017-10-13 Richard Biener <rguenther@suse.de>
6328 * graphite-isl-ast-to-gimple.c
6329 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6330 parameters and dominance check.
6331 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6332 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6333 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6334 Do not update SSA form here or do intermediate IL verification.
6335 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6336 (graphite_initialize): Remove check on the number of loops in
6337 the function and inline into graphite_transform_loops.
6338 (graphite_finalize): Inline into graphite_transform_loops.
6339 (graphite_transform_loops): Perform SSA update and IL verification
6341 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6343 2017-10-13 Richard Biener <rguenther@suse.de>
6345 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6346 graphite_expression_type_precision): Avoid global constructor
6348 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6349 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6350 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6351 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6352 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6353 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6355 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6358 * config/i386/i386.h (ix86_red_zone_size): New.
6359 * config/i386/i386.md (push peephole2s): Replace
6360 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6362 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6363 Alan Hayward <alan.hayward@arm.com>
6364 David Sherwood <david.sherwood@arm.com>
6366 * combine.c (can_change_dest_mode): Reject changes in
6367 REGMODE_NATURAL_SIZE.
6369 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6370 Alan Hayward <alan.hayward@arm.com>
6371 David Sherwood <david.sherwood@arm.com>
6373 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6374 (expand_debug_source_expr): Likewise.
6375 * combine.c (combine_simplify_rtx): Likewise.
6376 * cse.c (fold_rtx): Likewise.
6377 * fwprop.c (canonicalize_address): Likewise.
6378 * targhooks.c (default_shift_truncation_mask): Likewise.
6380 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6381 Alan Hayward <alan.hayward@arm.com>
6382 David Sherwood <david.sherwood@arm.com>
6384 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6385 (widened_mode): Likewise.
6386 (expand_unop): Likewise.
6387 * ree.c (transform_ifelse): Likewise.
6388 (merge_def_and_ext): Likewise.
6389 (combine_reaching_defs): Likewise.
6390 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6392 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6393 Alan Hayward <alan.hayward@arm.com>
6394 David Sherwood <david.sherwood@arm.com>
6396 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6397 * combine.c (gen_lowpart_for_combine): Likewise.
6398 * dwarf2out.c (rtl_for_decl_location): Likewise.
6399 * final.c (alter_subreg): Likewise.
6400 * rtlhooks.c (gen_lowpart_general): Likewise.
6401 (gen_lowpart_if_possible): Likewise.
6403 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6404 Alan Hayward <alan.hayward@arm.com>
6405 David Sherwood <david.sherwood@arm.com>
6407 * calls.c (expand_call): Use subreg_lowpart_offset.
6408 * cse.c (cse_insn): Likewise.
6409 * regcprop.c (copy_value): Likewise.
6410 (copyprop_hardreg_forward_1): Likewise.
6412 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6415 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6416 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6417 =Q constraints to +Q and into insn condition add check
6418 that operands[0] and operands[1] are equal.
6419 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6420 =Q constraints to +Q and into insn condition add check
6421 that operands[0] is equal to either operands[1] or operands[2].
6424 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6425 instead of handling MINUS_EXPR twice (once for each argument),
6426 canonicalize operand order and handle just once, use rtype where
6427 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6430 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6431 any values of __C while still being pattern recognizable as a simple
6434 2017-10-13 Richard Biener <rguenther@suse.de>
6436 PR tree-optimization/82451
6438 2017-10-02 Richard Biener <rguenther@suse.de>
6440 PR tree-optimization/82355
6441 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6442 a mapping for the enclosing loop but avoid generating one for
6444 (copy_bb_and_scalar_dependences): Remove premature codegen
6445 error on PHIs in blocks duplicated into multiple places.
6446 * graphite-scop-detection.c
6447 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6448 in the region use it as loop and nest to analyze the DR in.
6449 (try_generate_gimple_bb): Likewise.
6450 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6451 (add_loop_constraints): For blocks in a loop not in the region
6452 create a dimension with a single iteration.
6453 * sese.h (gbb_loop_at_index): Remove assert.
6455 * cfgloop.c (loop_preheader_edge): For the loop tree root
6456 return the single successor of the entry block.
6457 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6458 Reset the SCEV hashtable and niters.
6459 * graphite-scop-detection.c
6460 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
6461 assert that we only have POLYNOMIAL_CHREC that vary in loops
6462 contained in the region.
6463 (scop_detection::graphite_can_represent_expr): Adjust.
6464 (scop_detection::stmt_has_simple_data_refs_p): For loops
6465 not in the region set loop to NULL. The nest is now the
6466 entry edge to the region.
6467 (try_generate_gimple_bb): Likewise.
6468 * sese.c (scalar_evolution_in_region): Adjust for
6469 instantiate_scev change.
6470 * tree-data-ref.h (graphite_find_data_references_in_stmt):
6471 Make nest parameter the edge into the region.
6472 (create_data_ref): Likewise.
6473 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
6474 entry edge into a region and adjust instantiate_scev calls.
6475 (create_data_ref): Likewise.
6476 (graphite_find_data_references_in_stmt): Likewise.
6477 (find_data_references_in_stmt): Pass the loop preheader edge
6478 from the nest argument.
6479 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
6480 parameter the edge into the region.
6481 (instantiate_parameters): Use the loop preheader edge as entry.
6482 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
6484 (get_instantiated_value_entry): Make instantiate_below parameter
6485 the edge into the region.
6486 (instantiate_scev_name): Likewise. Adjust dominance checks,
6487 when we cannot use loop-based instantiation instantiate by
6488 walking use-def chains.
6489 (instantiate_scev_poly): Adjust.
6490 (instantiate_scev_binary): Likewise.
6491 (instantiate_scev_convert): Likewise.
6492 (instantiate_scev_not): Likewise.
6493 (instantiate_array_ref): Remove.
6494 (instantiate_scev_3): Likewise.
6495 (instantiate_scev_2): Likewise.
6496 (instantiate_scev_1): Likewise.
6497 (instantiate_scev_r): Do not blindly handle N-operand trees.
6498 Do not instantiate array-refs. Handle all constants and invariants.
6499 (instantiate_scev): Make instantiate_below parameter
6500 the edge into the region.
6501 (resolve_mixers): Use the loop preheader edge for the region
6502 parameter to instantiate_scev_r.
6503 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6505 2017-10-13 Richard Biener <rguenther@suse.de>
6507 PR tree-optimization/82525
6508 * graphite-isl-ast-to-gimple.c
6509 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6511 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6512 Fail code generation when we cannot represent the isl integer.
6513 (binary_op_to_tree): Elide modulo operations that are no-ops
6514 in the type we code generate. Remove now superfluous code
6517 2017-10-13 Richard Biener <rguenther@suse.de>
6519 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6520 (scop_detection::harmful_loop_in_region): Remove premature
6521 IV type restriction.
6522 (scop_detection::graphite_can_represent_scev): We can handle
6523 pointer IVs just fine.
6525 2017-10-13 Alan Modra <amodra@gmail.com>
6527 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6528 "Clobbers and Scratch Registers". Add paragraph on
6529 alternative to clobbers for scratch registers and OpenBLAS
6532 2017-10-13 Alan Modra <amodra@gmail.com>
6534 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6535 example of a memory input for a string of known length. Move
6536 commentary out of table. Add a number of new examples
6537 covering array memory inputs.
6539 2017-10-12 Martin Liska <mliska@suse.cz>
6541 PR tree-optimization/82493
6542 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6543 (test_range_functions): New function.
6544 (sbitmap_c_tests): Likewise.
6545 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6546 * selftest.h (sbitmap_c_tests): New function.
6548 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6550 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6552 * config/rs6000/amo.h: Fix spacing issue.
6554 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6557 * config/i386/i386.md (*ashl<mode>3_mask_1,
6558 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6559 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6562 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6564 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6565 (profile_probability): Set max_probability
6566 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6568 * profile-count.c (profile_probability::differs_from_p): Do not
6569 rely on max_probaiblity == 10000
6571 2017-10-12 Jeff Law <law@redhat.com>
6573 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6576 2017-10-12 Martin Sebor <msebor@redhat.com>
6580 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6581 (handle_alias_pairs): Call it.
6582 * common.opt (-Wattribute-alias): New option.
6583 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6584 * doc/invoke.texi (-Wattribute-alias): Document.
6586 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6589 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6591 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6593 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6594 (make_hard_regno_born, make_hard_regno_dead): Update
6595 bb_killed_pseudos and bb_gen_pseudos.
6597 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6599 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6601 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6603 * config/alpha/alpha.c (alpha_split_conditional_move):
6604 Use std::swap instead of manually swapping.
6605 (alpha_stdarg_optimize_hook): Ditto.
6606 (alpha_canonicalize_comparison): Ditto.
6608 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6610 * tree-loop-distribution.c (struct builtin_info): New struct.
6611 (struct partition): Refactor fields into struct builtin_info.
6612 (partition_free): Free struct builtin_info.
6613 (build_size_arg_loc, build_addr_arg_loc): Delete.
6614 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6615 information from struct builtin_info.
6616 (find_single_drs): New function refactored from classify_partition.
6617 Also moved builtin validity checks to this function.
6618 (compute_access_range, alloc_builtin): New functions.
6619 (classify_builtin_st, classify_builtin_ldst): New functions.
6620 (classify_partition): Refactor code into functions find_single_drs,
6621 classify_builtin_st and classify_builtin_ldst.
6622 (distribute_loop): Don't do runtime alias check when distributing
6624 (find_seed_stmts_for_distribution): New function.
6625 (pass_loop_distribution::execute): Refactor code finding seed
6626 stmts into above function. Support distribution for the innermost
6627 two-level loop nest. Adjust dump information.
6629 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6631 * tree-loop-distribution.c: Adjust the general comment.
6632 (NUM_PARTITION_THRESHOLD): New macro.
6633 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6634 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6635 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6636 in call to build_partition_graph.
6637 (finalize_partitions): New parameter. Make loop distribution more
6638 conservative by fusing more partitions.
6639 (distribute_loop): Don't do runtime alias check in case of loop nest
6641 (find_seed_stmts_for_distribution): New function.
6642 (prepare_perfect_loop_nest): New function.
6643 (pass_loop_distribution::execute): Refactor code finding seed stmts
6644 and loop nest into above functions. Support loop nest distribution.
6645 Adjust dump information accordingly.
6647 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6649 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6650 and set PTYPE_SEQUENTIAL for merged partition.
6652 2017-10-12 Richard Biener <rguenther@suse.de>
6654 PR tree-optimization/69728
6656 2017-09-19 Richard Biener <rguenther@suse.de>
6658 PR tree-optimization/69728
6659 * graphite-sese-to-poly.c (schedule_error): New global.
6660 (add_loop_schedule): Handle empty domain by failing the
6662 (build_original_schedule): Handle schedule_error.
6664 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6665 domain by returning an unchanged schedule.
6667 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6669 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6670 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6672 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6674 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6677 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6680 * expr.c (store_field): Don't optimize away bitsize == 0 store
6681 from CALL_EXPR with addressable return type.
6683 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6685 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6686 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6687 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6688 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6689 TARGET_ISEL instead of TARGET_ISEL<sel>.
6691 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6693 * config/rs6000/rs6000.c
6694 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6696 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6698 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6700 (reg_or_cint_operand): Fix comment.
6701 (reg_or_zero_operand): New predicate.
6702 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6703 * config/rs6000/rs6000.c (output_isel): Delete.
6704 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6705 instead of reg_or_cint_operand. Output instruction directly (not via
6707 (isel_unsigned_<mode>): Ditto.
6708 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6709 gpc_reg_operand. Add an instruction alternative for this. Output
6710 instruction directly.
6711 (*isel_reversed_unsigned_<mode>): Ditto.
6713 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6715 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6716 (TARGET_CANONICALIZE_COMPARISON): Define.
6718 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6721 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6722 Check whether the dest is REG before adding REG_EQUIV note.
6724 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6727 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6729 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6730 (make_hard_regno_born, make_hard_regno_dead): Update
6731 bb_killed_pseudos and bb_gen_pseudos.
6733 2017-10-11 Nathan Sidwell <nathan@acm.org>
6735 * incpath.h (enum incpath_kind): Name enum, prefix values.
6736 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6737 * incpath.c (heads, tails): Use INC_MAX.
6738 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6739 (merge_include_chains, split_quote_chain,
6740 register_include_chains): Update incpath_kind names.
6741 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6742 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6744 (add_framework_path, darwin_register_objc_includes): Likewise.
6745 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6747 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6749 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6750 Do not use float_operator operator predicate.
6751 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6752 * config/i386/predicates.md (float_operator): Remove predicate.
6754 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6756 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6757 (*jccxf_i387): Ditto.
6758 (*jcc<mode>_i387): Ditto.
6759 (*jccu<mode>_i387): Ditto.
6760 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6761 (*jcc_*_i387 splitters): Remove.
6762 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6763 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6764 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6767 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6769 * profile-count.h (slow_safe_scale_64bit): New function.
6770 (safe_scale_64bit): New inline.
6771 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6772 * profile-count.c: Include wide-int.h
6773 (slow_safe_scale_64bit): New.
6775 2017-10-11 Nathan Sidwell <nathan@acm.org>
6777 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6778 HAS_DECL_ASSEMBLER_NAME_P.
6779 * gimple-expr.c (gimple_decl_printable_name: Check
6780 HAS_DECL_ASSEMBLER_NAME_P too.
6781 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6782 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6783 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6784 * passes.c (rest_of_decl_compilation): Check
6785 HAS_DECL_ASSEMBLER_NAME_P too.
6786 * recog.c (verify_changes): Likewise.
6787 * tree-pretty-print.c (dump_decl_name): Likewise.
6788 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6790 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6791 (SET_DECL_ASSEMBLER_NAME): Use it.
6792 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6793 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6794 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6796 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6798 * config.gcc (i386, x86_64): Add extra objects.
6799 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6800 (ix86_min_insn_size): Declare.
6801 (ix86_issue_rate): Declare.
6802 (ix86_adjust_cost): Declare.
6803 (ia32_multipass_dfa_lookahead): Declare.
6804 (ix86_macro_fusion_p): Declare.
6805 (ix86_macro_fusion_pair_p): Declare.
6806 (ix86_bd_has_dispatch): Declare.
6807 (ix86_bd_do_dispatch): Declare.
6808 (ix86_core2i7_init_hooks): Declare.
6809 (ix86_atom_sched_reorder): Declare.
6810 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6811 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6812 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6813 (rip_relative_addr_p): Rename to ...
6814 (ix86_rip_relative_addr_p): ... this one; export.
6815 (memory_address_length): Update.
6816 (ix86_issue_rate): Move to x86-tune-sched.c.
6817 (ix86_flags_dependent): Move to x86-tune-sched.c.
6818 (ix86_agi_dependent): Move to x86-tune-sched.c.
6819 (exact_dependency_1): Move to x86-tune-sched.c.
6820 (exact_store_load_dependency): Move to x86-tune-sched.c.
6821 (ix86_adjust_cost): Move to x86-tune-sched.c.
6822 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6823 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6824 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6825 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6826 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6827 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6828 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6829 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6830 (min_insn_size): Rename to ...
6831 (ix86_min_insn_size): ... this one; export.
6832 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6833 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6834 (core2i7_first_cycle_multipass_backtrack): Move to
6835 x86-tune-sched-core.c.
6836 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6837 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6838 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6839 (ix86_avoid_jump_mispredicts): Update.
6840 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6841 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6842 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6843 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6844 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6845 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6846 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6847 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6848 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6849 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6850 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6851 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6852 (BIG): Move to ix86-tune-sched-bd.c.
6853 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6854 (enum insn_path): Move to ix86-tune-sched-bd.c.
6855 (get_mem_group): Move to ix86-tune-sched-bd.c.
6856 (is_cmp): Move to ix86-tune-sched-bd.c.
6857 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6858 (is_branch): Move to ix86-tune-sched-bd.c.
6859 (is_prefetch): Move to ix86-tune-sched-bd.c.
6860 (init_window): Move to ix86-tune-sched-bd.c.
6861 (allocate_window): Move to ix86-tune-sched-bd.c.
6862 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6863 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6864 (process_end_window): Move to ix86-tune-sched-bd.c.
6865 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6866 (find_constant): Move to ix86-tune-sched-bd.c.
6867 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6868 (has_immediate): Move to ix86-tune-sched-bd.c.
6869 (get_insn_path): Move to ix86-tune-sched-bd.c.
6870 (get_insn_group): Move to ix86-tune-sched-bd.c.
6871 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6872 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6873 (add_insn_window): Move to ix86-tune-sched-bd.c.
6874 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6875 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6876 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6877 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6878 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6879 (do_dispatch): Move to ix86-tune-sched-bd.c.
6880 (has_dispatch): Move to ix86-tune-sched-bd.c.
6881 * config/i386/t-i386: Add new object files.
6882 * config/i386/x86-tune-costs.h: New file.
6883 * config/i386/x86-tune-sched-atom.c: New file.
6884 * config/i386/x86-tune-sched-bd.c: New file.
6885 * config/i386/x86-tune-sched-core.c: New file.
6886 * config/i386/x86-tune-sched.c: New file.
6888 2017-10-11 Liu Hao <lh_mouse@126.com>
6890 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6891 the generic version below instead.
6892 (should_colorize): Recognize Windows consoles as terminals
6894 * pretty-print.c [__MINGW32__] (write_all): New function.
6895 [__MINGW32__] (find_esc_head): Likewise.
6896 [__MINGW32__] (find_esc_terminator): Likewise.
6897 [__MINGW32__] (eat_esc_sequence): Likewise.
6898 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6900 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6903 2017-10-11 Richard Biener <rguenther@suse.de>
6905 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6906 Properly call analyze_scalar_evolution with the loop of the stmt.
6908 2017-10-11 Richard Biener <rguenther@suse.de>
6910 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6911 * tree-core.h (tree_base): Add chrec_var union member.
6912 * tree.h (CHREC_VAR): Remove.
6913 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6914 * tree-chrec.h (build_polynomial_chrec): Adjust.
6915 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6916 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6918 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6920 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6921 * match.pd: ... here.
6922 ((T) X == (T) Y): Relax condition.
6924 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6926 PR tree-optimization/82472
6927 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6929 (break_alias_scc_partitions): Update postorder number.
6931 2017-10-11 Martin Liska <mliska@suse.cz>
6934 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6936 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6938 2017-10-11 Martin Liska <mliska@suse.cz>
6940 * print-rtl.c (print_insn): Move declaration of idbuf
6941 to same scope as name.
6943 2017-10-11 Martin Liska <mliska@suse.cz>
6948 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6951 2017-10-11 Martin Liska <mliska@suse.cz>
6954 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6957 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6959 * config.gcc (default_gnu_indirect_function): Default to yes for
6960 arm*-*-linux* with glibc.
6962 2017-10-11 Richard Biener <rguenther@suse.de>
6964 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6965 default-defs and types we do not want to analyze.
6966 (interpret_loop_phi): Replace unreachable code with an assert.
6967 (compute_scalar_evolution_in_loop): Remove and inline ...
6968 (analyze_scalar_evolution_1): ... here, replacing condition with
6969 what makes the intent clearer. Remove handling of cases
6970 get_scalar_evolution now handles.
6972 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6974 PR rtl-optimization/81434
6975 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6976 comment for main loop. In sched_group_found if, also add checks for
6977 pass and min_cost_group.
6979 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6981 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6982 (rs6000_insn_cost): New function.
6983 * config/rs6000/rs6000.md (cost): New attribute.
6985 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6986 H.J. Lu <hongjiu.lu@intel.com>
6990 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6991 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6992 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6993 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6994 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6995 OPTION_MASK_ISA_MMX and some other bit set require both
6996 mmx and the ISAs without the mmx bit.
6997 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6998 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6999 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7000 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7001 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7002 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7003 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7004 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7005 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7006 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7007 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7008 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7009 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7010 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7011 Add OPTION_MASK_ISA_MMX.
7013 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7015 * config.gcc (armv7*-*-freebsd*): New target.
7016 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7018 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7020 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7021 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7024 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7026 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7027 a template parameter.
7028 (WIDE_INT_REF_FOR): Update accordingly.
7029 * tree.h (wi::int_traits <const_tree>): Delete.
7030 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7031 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7032 (wi::tree_to_wide_ref): New typedef.
7033 (wi::to_wide): New function.
7034 * calls.c (get_size_range): Use wi::to_wide when operating on
7036 * cgraph.c (cgraph_node::create_thunk): Likewise.
7037 * config/i386/i386.c (ix86_data_alignment): Likewise.
7038 (ix86_local_alignment): Likewise.
7039 * dbxout.c (stabstr_O): Likewise.
7040 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7041 * expr.c (const_vector_from_tree): Likewise.
7042 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7043 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7044 (fold_negate_expr_1, int_const_binop_1, const_binop)
7045 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7046 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7047 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7048 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7049 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7050 (fold_not_const, round_up_loc): Likewise.
7051 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7052 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7053 (alloca_call_type): Likewise.
7054 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7055 * godump.c (go_output_typedef): Likewise.
7056 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7057 * internal-fn.c (get_min_precision): Likewise.
7058 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7059 * ipa-polymorphic-call.c
7060 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7061 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7062 (ipa_modify_call_arguments): Likewise.
7063 * match.pd: Likewise.
7064 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7065 * print-tree.c (print_node_brief, print_node): Likewise.
7066 * stmt.c (expand_case): Likewise.
7067 * stor-layout.c (layout_type): Likewise.
7068 * tree-affine.c (tree_to_aff_combination): Likewise.
7069 * tree-cfg.c (group_case_labels_stmt): Likewise.
7070 * tree-data-ref.c (dr_analyze_indices): Likewise.
7071 (prune_runtime_alias_test_list): Likewise.
7072 * tree-dump.c (dequeue_and_dump): Likewise.
7073 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7074 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7075 * tree-pretty-print.c (dump_generic_node): Likewise.
7076 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7077 (simple_iv_with_niters): Likewise.
7078 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7079 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7080 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7081 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7082 (refine_value_range_using_guard, number_of_iterations_ne_max)
7083 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7084 (get_cst_init_from_scev, record_nonwrapping_iv)
7085 (scev_var_range_cant_overflow): Likewise.
7086 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7087 * tree-ssa-pre.c (compute_avail): Likewise.
7088 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7089 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7090 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7091 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7092 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7093 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7094 (dump_case_nodes, try_switch_expansion): Likewise.
7095 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7096 (vect_do_peeling): Likewise.
7097 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7098 * tree-vect-stmts.c (vectorizable_load): Likewise.
7099 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7100 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7101 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7102 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7103 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7104 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7105 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7106 (evrp_dom_walker::before_dom_children): Likewise.
7107 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7108 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7109 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7110 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7111 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7112 (get_range_pos_neg): Likewise.
7113 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7114 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7115 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7116 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7117 * config/avr/avr.c (avr_fold_builtin): Likewise.
7118 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7119 * config/msp430/msp430.c (msp430_attr): Likewise.
7120 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7121 * config/powerpcspe/powerpcspe-c.c
7122 (altivec_resolve_overloaded_builtin): Likewise.
7123 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7124 (rs6000_expand_ternop_builtin): Likewise.
7125 * config/rs6000/rs6000-c.c
7126 (altivec_resolve_overloaded_builtin): Likewise.
7127 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7128 (rs6000_expand_ternop_builtin): Likewise.
7129 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7131 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7133 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7134 when copying loop nest with only one inner loop.
7136 2017-10-10 Richard Biener <rguenther@suse.de>
7138 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7139 blocks if SCEV is active.
7140 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7142 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7143 (scev_initialize): Assert we are not yet initialized.
7145 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7147 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7148 inner loop's exit stmt by making it always exit the loop, otherwise
7149 we would generate an infinite empty loop.
7151 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7153 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7154 renaming variables in new preheader if it's deleted.
7156 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7158 * tree-loop-distribution.c (struct partition): Remove unused field
7159 loops of the structure.
7160 (partition_alloc, partition_free): Ditto.
7161 (build_rdg_partition_for_vertex): Ditto.
7163 2017-10-09 Jeff Law <law@redhat.com>
7165 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7166 return type to match prototype and documentation.
7168 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7170 * config/rs6000/rs6000.c (processor_costs): Move to ...
7171 * config/rs6000/rs6000.h: ... here.
7172 (rs6000_cost): Declare.
7174 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7176 * except.c (setjmp_fn): New global variable.
7177 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7178 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7179 if DONT_USE_BUILTIN_SETJMP is defined.
7181 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7183 * target.def (insn_cost): New hook.
7184 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7185 * doc/tm.texi: Regenerate.
7186 * rtlanal.c (insn_cost): Use the new hook.
7188 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7190 * combine.c (combine_validate_cost): Compute the new insn_cost,
7191 not just pattern_cost.
7192 (try_combine): Adjust comment.
7194 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7196 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7198 * combine.c (uid_insn_cost): Adjust comment.
7199 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7201 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7202 * dse.c (find_shift_sequence): Ditto.
7203 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7204 (bb_valid_for_noce_process_p): Use pattern_cost.
7205 * rtl.h (insn_rtx_cost): Delete.
7206 (pattern_cost): New prototype.
7207 (insn_cost): New prototype.
7208 * rtlanal.c (insn_rtx_cost): Rename to...
7209 (pattern_cost): ... this.
7212 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7214 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7215 (*jcc<mode>_0_r_i387): Ditto.
7216 (*jccxf_r_i387): Ditto.
7217 (*jcc<mode>_r_i387): Ditto.
7218 (*jccu<mode>_r_i387): Ditto.
7219 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7220 (*jcc): Rename from *jcc_1.
7222 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7224 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7225 deferred rescans after the lvx/stvx recombination pre-pass.
7227 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7229 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7230 memory operation instruction support.
7231 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7232 (rs6000-ibm-aix[789]*): Likewise.
7233 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7234 Document new functions.
7236 2017-10-09 Richard Biener <rguenther@suse.de>
7238 PR tree-optimization/82397
7239 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7240 equality only for semantically equal trees.
7242 2017-10-09 Richard Biener <rguenther@suse.de>
7244 PR tree-optimization/82449
7245 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7246 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7247 allow constant addresses.
7248 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7251 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7253 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7256 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7259 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7262 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7265 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7267 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7270 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7271 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7273 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7275 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7276 (WI_BINARY_PREDICATE_RESULT): Likewise.
7277 (wi::binary_traits::operator_result): New type.
7278 (wi::binary_traits::predicate_result): Likewise.
7279 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7280 (generic_wide_int::operator==, generic_wide_int::operator!=)
7281 (generic_wide_int::operator&, generic_wide_int::and_not)
7282 (generic_wide_int::operator|, generic_wide_int::or_not)
7283 (generic_wide_int::operator^, generic_wide_int::operator+
7284 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7286 (operator~, unary operator-, operator==, operator!=, operator&)
7287 (operator|, operator^, operator+, binary operator-, operator*): New
7289 * expr.c (get_inner_reference): Use wi::bit_and_not.
7290 * fold-const.c (fold_binary_loc): Likewise.
7291 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7292 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7293 (bit_value_binop): Likewise.
7294 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7295 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7296 (extract_range_from_binary_expr_1): Likewise.
7297 (masked_increment): Likewise.
7298 (simplify_bit_ops_using_ranges): Likewise.
7300 2017-10-09 Martin Jambor <mjambor@suse.cz>
7303 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7304 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7305 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7307 (hsa_fixup_mov_insn_type): New function.
7308 (hsa_op_with_type::get_in_type): Use it.
7309 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7310 immediates in an assert.
7311 (hsa_op_with_type::extend_int_to_32bit): New method.
7312 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7313 types. Convert to dest type if necessary.
7314 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7315 (reg_for_gimple_ssa): Pass false as min32int to
7316 hsa_type_for_scalar_tree_type.
7317 (gen_hsa_addr): Fixup type when creating addresable temporary.
7318 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7319 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7320 necessary. Call hsa_fixup_mov_insn_type.
7321 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7322 extend operands and convert to dest type if necessary.
7323 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7324 to dest type if necessary.
7325 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7326 if conversion nt necessary and size matches.
7327 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7328 to dest type if necessary.
7329 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7330 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7332 (gen_hsa_clrsb): Likewise.
7333 (gen_hsa_ffs): Likewise.
7334 (gen_hsa_divmod): Extend operands and convert to dest type if
7336 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7338 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7340 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7341 Remove empty default arguments. Use a brace block as output
7343 (conditional return): Ditto.
7345 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7346 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7348 (group_ending_nop): Ditto.
7349 (doloop_end): Ditto.
7350 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7351 (splitters for those): Ditto.
7353 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7355 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7356 a conditional jump (and the compare for it) so that pc_rtx is the
7358 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7359 for the deleted and renamed ctr<mode>_internal[234] patterns.
7360 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7361 Delete second conditional return pattern.
7362 (ctr<mode>_internal2): Delete this second bdnz pattern.
7363 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7364 (ctr<mode>_internal4): Delete this second bdz pattern.
7366 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7368 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7369 (always_initialized_rtx_for_ssa_name_p): New predicate.
7370 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7371 (finish_out_of_ssa): Free new field of SA.
7372 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7373 * tree-ssa-coalesce.c: Include tree-ssa.h.
7374 (get_parm_default_def_partitions): Remove extern keyword.
7375 (get_undefined_value_partitions): New function.
7376 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7377 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7379 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7381 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7383 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7385 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7387 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7388 for targets that preffer 128bit.
7390 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7392 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7394 2017-10-08 Olivier Hainque <hainque@adacore.com>
7396 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7397 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7399 (thumb_set_return_address): Likewise.
7401 2017-10-08 Olivier Hainque <hainque@adacore.com>
7403 * common/config/arm/arm-common.c (arm_except_unwind_info):
7404 Handle DWARF2_UNWIND_INFO.
7406 2017-10-07 Michael Collison <michael.collison@arm.com>
7408 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7411 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7413 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7414 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7415 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7416 defined, force the creation of a new block for a dispatch label.
7418 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7420 * invoke.texi (Wsuggest-attribute=cold): Document.
7421 * common.opt (Wsuggest-attribute=cold): New
7422 * ipa-pure-const.c (warn_function_cold): New function.
7423 * predict.c (compute_function_frequency): Use it.
7424 * predict.h (warn_function_cold): Declare.
7426 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7428 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7431 2017-10-06 Martin Liska <mliska@suse.cz>
7433 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7434 keyword for member functions.
7435 (struct sanopt_tree_couple): New struct.
7436 (struct sanopt_tree_couple_hash): New function.
7437 (struct sanopt_ctx): Add new hash_map.
7438 (has_dominating_ubsan_ptr_check): New function.
7439 (record_ubsan_ptr_check_stmt): Likewise.
7440 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7441 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7442 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7444 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7447 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7448 aarch64_simd_valid_immediate on CONST_VECTORs.
7449 (aarch64_reg_or_bic_imm): Likewise.
7451 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7453 PR rtl-optimization/82396
7454 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7456 2017-10-06 Sebastian Pop <sebpop@gmail.com>
7458 * graphite-dependences.c (scop_get_reads): Move code to...
7459 (scop_get_must_writes): Move code to...
7460 (scop_get_may_writes): Move code to...
7461 (scop_get_reads_and_writes): ... here.
7462 (scop_get_dependences): Call scop_get_reads_and_writes.
7464 2017-10-06 Jakub Jelinek <jakub@redhat.com>
7466 PR tree-optimization/82434
7467 * fold-const.h (can_native_encode_type_p,
7468 can_native_encode_string_p): Remove.
7469 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
7470 don't encode anything, just return what would be otherwise returned.
7471 (native_encode_fixed, native_encode_complex, native_encode_vector):
7473 (native_encode_string): Likewise. Inline by hand
7474 can_native_encode_string_p.
7475 (can_native_encode_type_p): Remove.
7476 (can_native_encode_string_p): Remove.
7477 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
7478 STRING_CSTs using can_native_encode_string_p, test all
7479 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
7480 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
7481 argument from native_encode_expr.
7482 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
7483 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
7486 2017-10-06 Richard Biener <rguenther@suse.de>
7488 PR tree-optimization/82397
7489 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
7490 operand_equal_p but rely on data_ref_compare_tree for detecting
7492 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7493 to match up with dr_group_sort_cmp.
7495 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7498 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7500 * config/s390/s390-builtin-types.def: Regenerate.
7502 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7505 * config/s390/s390-builtin-types.def: Regenerate.
7506 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7507 Change flag from B_VXE to B_VX.
7508 (s390_vec_min_dbl): Remove B_VXE flag.
7510 2017-10-06 Richard Biener <rguenther@suse.de>
7512 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7513 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7514 translate_isl_ast_to_gimple::is_valid_rename,
7515 translate_isl_ast_to_gimple::get_rename,
7516 translate_isl_ast_to_gimple::get_def_bb_for_const,
7517 translate_isl_ast_to_gimple::get_new_name,
7518 translate_isl_ast_to_gimple::collect_all_ssa_names,
7519 translate_isl_ast_to_gimple::copy_loop_phi_args,
7520 translate_isl_ast_to_gimple::collect_all_ssa_names,
7521 translate_isl_ast_to_gimple::copy_loop_phi_args,
7522 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7523 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7524 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7525 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7526 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7527 translate_isl_ast_to_gimple::copy_cond_phi_args,
7528 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7529 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7530 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7531 translate_isl_ast_to_gimple::rename_uses,
7532 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7533 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7534 (set_rename_for_each_def): Likewise.
7535 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7536 here. Handle rewriting SCEV analyzable uses here.
7537 (copy_bb_and_scalar_dependences): Generate code for PHI
7539 (graphite_regenerate_ast_isl): Adjust.
7540 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7541 (add_write, add_read): New functions.
7542 (build_cross_bb_scalars_def): Use it and simplify.
7543 (build_cross_bb_scalars_use): Likewise.
7544 (graphite_find_cross_bb_scalar_vars): Inline into...
7545 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7546 simulating out-of-SSA. Compute liveout and add dependencies.
7547 (build_scops): Force an empty entry block.
7548 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7550 (sese_build_liveouts): Declare.
7551 (sese_trivially_empty_bb_p): Likewise.
7552 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7553 compute liveout and debug_liveout.
7554 (sese_bad_liveouts_use): Remove.
7555 (sese_reset_debug_liveouts_bb): Likewise.
7556 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7557 (sese_build_liveouts): Build liveout and debug_liveout and store
7559 (new_sese_info): Adjust.
7560 (free_sese_info): Likewise.
7561 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7562 do not build liveout here.
7563 (move_sese_in_condition): Adjust region entry.
7564 (scev_analyzable_p): Match up with chrec_apply requirements.
7565 (sese_trivially_empty_bb_p): New.
7566 * tree-into-ssa.c (get_reaching_def): Properly support generating
7567 default-defs for incremental rewrite of anonymous names.
7569 2017-10-06 Richard Biener <rguenther@suse.de>
7571 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7572 precision do not perform modulo reduction.
7574 2017-10-06 Richard Biener <rguenther@suse.de>
7576 PR tree-optimization/82436
7577 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7578 conservatively choose the vectorization factor when checking
7579 whether we can perform the required load permutation.
7580 (vect_transform_slp_perm_load): Assert when we may not fail.
7582 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7584 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7585 message for incompatible -msdata=* and -mcall-* options.
7587 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7589 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7590 rate for post-reload scheduling.
7592 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7594 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7596 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7598 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7599 to improve monte carlo in scimark.
7601 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7603 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7604 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7605 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7606 pentium4_cost, nocona_cost): Set reassociation width to 1.
7607 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7608 width to 2 for fp operations and 1 otherwise.
7609 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7611 (atom_cost): Set reassociation width to 2.
7612 (slm_cost, generic_cost): Set fp reassociation width
7613 to 2 and 1 otherwise.
7614 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7615 (core_cost): Set fp reassociation width to 4 and vector to 2.
7616 (ix86_reassociation_width): Rewrite using cost table; special case
7617 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7618 and TARGET_AVX128_OPTIMAL.
7619 * config/i386/i386.h (processor_costs): Add
7620 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7621 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7622 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7623 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7624 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7625 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7627 2017-10-05 Nathan Sidwell <nathan@acm.org>
7629 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7631 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7633 * config/arm/arm.c (arm_test_fpu_data): New.
7634 (arm_run_selftests): Call arm_test_fpu_data.
7636 2017-10-04 Nathan Sidwell <nathan@acm.org>
7638 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7640 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7642 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7644 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7645 check type for aarch64_simd_valid_immediate.
7646 (aarch64_output_simd_mov_immediate): Update prototype.
7647 (aarch64_simd_valid_immediate): Update prototype.
7648 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7649 support for ORR-immediate.
7650 (and<mode>3): modified pattern to add support for BIC-immediate.
7651 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7652 now checks for valid immediate for BIC and ORR based on new enum
7654 (aarch64_output_simd_mov_immediate): Function now used to output
7655 BIC/ORR imm as well based on new enum argument.
7656 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7658 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7659 (aarch64_reg_or_bic_imm): Likewise.
7661 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7663 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7664 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7666 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7670 PR rtl-optimization/82396
7671 * haifa-sched.c (autopref_multipass_init): Simplify
7673 (autopref_rank_data): Simplify sort order.
7674 * sched-int.h (autopref_multipass_data_): Remove
7675 multi_mem_insn_p, min_offset and max_offset.
7677 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7679 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7681 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7683 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7686 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7688 * doc/sourcebuild.texi: Document vect_long_mult.
7690 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7692 PR tree-optimization/82413
7693 * fold-const.c (build_range_check): Use widest_int when comparing
7694 the maximum ETYPE value with HIGH.
7696 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7698 PR rtl-optimization/82396
7699 * haifa-sched.c (autopref_multipass_init): Simplify
7701 (autopref_rank_data): Simplify sort order.
7702 * sched-int.h (autopref_multipass_data_): Remove
7703 multi_mem_insn_p, min_offset and max_offset.
7705 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7707 PR tree-optimization/82381
7708 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7709 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7712 PR tree-optimization/82374
7713 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7714 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7715 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7716 current_function_decl to the new decl.
7718 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7720 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7721 helper macro for IEEE float128 hardware built-in functions.
7722 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7724 (TRUNCF128_ODD): Likewise.
7725 (ADDF128_ODD): Likewise.
7726 (SUBF128_ODD): Likewise.
7727 (MULF128_ODD): Likewise.
7728 (DIVF128_ODD): Likewise.
7729 (FMAF128_ODD): Likewise.
7730 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7731 UNSPEC_TRUNC_ROUND_TO_ODD.
7732 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7733 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7734 floating point round to odd instructions.
7735 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7736 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7737 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7738 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7739 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7740 (trunc<mode>sf2_hw): Change the truncate with round to odd
7741 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7742 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7743 to odd hardware instructions.
7744 (sub<mode>3_odd): Likewise.
7745 (mul<mode>3_odd): Likewise.
7746 (div<mode>3_odd): Likewise.
7747 (sqrt<mode>2_odd): Likewise.
7748 (fma<mode>4_odd): Likewise.
7749 (fms<mode>4_odd): Likewise.
7750 (nfma<mode>4_odd): Likewise.
7751 (nfms<mode>4_odd): Likewise.
7752 (trunc<mode>df2_odd): Change the truncate with round to odd
7753 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7755 * doc/extend.texi (PowerPC built-in functions): Update documentation
7756 for existing IEEE float128-bit built-in functions. Add built-in
7757 functions that generate the IEEE 128-bit floating point round to
7760 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7762 PR rtl-optimization/77729
7763 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7764 to (X&(C1&~C2))|C2 transformations.
7766 2017-10-03 Martin Jambor <mjambor@suse.cz>
7768 PR tree-optimization/82363
7769 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7770 mismatch, mark lacc written regardless of racc.
7772 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7774 PR tree-optimization/82381
7775 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7776 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7777 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7778 fallthrough into reassoc_stmt_dominates_stmt_p.
7781 * combine.c (combine_instructions): Don't combine in unreachable
7784 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7787 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7788 function to not use the have_cpu variable. Do not set cpu_index,
7789 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7791 (rs6000_valid_attribute_p): Remove duplicate initializations of
7792 old_optimize and func_optimize.
7793 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7794 (rs6000_activate_target_options): Make global.
7795 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7798 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7800 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7801 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7802 Set *poffset to 0 and *psize and *pmax_size to -1 if
7803 *poffset + *psize overflows in HOST_WIDE_INT.
7805 PR tree-optimization/82387
7806 PR tree-optimization/82388
7807 PR tree-optimization/82389
7808 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7809 instead of live_bytes non-NULL.
7811 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7814 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7817 2017-10-02 Richard Biener <rguenther@suse.de>
7819 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7820 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7821 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7823 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7825 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7826 requested precision matches the type's.
7827 * calls.c (alloc_max_size): Calculate the new candidate size as
7828 a widest_int and use wi::to_widest when comparing it with the
7829 current candidate size.
7830 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7831 zero rather than integer_zero_node.
7832 * match.pd: Check for a no-op conversion before using wi::add
7833 rather than after. Use tree_to_uhwi when summing small shift
7834 counts into an unsigned int.
7836 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7837 Alan Hayward <alan.hayward@arm.com>
7838 David Sherwood <david.sherwood@arm.com>
7841 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7842 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7843 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7844 POINTER_AND_FP_REGS.
7846 2017-10-02 Richard Biener <rguenther@suse.de>
7848 PR tree-optimization/82355
7849 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7850 a mapping for the enclosing loop but avoid generating one for
7852 (copy_bb_and_scalar_dependences): Remove premature codegen
7853 error on PHIs in blocks duplicated into multiple places.
7854 * graphite-scop-detection.c
7855 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7856 in the region use it as loop and nest to analyze the DR in.
7857 (try_generate_gimple_bb): Likewise.
7858 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7859 (add_loop_constraints): For blocks in a loop not in the region
7860 create a dimension with a single iteration.
7861 * sese.h (gbb_loop_at_index): Remove assert.
7863 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7865 * omp-expand.c (adjust_context_scope): New function.
7866 (expand_parallel_call): Call adjust_context_scope.
7868 2017-10-01 Jeff Law <law@redhat.com>
7870 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7871 dom_opt_dom_walker class with direct access to private members.
7872 Add comments. Call test_for_singularity.
7873 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7874 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7875 m_dummy_cond anymore.
7876 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7878 (pass_dominator:execute): Build the dummy_cond here and pass it
7879 to the dom_opt_dom_walker ctor.
7880 (test_for_singularity): New function.
7882 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7883 Maya Rashish <coypu@sdf.org>
7885 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7886 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7887 (alpha*-*-netbsd*) Use nbsd_tm_file.
7888 (arm*-*-netbsdelf*) Likewise.
7889 (i[34567]86-*-netbsdelf*) Likewise.
7890 (x86_64-*-netbsd*) Likewise.
7891 (mips*-*-netbsd*) Likewise.
7892 (powerpc-*-netbsd*) Likewise.
7893 (sh*-*-netbsd*) Likewise.
7894 (sparc-*-netbsdelf*) Likewise.
7895 (sparc64-*-netbsd*) Likewise.
7896 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7898 (vax-*-netbsdelf*) Likewise.
7899 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7900 (UINT_FAST8_TYPE) Likewise.
7901 (INT_FAST16_TYPE) Check CHAR_FAST16.
7902 (UINT_FAST16_TYPE) Likewise.
7904 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7907 * config/i386/i386.md
7908 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7909 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7910 *divmodsi4_zext_2): New define_insn_and_split.
7911 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7912 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7913 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7914 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7915 New define_insn_and_split.
7916 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7917 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7918 operands[1] having DImode when mode is SImode.
7920 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7921 always SImode for DIV and MOD in REG_EQUAL notes.
7923 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7926 * match.pd: Fix handling of NaNs in pattern.
7928 2017-09-29 Jeff Law <law@redhat.com>
7930 * sbitmap.c (bitmap_bit_in_range_p): New function.
7931 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7932 * tree-ssa-dse.c (live_bytes_read): New function.
7933 (dse_classify_store): Ignore reads of dead bytes.
7935 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7936 typos and whitespace errors.
7937 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7938 * config/s390/s390.c (s390_emit_prologue): Likewise.
7940 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7943 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7944 with a symbol for LRA.
7946 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7948 PR rtl-optimization/82338
7949 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7951 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7953 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7954 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7955 (qsort_chk): Declare.
7956 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7957 (qsort_chk): New function.
7959 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7961 PR tree-optimization/82337
7962 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7963 phi definition if the PHI result appears in an abnormal PHI.
7964 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7965 basis appears in an abnormal PHI.
7967 2017-09-29 Richard Biener <rguenther@suse.de>
7969 * graphite-isl-ast-to-gimple.c
7970 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7971 (binary_op_to_tree): Use it.
7972 (get_rename_from_scev): Likewise.
7973 (copy_loop_phi_nodes): Likewise.
7974 (copy_bb_and_scalar_dependences): Likewise.
7975 (translate_pending_phi_nodes): Likewise.
7977 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7980 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7981 for movabsq $(i32 << shift), r64.
7983 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7985 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7986 index when encoding %esp as %rsp to avoid 0x67 prefix.
7988 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7990 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7991 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7993 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7995 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7996 Extensions with more than 16 double VFP registers.
7997 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7998 to_clear_mask and all code related to it. Replace the remaining
7999 entry by a sbitmap and adapt code accordingly.
8001 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8003 * brig-builtins.def: Change pure attributes to const.
8005 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8007 * config.gcc (default_gnu_indirect_function): Default to yes for
8008 sparc*-*-linux* with glibc.
8010 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8012 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8013 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8014 when creating .init_array and .fini_array sections with priority
8017 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8020 * config/aarch64/aarch64.c
8021 (aarch64_builtin_support_vector_misalignment): Always return false
8022 when misalignment is unknown.
8024 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8026 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8027 this function to return false if the definition used by the swap
8028 instruction is artificial, or if the memory address from which the
8029 constant value is loaded is not represented by a base address held
8030 in a register or if the base address register is a frame or stack
8031 pointer. Additionally, return false if the base address of the
8032 loaded constant is a SYMBOL_REF but is not considered to be a
8034 (replace_swapped_load_constant): New function.
8035 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8036 loaded constant vector with a load of a swapped constant vector.
8038 2017-09-27 Carl Love <cel@us.ibm.com>
8040 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8041 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8042 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8045 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8047 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8048 first, always call autopref_rank_data otherwise.
8050 2017-09-27 Richard Biener <rguenther@suse.de>
8052 * graphite-scop-detection.c (find_scop_parameters): Move
8053 loop bound handling ...
8054 (gather_bbs::before_dom_children): ... here, avoiding the need
8055 to build scop_info->loop_nest.
8056 (record_loop_in_sese): Remove.
8057 * sese.h (sese_info_t::loop_nest): Remove.
8058 * sese.c (new_sese_info): Do not allocate loop_nest.
8059 (free_sese_info): Do not free loop_nest.
8061 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8064 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8065 lhs from calls if the lhs has addressable type.
8067 2017-09-27 Richard Biener <rguenther@suse.de>
8069 * graphite.h (scop::max_alias_set): New member.
8070 * graphite-scop-detection.c: Remove references to non-existing
8071 --param in comments.
8072 (build_alias_sets): Record the maximum alias set used for drs.
8073 (build_scops): Support zero as unlimited for
8074 --param graphite-max-arrays-per-scop.
8075 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8077 (build_poly_sr_1): ... here. Compute alias set based on the
8078 maximum alias set used for drs rather than
8079 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8081 2017-09-27 Richard Biener <rguenther@suse.de>
8083 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8084 --param loop-block-tile-size=0 to disable tiling.
8086 2017-09-27 Richard Biener <rguenther@suse.de>
8088 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8089 (graphite-max-nb-scop-params): Document special value zero.
8090 * domwalk.h (dom_walker::STOP): New symbolical constant.
8091 (dom_walker::dom_walker): Add optional parameter for bb to
8093 (dom_walker::~dom_walker): Declare.
8094 (dom_walker::before_dom_children): Document STOP return value.
8095 (dom_walker::m_user_bb_to_rpo): New member.
8096 (dom_walker::m_bb_to_rpo): Likewise.
8097 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8098 mapping here if not provided by the user.
8099 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8100 provided by the user.
8101 (dom_walker::STOP): Define.
8102 (dom_walker::walk): Do not compute bb to RPO mapping here.
8103 Support STOP return value from before_dom_children to stop
8105 * graphite-optimize-isl.c (optimize_isl): If the schedule
8106 is the same still generate code if -fgraphite-identity
8107 or -floop-parallelize-all are given.
8108 * graphite-scop-detection.c: Include cfganal.h.
8109 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8111 (gather_bbs::before_dom_children): Return STOP for BBs
8113 (build_scops): Compute bb to RPO mapping and pass it to
8114 the domwalk. Treat --param graphite-max-nb-scop-params=0
8115 as not limiting the number of params.
8116 * graphite.c (graphite_initialize): Remove limit on the
8117 number of basic-blocks in a function.
8118 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8119 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8120 default value of 10.
8122 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8124 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8125 Adjust code to eliminate needing to do the shift right 32-bits
8126 operation after XSCVDPSPN.
8128 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8130 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8131 ((X / Y) != 0 -> X >= Y): Likewise.
8133 2017-09-26 Carl Love <cel@us.ibm.com>
8135 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8136 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8137 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8138 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8139 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8140 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8141 definitions and overloading.
8142 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8143 statement for P9V_BUILTIN_XST_LEN_R.
8144 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8145 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8146 define_expand and define_insn for the instructions and builtins.
8147 * doc/extend.texi: Update the built-in documentation file for the new
8149 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8150 define_insn for the instructions
8152 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8155 * gcc/config/netbsd-protos.h: New file.
8156 * gcc/config/netbsd.c: New file.
8157 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8158 * gcc/config/t-netbsd: New file.
8159 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8160 (tmake_file) Add t-netbsd.
8161 (extra_objs) Add netbsd.o.
8163 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8167 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8169 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8171 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8172 sign extension from a vector register to a GPR by doing a 32-bit
8173 direct move and then an EXTSW.
8174 (extendsi<mode>2 splitter): Likewise.
8175 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8176 right or vector extract after doing XSCVDPSPN. Use
8177 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8179 (movdi_from_sf_zero_ext): Likewise.
8180 (reload_gpr_from_vsxsf): Likewise.
8181 (p8_mfvsrd_4_disf): Delete, no longer used.
8182 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8183 then needing to move the SFmode to a GPR to use the XSCVDPSP
8184 instruction instead of FRSP and XSCVDPSPN.
8185 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8186 it is adjacent to the other XSCVSPDP insns.
8187 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8188 SFmode to be in traditional Altivec registers.
8189 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8190 (vsx_xscvspdpn): Likewise.
8191 (vsx_xscvspdpn_scalar): Likewise.
8193 2017-09-26 Martin Jambor <mjambor@suse.cz>
8195 * tree-sra.c (compare_access_positions): Put integral types first,
8196 stabilize sorting of integral types, remove conditions putting
8197 non-full-precision integers last.
8198 (sort_and_splice_var_accesses): Disable scalarization if a
8199 non-integert would be represented by a non-full-precision integer.
8201 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8203 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8204 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8205 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8206 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8207 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8208 conditionals inside the function instead of around it. Call
8209 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8210 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8212 2017-09-26 Richard Biener <rguenther@suse.de>
8214 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8216 (scop_detection::build_scop_breadth): ... this. Removed.
8217 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8218 (scop_detection::harmful_stmt_in_bb): Likewise.
8219 (scop_detection::graphite_can_represent_stmt): Likewise.
8220 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8221 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8222 (scop_detection::can_represent_loop_1): ... this. Removed.
8223 (scop_detection::harmful_loop_in_region): Simplify after inlining
8224 the above and remove more quadraticness.
8225 (build_scops): Adjust.
8226 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8229 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8232 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8233 REGNO (base) == SP_REG if base is a REG.
8236 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8237 if it is different SSA_NAME.
8238 (optimize_range_tests_cmp_bitwise): New function.
8239 (optimize_range_tests): Call it.
8241 2017-09-26 Richard Biener <rguenther@suse.de>
8243 PR tree-optimization/82321
8244 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8245 for the def being inside the loop.
8247 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8249 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8251 * config/s390/s390-builtins.def: Fix constraint on op4.
8253 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8255 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8256 independent expanders.
8257 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8258 ("vec_ordered", "vec_unordered"): New expanders.
8260 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8262 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8265 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8267 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8268 vec_unpacks_lo_v16qi.
8269 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8271 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8273 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8274 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8275 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8277 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8279 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8281 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8283 ("*vec_slb<mode>"): New insn pattern.
8284 ("vec_shr_<mode>"): New expander.
8285 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8286 and force the shift count operand to V16QImode.
8287 ("vec_srb<mode>"): Set shift count mode to V16QI.
8289 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8291 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8292 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8293 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8295 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8298 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8300 2017-09-26 Richard Biener <rguenther@suse.de>
8302 PR tree-optimization/82320
8303 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8306 2017-09-25 Jeff Law <law@redhat.com>
8308 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8309 prototype for new argument.
8310 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8311 mostly extracted from rs6000_emit_allocate_stack.
8312 (rs6000_emit_probe_stack_range_stack_clash): New function.
8313 (rs6000_emit_allocate_stack): Call
8314 rs6000_emit_probe_stack_range_stack_clash as needed.
8315 (rs6000_emit_probe_stack_range): Add additional argument
8316 to call to gen_probe_stack_range{si,di}.
8317 (output_probe_stack_range): New.
8318 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8319 (output_probe_stack_range_stack_clash): New.
8320 (rs6000_emit_prologue): Emit notes into dump file as requested.
8321 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8322 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8323 Add additional operand and pass it to output_probe_stack_range.
8325 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8327 PR tree-optimization/82163
8328 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8329 (checking_verify_loop_closed_ssa): New parameter.
8330 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8331 (check_loop_closed_ssa_stmt): Delete.
8332 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8333 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8334 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8337 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8339 * brig-builtins.def: Treat HSAIL barrier builtins as
8340 setjmp/longjump style functions.
8342 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8344 * target.def (constant_alignment): New hook.
8345 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8346 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8347 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8348 * doc/tm.texi: Regenerate.
8349 * targhooks.h (default_constant_alignment): Declare.
8350 (constant_alignment_word_strings): Likewise.
8351 * targhooks.c (default_constant_alignment): New function.
8352 (constant_alignment_word_strings): Likewise.
8353 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8354 instead of CONSTANT_ALIGNMENT.
8355 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8356 (force_const_mem): Likewise.
8357 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8358 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8359 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8360 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8361 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8363 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8364 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8365 constant_alignment_word_strings.
8366 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8367 (CONSTANT_ALIGNMENT): Likewise.
8368 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8369 (arm_constant_alignment): New function.
8370 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8371 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8372 constant_alignment_word_strings.
8373 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8374 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8375 constant_alignment_word_strings.
8376 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8377 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8378 (cris_constant_alignment): New function.
8379 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8380 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8381 (epiphany_constant_alignment): New function.
8382 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8383 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8384 constant_alignment_word_strings.
8385 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8386 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8387 constant_alignment_word_strings.
8388 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8389 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8390 constant_alignment_word_strings.
8391 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8392 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8393 * config/i386/i386.c (ix86_constant_alignment): Make static.
8394 Use the same interface as the target hook.
8395 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8396 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8397 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8398 constant_alignment_word_strings.
8399 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8400 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8401 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8402 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8403 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8404 constant_alignment_word_strings.
8405 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8406 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8407 constant_alignment_word_strings.
8408 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8409 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8410 constant_alignment_word_strings.
8411 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8412 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8414 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8415 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8416 * config/mips/mips.c (mips_constant_alignment): New function.
8417 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8418 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8419 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8420 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8421 (mmix_constant_alignment): Make static. Use the same interface
8423 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8424 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8425 constant_alignment_word_strings.
8426 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8427 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8428 constant_alignment_word_strings.
8429 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8430 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8431 constant_alignment_word_strings.
8432 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8433 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8434 (rs6000_constant_alignment): New function.
8435 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8436 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8437 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8438 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8439 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8440 (rs6000_constant_alignment): New function.
8441 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8442 * config/s390/s390.c (s390_constant_alignment): New function.
8443 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8444 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8445 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8446 constant_alignment_word_strings.
8447 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8448 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8449 (sparc_constant_alignment): New function.
8450 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8451 * config/spu/spu.c (spu_constant_alignment): New function.
8452 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8453 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8454 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8455 constant_alignment_word_strings.
8456 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8457 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8458 constant_alignment_word_strings.
8459 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
8460 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8461 constant_alignment_word_strings.
8462 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
8463 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8464 (visium_constant_alignment): New function.
8465 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
8466 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8467 (xtensa_constant_alignment): New function.
8468 * system.h (CONSTANT_ALIGNMENT): Poison.
8470 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
8472 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8473 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
8474 (rs6000_builtin_valid_without_lhs): New helper function.
8475 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8476 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
8478 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8480 * target.h (vec_perm_indices): Use unsigned short rather than
8482 (auto_vec_perm_indices): Likewise.
8483 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
8484 Use unsigned int rather than unsigned char.
8485 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
8487 2017-09-25 Richard Biener <rguenther@suse.de>
8489 * cfgloop.h (sort_sibling_loops): Declare.
8490 * cfgloop.c (sort_sibling_loops_cmp): New helper.
8491 (sort_sibling_loops): New function sorting the sibling loop list
8493 * graphite.c (graphite_transform_loops): Sort sibling loops.
8495 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8497 * target.def (vec_perm_const_ok): Change sel parameter to
8499 * optabs-query.c (can_vec_perm_p): Update accordingly.
8500 * doc/tm.texi: Regenerate.
8501 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8502 auto_vec_perm_indices and remove separate nelt field.
8503 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8504 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8505 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8506 (aarch64_expand_vec_perm_const): Update accordingly.
8507 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8508 to vec_perm_indices.
8509 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8510 auto_vec_perm_indices and remove separate nelt field.
8511 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8512 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8513 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8515 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8516 to vec_perm_indices.
8517 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8518 sel to vec_perm_indices.
8519 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8520 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8521 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8523 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8526 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8529 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8530 on the FUNCTION_DECL function context if it has a DIE that is a
8533 2017-09-25 Richard Biener <rguenther@suse.de>
8535 PR tree-optimization/82285
8536 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8539 2017-09-25 Tom de Vries <tom@codesourcery.com>
8543 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8546 2017-09-25 Richard Biener <rguenther@suse.de>
8548 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8549 ISL errors other than isl_error_quota happen. Dump if the
8550 schedule is the same.
8551 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8552 errors instead of aborting inside ISL.
8554 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8557 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8558 of libgcc_eh for m64.
8559 * config/i386/darwin64.h: Likewise.
8561 2017-09-25 Richard Biener <rguenther@suse.de>
8564 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8565 attribute for incomplete types nor twice for complete ones.
8567 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8570 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8571 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8574 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8577 * config/i386/i386.opt (mprefer-avx256): Use
8578 ix86_target_flags variable.
8579 * config/i386/i386.c (ix86_target_string): Move
8580 -mprefer-avx256 to flag2_opts.
8582 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8585 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8586 and x != -1 | y != -1 into (x & y) != -1.
8588 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8590 * config.gcc: Add new case statement to set
8591 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8592 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8593 s390x-*-linux* case statements. Added aarch64 to the list of
8594 supported architectures.
8596 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8598 PR tree-optimization/82289
8599 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8600 STMT_VINFO_RELEVANT_P.
8602 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8603 Alan Hayward <alan.hayward@arm.com>
8604 David Sherwood <david.sherwood@arm.com>
8606 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8607 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8608 (extract_range_from_binary_expr_1): Don't call
8609 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8611 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8612 Alan Hayward <alan.hayward@arm.com>
8613 David Sherwood <david.sherwood@arm.com>
8615 * target.def (preferred_vector_alignment): New hook.
8616 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8618 * doc/tm.texi: Regenerate.
8619 * targhooks.h (default_preferred_vector_alignment): Declare.
8620 * targhooks.c (default_preferred_vector_alignment): New function.
8621 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8623 (DR_TARGET_ALIGNMENT): New macro.
8624 (aligned_access_p): Update commentary.
8625 (vect_known_alignment_in_bytes): New function.
8626 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8628 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8629 Calculate the misalignment based on the target alignment rather than
8631 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8632 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8633 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8634 the target alignment, rather than masking the element misalignment
8635 with the number of elements in a vector. Also use the target
8636 alignment when calculating the maximum number of peels.
8637 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8638 instead of TYPE_ALIGN_UNIT.
8639 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8640 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8641 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8642 (vect_create_data_ref_ptr): Likewise.
8643 (vect_setup_realignment): Realign by ANDing with
8644 -DR_TARGET_MISALIGNMENT.
8645 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8646 the number of peels based on DR_TARGET_ALIGNMENT.
8647 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8648 with the guaranteed alignment boundary when deciding whether
8650 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8651 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8652 (ensure_base_align): Remove stmt_info parameter. Get the
8653 target base alignment from DR_TARGET_ALIGNMENT.
8654 (vectorizable_store): Update call accordingly. Interpret
8655 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8657 (vectorizable_load): Likewise.
8659 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8660 Alan Hayward <alan.hayward@arm.com>
8661 David Sherwood <david.sherwood@arm.com>
8663 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8664 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8665 (vect_enhance_data_refs_alignment): Likewise.
8667 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8669 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8670 error. Only quit immediately if parsing is complete.
8671 (BEGIN): Initialize fatal_err and parse_done.
8672 (begin fpu, end fpu): Check number of arguments.
8673 (begin arch, end arch): Likewise.
8674 (begin cpu, end cpu): Likewise.
8675 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8676 (optalias): Likewise.
8678 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8680 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8681 * config/arm/arm-isa.h: Delete. Move definitions to ...
8682 * arm-cpus.in: ... here. Use new feature and fgroup values.
8683 * config/arm/arm.c (arm_option_override): Use lower case for feature
8685 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8686 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8687 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8689 (print_isa_bits_for): New function.
8690 (gen_isa): New function.
8691 (gen_comm_data): Use print_isa_bits_for.
8692 (define feature): New keyword.
8693 (define fgroup): New keyword.
8694 * config/arm/t-arm (TM_H): Remove.
8695 (GTM_H): Add arm-isa.h.
8696 (arm-isa.h): Add rule to generate file.
8697 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8698 case for feature bit names.
8700 2017-09-22 Richard Biener <rguenther@suse.de>
8702 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8704 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8705 print of no dependency loops ...
8706 * graphite.c (graphite_transform_loops): ... here.
8707 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8709 (same_close_phi_node, remove_duplicate_close_phi,
8710 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8711 (canonicalize_loop_closed_ssa): ... here and simplify.
8712 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8713 (optimize_isl): Use dump_printf_loc to tell when we stopped
8714 optimizing because of an ISL timeout.
8716 2017-09-22 Richard Biener <rguenther@suse.de>
8718 PR tree-optimization/82291
8719 * tree-if-conv.c (predicate_mem_writes): Make sure to
8720 remove writes in blocks predicated with false.
8722 2017-09-22 Richard Biener <rguenther@suse.de>
8724 * sese.c: Include cfganal.h.
8725 (if_region_set_false_region): Remove.
8726 (create_if_region_on_edge): Likewise.
8727 (move_sese_in_condition): Re-implement without destroying
8730 2017-09-22 Richard Biener <rguenther@suse.de>
8732 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8733 Verify both BBs contain loop PHI nodes before dispatching to
8735 (graphite_regenerate_ast_isl): Do not recompute dominators,
8736 do not verify three times. Restructure for clarity.
8737 * graphite-scop-detection.c (same_close_phi_node,
8738 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8739 defined_in_loop_p, canonicalize_loop_closed_ssa,
8740 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8741 checking and SSA rewrite, move to ...
8742 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8743 (graphite_initialize): Do not pass in ctx, do not reset the
8744 SCEV cache, compute only dominators.
8745 (graphite_transform_loops): Allocate ISL ctx after
8746 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8747 Maintain post-dominators only around build_scops.
8748 * sese.c (if_region_set_false_region): Make static. Free
8749 and recompute dominators.
8750 (move_sese_in_condition): Assert we don't get called with
8751 post-dominators computed.
8752 * sese.h (if_region_set_false_region): Remove.
8754 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8756 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8757 mode attribute for TARGET_AVX512VL.
8759 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8761 * config/i386/i386.opt (mprefer-avx256): New option.
8762 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8764 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8765 for TARGET_PREFER_AVX256.
8766 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8768 2017-09-21 Jeff Law <law@redhat.com>
8770 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8771 Fix dump output if the only stack space is for pushed registers.
8773 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8775 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8778 2017-09-21 Martin Sebor <msebor@redhat.com>
8781 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8782 code (in C++) or code that triggers warnings.
8784 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8786 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8788 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8790 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8791 * sched-rgn.c: Ditto.
8792 * sel-sched-ir.c: Ditto.
8794 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8796 * toplev.h (set_random_seed): Adjust return type.
8797 * toplev.c (init_local_tick): Move eager initialization of random_seed
8798 to get_random_seed. Adjust comment.
8799 (init_random_seed): Inline to get_random_seed, delete.
8800 (get_random_seed): Initialize random_seed lazily.
8801 (set_random_seed): Do not return previous value.
8802 (print_switch_value): Do not call get_random_seed.
8804 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8806 * cgraph.c (delete_function_version): New, broken out from...
8807 (cgraph_node::delete_function_version): ...here. Rename to
8808 cgraph_node::delete_function_version_by_decl. Update all uses.
8809 (cgraph_node::remove): Call delete_function_version.
8811 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8814 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8815 VAR_DECLs to which addressable non-volatile parameters are mapped
8816 and for id->retvar after the return value assignment. Clear
8817 id->retval and id->retbnd after inlining.
8819 2017-09-21 Richard Biener <rguenther@suse.de>
8821 PR tree-optimization/82276
8822 PR tree-optimization/82244
8823 * tree-vrp.c (build_assert_expr_for): Set
8824 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8826 (remove_range_assertions): Revert earlier change.
8828 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8831 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8833 2017-09-21 Richard Biener <rguenther@suse.de>
8835 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8836 Restore valid IL after code generation errors.
8837 * graphite.c (graphite_transform_loops): Diagnose code
8838 generation issues as MSG_MISSED_OPTIMIZATION and continue
8839 with processing SCOPs.
8841 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8842 Alan Hayward <alan.hayward@arm.com>
8843 David Sherwood <david.sherwood@arm.com>
8845 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8846 rather than choosing between plus_constant and gen_rtx_<CODE>.
8847 * expr.c (emit_push_insn): Likewise.
8848 (expand_expr_real_2): Likewise.
8850 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8851 Alan Hayward <alan.hayward@arm.com>
8852 David Sherwood <david.sherwood@arm.com>
8854 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8856 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8857 Alan Hayward <alan.hayward@arm.com>
8858 David Sherwood <david.sherwood@arm.com>
8860 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8861 calling tree_to_uhwi.
8863 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8864 Alan Hayward <alan.hayward@arm.com>
8865 David Sherwood <david.sherwood@arm.com>
8867 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8868 INTEGER_CST rather than a negative test for ADDR_EXPR.
8870 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8871 Alan Hayward <alan.hayward@arm.com>
8872 David Sherwood <david.sherwood@arm.com>
8874 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8875 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8877 2017-09-21 Richard Biener <rguenther@suse.de>
8879 PR tree-optimization/71351
8880 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8881 graphite_create_new_loop_guard): Remove, fold remaining parts
8883 (translate_isl_ast_node_for): ... here and simplify.
8885 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8888 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8889 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8890 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8891 alternative always use QI mode, for -Os imov (=R,R) alternative
8892 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8895 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8896 Jeff Law <law@redhat.com>
8898 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8899 (allocate_stack_space): New function, partially extracted from
8901 (s390_emit_prologue): Track offset to most recent stack probe.
8902 Code to allocate space moved into allocate_stack_space.
8903 Dump actions when no stack is allocated.
8904 (s390_prologue_plus_offset): New function.
8905 (s390_emit_stack_probe): Likewise.
8907 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8909 * common.opt (Wa, Wl, Wp, g, gz=): Add
8911 (gno-column-info): Remove.
8912 (gcolumn-info): Drop RejectNegative.
8914 (gno-record-gcc-switches): Remove.
8915 (grecord-gcc-switches): Drop RejectNegative.
8916 (gno-split-dwarf): Remove.
8917 (gsplit-dwarf): Drop RejectNegative.
8918 (gno-strict-dwarf): Remove.
8919 (gstrict-dwarf): Drop RejectNegative.
8920 * config/darwin.opt (gfull, gused): Add RejectNegative.
8921 * dwarf2out.c (gen_producer_string): Drop
8922 gno-record-gcc-switches handler.
8923 * optc-gen.awk: Add g to prefixes with negative forms.
8924 * opts-common.c (remapping_prefix_p): New.
8925 (find_opt): Check it.
8926 (generate_canonical_option): Test g prefix.
8927 (option_map): Add -gno- mapping.
8928 (add_misspelling_candidates): Check remapping_prefix_p.
8930 2017-09-20 Jeff Law <law@redhat.com>
8932 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8933 thinko in stack clash protection support.
8935 * explow.c (compute_stack_clash_protection_loop_data): Use
8936 CONST_INT_P instead of explicit test. Verify object is a
8937 CONST_INT_P before looking at INTVAL.
8938 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8939 instead of explicit test.
8941 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8944 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8945 address instead of to r1 and r11.
8947 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8949 * config.gcc: Support "knm".
8950 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8951 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8953 * config/i386/i386.c (m_KNM): Define.
8954 (processor_target_table): Add "knm".
8956 (ix86_option_override_internal): Add "knm".
8957 (ix86_issue_rate): Add PROCESSOR_KNM.
8958 (ix86_adjust_cost): Ditto.
8959 (ia32_multipass_dfa_lookahead): Ditto.
8960 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8961 (fold_builtin_cpu): Add M_INTEL_KNM.
8962 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8963 (processor_type): Add PROCESSOR_KNM.
8964 * config/i386/x86-tune.def: Add m_KNM.
8965 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8967 2017-09-20 Richard Biener <rguenther@suse.de>
8969 PR tree-optimization/80213
8970 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8971 are allowed in empty BBs as well.
8972 (canonicalize_loop_closed_ssa): Also look for other complex
8974 (scop_detection::get_sese): Include the loop-closed PHI block
8976 (scop_detection::merge_sese): Remove code adding extra blocks.
8977 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8978 (build_scops): Assert the final returned scop is invalid.
8980 2017-09-20 Richard Biener <rguenther@suse.de>
8982 PR tree-optimization/82264
8983 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8985 (vn_phi_lookup): Likewise.
8986 (vn_phi_insert): Likewise.
8988 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8990 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8991 that fits into uhwi or shwi, add DW_AT_const_value regardless
8992 of early_dwarf without going through RTL, using add_AT_unsigned
8995 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8996 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8997 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8998 (DEBUG_MACRO_SECTION): Likewise.
8999 (DEBUG_LTO_MACRO_SECTION): Likewise.
9000 (DEBUG_STR_DWO_SECTION): Likewise.
9001 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9002 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9003 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9004 (DEBUG_LTO_LINE_STR_SECTION): Define.
9005 (init_sections_and_labels): Initialize debug_line_str_section
9006 variable. Initialize debug_loc_section for -gdwarf-5 to
9007 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9009 2017-09-20 Richard Biener <rguenther@suse.de>
9011 * graphite-sese-to-poly.c (extract_affine): Properly handle
9012 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9014 2017-09-20 Richard Biener <rguenther@suse.de>
9016 PR tree-optimization/81373
9017 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9018 Force SESE live-out defs to be handled even if they are
9021 2017-09-19 Jeff Law <law@redhat.com>
9023 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9024 nothing for stack adjustments with REG_STACK_CHECK.
9025 * sched-deps.c (parse_add_or_inc): Reject insns with
9026 REG_STACK_CHECK from dependency breaking.
9027 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9028 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9029 * reg-notes.def (STACK_CHECK): New note.
9031 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9032 (ix86_expand_prologue): Dump stack clash info as needed.
9033 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9035 * function.c (dump_stack_clash_frame_info): New function.
9036 * function.h (dump_stack_clash_frame_info): Prototype.
9037 (enum stack_clash_probes): New enum.
9039 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9040 flag_stack_clash_protection.
9041 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9042 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9043 (arm_frame_pointer_required): Likewise.
9044 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9045 (ia64_expand_prologue): Likewise.
9046 * config/mips/mips.c (mips_expand_prologue): Likewise.
9047 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9048 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9049 (sparc_flat_expand_prologue): Likewise.
9050 * config/spu/spu.c (spu_expand_prologue): Likewise.
9052 * explow.c: Include "params.h".
9053 (anti_adjust_stack_and_probe_stack_clash): New function.
9054 (get_stack_check_protect): Likewise.
9055 (compute_stack_clash_protection_loop_data): Likewise.
9056 (emit_stack_clash_protection_loop_start): Likewise.
9057 (emit_stack_clash_protection_loop_end): Likewise.
9058 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9059 Use anti_adjust_stack_and_probe_stack_clash.
9060 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9061 (emit_stack_clash_protection_loop_start): Likewise.
9062 (emit_stack_clash_protection_loop_end): Likewise.
9063 * rtl.h (get_stack_check_protect): Prototype.
9064 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9065 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9066 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9068 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9070 * doc/tm.texi: Rebuilt.
9071 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9072 get_stack_check_protect.
9073 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9074 * config/arm/arm.c (arm_expand_prologue): Likewise.
9075 (arm_frame_pointer_required): Likewise.
9076 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9077 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9078 * config/mips/mips.c (mips_expand_prologue): Likewise.
9079 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9080 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9081 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9082 (sparc_flat_expand_prologue): Likewise.
9084 * common.opt (-fstack-clash-protection): New option.
9085 * flag-types.h (enum stack_check_type): Note difference between
9086 -fstack-check= and -fstack-clash-protection.
9087 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9088 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9089 * toplev.c (process_options): Issue warnings/errors for cases
9090 not handled with -fstack-clash-protection.
9091 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9092 (-fstack-check): Note additional problem with -fstack-check=generic.
9093 Note that -fstack-check is primarily for Ada and refer users
9094 to -fstack-clash-protection for stack-clash-protection.
9095 Document new params for stack clash protection.
9097 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9099 * config/i386/i386.c (ix86_split_long_move): Do not handle
9100 address used for LEA in a special way.
9102 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9104 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9106 2017-09-19 Martin Sebor <msebor@redhat.com>
9109 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9110 of incompatible types.
9112 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9114 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9115 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9116 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9117 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9119 2017-09-19 Richard Biener <rguenther@suse.de>
9121 PR tree-optimization/82244
9122 * tree-vrp.c (remove_range_assertions): Do not propagate
9123 a constant to abnormals but replace the assert with a copy.
9125 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9127 PR rtl-optimization/57878
9128 PR rtl-optimization/68988
9129 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9130 avoidance test involving non_reload_pseudos. Move frequency test
9131 below the general fragmentation avoidance test.
9133 2017-09-19 Richard Biener <rguenther@suse.de>
9135 PR tree-optimization/69728
9136 * graphite-sese-to-poly.c (schedule_error): New global.
9137 (add_loop_schedule): Handle empty domain by failing the
9139 (build_original_schedule): Handle schedule_error.
9141 2017-09-19 Richard Biener <rguenther@suse.de>
9143 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9144 Do not iterate to sibling loops but only to siblings of inner
9147 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9150 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9151 that the registers are different.
9153 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9155 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9156 to processor_model and "amdfam17h" to arch_names_table.
9157 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9159 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9162 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9164 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9165 Alan Hayward <alan.hayward@arm.com>
9166 David Sherwood <david.sherwood@arm.com>
9168 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9170 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9172 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9173 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9174 vect_schedule_slp_instance.
9175 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9176 Update call to vect_slp_analyze_node_operations. Simplify return
9178 (vect_slp_analyze_bb_1): Update call accordingly.
9179 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9180 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9181 (vect_schedule_slp): Update call accordingly.
9183 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9184 Alan Hayward <alan.hayward@arm.com>
9185 David Sherwood <david.sherwood@arm.com>
9187 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9188 with types that aren't in fact scalar.
9190 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9192 * tree-vect-slp.c (vect_record_max_nunits): New function,
9194 (vect_build_slp_tree_1): ...here.
9195 (vect_build_slp_tree_2): Call it for phis too.
9197 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9199 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9200 to vect_get_vec_def_for_operand when getting the mask operand.
9202 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9203 Alan Hayward <alan.hayward@arm.com>
9204 David Sherwood <david.sherwood@arm.com>
9206 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9209 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9210 Alan Hayward <alan.hayward@arm.com>
9211 David Sherwood <david.sherwood@arm.com>
9213 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9214 calculation for vector booleans.
9216 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9217 Alan Hayward <alan.hayward@arm.com>
9218 David Sherwood <david.sherwood@arm.com>
9220 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9222 (vect_transform_stmt): ...here.
9223 (vect_analyze_stmt): Use it instead of calling
9224 vectorizable_live_operation directly.
9226 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9228 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9229 non-SIMT targets in acc vector loops.
9231 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9233 * configure.ac: Add arc and check if assembler supports gdwarf2.
9234 * configure: Regenerate.
9236 2017-09-18 Richard Biener <rguenther@suse.de>
9238 PR tree-optimization/82220
9239 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9240 epilogue niters from the min_profitable_iters compute.
9242 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9245 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9246 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9247 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9249 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9252 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9253 switching to a new text section.
9255 2017-09-18 Richard Biener <rguenther@suse.de>
9257 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9259 (build_alias_set): Reject aliases with no access function.
9261 2017-09-18 Richard Biener <rguenther@suse.de>
9263 PR tree-optimization/79622
9264 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9266 (build_cross_bb_scalars_use): Likewise.
9268 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9270 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9272 2017-09-18 Alan Modra <amodra@gmail.com>
9275 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9276 stack_pointer_rtx for count 0. Update comments. Break up
9277 large rtl expression.
9279 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9281 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9282 Increase to 20 bytes.
9283 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9284 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9285 or avx version of the stub.
9287 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9290 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9291 compute the minimum stack alignment. Also update preferred stack
9292 boundary for leaf functions.
9294 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9296 PR tree-optimization/82228
9297 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9300 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9302 * common/config/nds32/nds32-common.c
9303 (nds32_option_optimization_table): Refine formatting.
9304 (nds32_option_optimization_table): Use -fsched-pressure and
9305 -fomit-frame-pointer for specific optimization level.
9307 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9309 * config/nds32/nds32.c: Refine formatting and comments.
9310 * config/nds32/nds32.h: Likewise.
9311 * config/nds32/nds32.md: Likewise.
9312 * config/nds32/nds32-cost.c: Likewise.
9313 * config/nds32/nds32-isr.c: Likewise.
9314 * config/nds32/nds32-md-auxiliary.c: Likewise.
9315 * config/nds32/nds32-multiple.md: Likewise.
9316 * config/nds32/nds32-predicates.c: Likewise.
9318 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9319 Jakub Jelinek <jakub@redhat.com>
9321 Add support for -std=c++2a.
9322 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9324 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9326 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9329 * doc/extend.texi (Common Function Attributes): Add
9330 references to ARM, AArch64, and S/390 specific attributes.
9331 (Function Specific Option Pragmas): Add AArch64 and S/390
9332 to list of back ends that support the target pragma.
9334 2017-09-15 Nathan Sidwell <nathan@acm.org>
9336 * doc/standards.texi: Fix C++17 description. Update URLs for
9339 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9341 * common.opt (Wcast-align=strict): New warning option.
9342 * doc/invoke.texi: Document -Wcast-align=strict.
9344 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9346 * cgraph.h (cgraph_thunk_info): Add comments.
9347 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9348 assert for VIRTUAL_* arguments stricter.
9350 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9352 PR tree-optimization/71026
9353 * match.pd: Move RDIV patterns from fold-const.c
9354 * fold-const.c (distribute_real_division): Removed.
9355 (fold_binary_loc): Remove calls to distribute_real_divison.
9357 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9359 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9360 c++1z and gnu++1z as deprecated. Change other references to
9361 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9362 Change -Wc++1z-compat to -Wc++17-compat.
9363 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9364 * dwarf2out.c (highest_c_language): Handle C++17.
9365 (gen_compile_unit_die): Likewise.
9367 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9369 PR rtl-optimization/82192
9370 * combine.c (make_extraction): Don't look through non-paradoxical
9371 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9374 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9375 Alan Hayward <alan.hayward@arm.com>
9376 David Sherwood <david.sherwood@arm.com>
9378 * target.def (function_arg_offset): New hook.
9379 * targhooks.h (default_function_arg_offset): Declare.
9380 * targhooks.c (default_function_arg_offset): New function.
9381 * function.c (locate_and_pad_parm): Use
9382 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9383 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9384 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9385 * doc/tm.texi: Regenerate.
9386 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9387 * config/spu/spu.c (spu_function_arg_offset): New function.
9388 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9389 * system.h (FUNCTION_ARG_OFFSET): Poison.
9391 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9392 Alan Hayard <alan.hayward@arm.com>
9393 David Sherwood <david.sherwood@arm.com>
9395 * target.def (truly_noop_truncation): New hook.
9396 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9397 than TRULY_NOOP_TRUNCATION.
9398 * hooks.h (hook_bool_uint_uint_true): Declare.
9399 * hooks.c (hook_bool_uint_uint_true): New function.
9400 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9401 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9402 * doc/tm.texi: Regenerate.
9403 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9404 rather than TRULY_NOOP_TRUNCATION in comments.
9405 (simplify_comparison): Likewise.
9406 (record_truncated_value): Likewise.
9407 * expmed.c (extract_bit_field_1): Likewise.
9408 (extract_split_bit_field): Likewise.
9409 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9410 instead of TRULY_NOOP_TRUNCATION.
9411 * function.c (assign_parm_setup_block): Likewise.
9412 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9413 * rtlhooks.c: Include target.h.
9414 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9415 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9416 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9417 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9418 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9419 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9420 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9421 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9422 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9423 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9424 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9425 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9426 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9427 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9428 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9429 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9430 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9431 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9432 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9433 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9434 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9435 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9436 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9437 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9438 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9439 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9440 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9441 rather than TRULY_NOOP_TRUNCATION in comments.
9442 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9443 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9444 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9445 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9446 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9447 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9448 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9449 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9450 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9451 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9452 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9453 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9454 rather than TRULY_NOOP_TRUNCATION in comments.
9455 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9456 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9457 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
9458 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
9459 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
9460 TRULY_NOOP_TRUNCATION condition.
9461 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
9462 (TRULY_NOOP_TRUNCATION): Delete.
9463 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
9464 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
9465 * config/spu/spu.c (spu_truly_noop_truncation): New function.
9466 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9467 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
9468 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
9469 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
9470 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9471 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9472 rather than TRULY_NOOP_TRUNCATION in comments.
9473 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
9474 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
9475 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
9476 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
9477 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
9478 * system.h (TRULY_NOOP_TRUNCATION): Poison.
9480 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
9483 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
9484 (*cmp_ior): Likewise.
9485 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
9486 (*ior_scc_scc_cmp): Likewise.
9487 (*and_scc_scc): Likewise.
9488 (*and_scc_scc_cmp): Likewise.
9490 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9491 Alan Hayard <alan.hayward@arm.com>
9492 David Sherwood <david.sherwood@arm.com>
9494 * target.def (can_change_mode_class): New hook.
9495 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9496 (hard_regno_nregs): Likewise.
9497 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9498 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9499 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9500 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9501 (LOAD_EXTEND_OP): Update accordingly.
9502 * doc/tm.texi: Regenerate.
9503 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9504 CANNOT_CHANGE_MODE_CLASS.
9505 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9506 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9507 * combine.c (simplify_set): Update accordingly.
9508 * emit-rtl.c (validate_subreg): Likewise.
9509 * recog.c (general_operand): Likewise.
9510 * regcprop.c (mode_change_ok): Likewise.
9511 * reload1.c (choose_reload_regs): Likewise.
9512 (inherit_piecemeal_p): Likewise.
9513 * rtlanal.c (simplify_subreg_regno): Likewise.
9514 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9515 instead of CANNOT_CHANGE_MODE_CLASS.
9516 (reload_cse_simplify_operands): Likewise.
9517 * reload.c (push_reload): Use targetm.can_change_mode_class
9518 instead of CANNOT_CHANGE_MODE_CLASS.
9519 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9520 REG_CANNOT_CHANGE_MODE_P.
9521 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9522 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9523 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9524 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9525 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9526 (arm_can_change_mode_class): New function.
9527 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9528 than CANNOT_CHANGE_MODE_CLASS in comments.
9529 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9530 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9531 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9532 (ix86_can_change_mode_class): ...this new function, inverting the
9533 sense of the return value.
9534 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9535 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9536 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9537 (ia64_can_change_mode_class): New function.
9538 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9539 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9540 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9541 (m32c_can_change_mode_class): ...this new function, inverting the
9542 sense of the return value.
9543 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9544 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9545 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9546 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9547 (mips_can_change_mode_class): ...this new function, inverting the
9548 sense of the return value.
9549 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9550 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9551 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9552 (msp430_can_change_mode_class): New function.
9553 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9554 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9555 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9556 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9557 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9558 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9559 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9560 (pa_cannot_change_mode_class): Replace with...
9561 (pa_can_change_mode_class): ...this new function, inverting the
9562 sense of the return value.
9563 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9564 than CANNOT_CHANGE_MODE_CLASS in comments.
9565 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9566 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9567 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9568 (pdp11_cannot_change_mode_class): Replace with...
9569 (pdp11_can_change_mode_class): ...this new function, inverting the
9570 sense of the return value.
9571 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9572 * config/powerpcspe/powerpcspe-protos.h
9573 (rs6000_cannot_change_mode_class_ptr): Delete.
9574 * config/powerpcspe/powerpcspe.c
9575 (rs6000_cannot_change_mode_class_ptr): Delete.
9576 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9577 (rs6000_option_override_internal): Assign to
9578 targetm.can_change_mode_class instead of
9579 rs6000_cannot_change_mode_class_ptr.
9580 (rs6000_cannot_change_mode_class): Replace with...
9581 (rs6000_can_change_mode_class): ...this new function, inverting the
9582 sense of the return value.
9583 (rs6000_debug_cannot_change_mode_class): Replace with...
9584 (rs6000_debug_can_change_mode_class): ...this new function.
9585 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9586 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9587 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9588 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9589 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9591 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9592 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9593 (rs6000_option_override_internal): Assign to
9594 targetm.can_change_mode_class instead of
9595 rs6000_cannot_change_mode_class_ptr.
9596 (rs6000_cannot_change_mode_class): Replace with...
9597 (rs6000_can_change_mode_class): ...this new function, inverting the
9598 sense of the return value.
9599 (rs6000_debug_cannot_change_mode_class): Replace with...
9600 (rs6000_debug_can_change_mode_class): ...this new function.
9601 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9602 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9603 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9604 (s390_can_change_mode_class): ...this new function, inverting the
9605 sense of the return value.
9606 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9607 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9608 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9609 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9610 (sh_cannot_change_mode_class): Replace with...
9611 (sh_can_change_mode_class): ...this new function, inverting the
9612 sense of the return value.
9613 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9614 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9615 (sparc_can_change_mode_class): New function.
9616 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9617 * config/spu/spu.c (spu_can_change_mode_class): New function.
9618 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9619 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9620 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9621 (visium_can_change_mode_class): New function.
9622 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9624 2017-09-15 Richard Biener <rguenther@suse.de>
9626 PR tree-optimization/82217
9627 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9628 but not undefined case.
9630 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9633 * postreload.c (reload_cse_simplify_operands): Skip
9634 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9636 2017-09-15 Richard Biener <rguenther@suse.de>
9638 PR tree-optimization/68823
9639 * graphite-scop-detection.c (build_alias_set): If we have a
9640 possible dependence check whether we can handle them by just
9641 looking at the DRs DR_ACCESS_FNs.
9642 (build_scops): If build_alias_set fails, fail the SCOP.
9644 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9646 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9647 to support float128 built-in functions that require the ISA 3.0
9649 (BU_FLOAT128_3_HW): Likewise.
9650 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9652 (FMAF128): Likewise.
9654 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9655 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9656 floating point instructions.
9657 (rs6000_invalid_builtin): Likewise.
9658 (rs6000_builtin_mask_names): Likewise.
9659 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9660 (RS6000_BTM_FLOAT128_HW): Likewise.
9661 (RS6000_BTM_COMMON): Likewise.
9662 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9664 * doc/extend.texi (RS/6000 built-in functions): Document the
9665 IEEE 128-bit floating point square root and fused multiply-add
9668 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9670 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9671 reg (r2) isn't in the set of registers defined in the prologue.
9673 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9674 Alan Hayward <alan.hayward@arm.com>
9675 David Sherwood <david.sherwood@arm.com>
9677 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9678 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9679 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9680 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9681 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9683 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9684 max_vectorization_factor.
9685 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9687 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9688 Alan Hayward <alan.hayward@arm.com>
9689 David Sherwood <david.sherwood@arm.com>
9691 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9692 (vect_worthwhile_without_simd_p): Declare.
9693 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9694 (vectorizable_reduction): Use it.
9695 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9696 (vectorizable_operation): Likewise.
9698 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9699 Alan Hayward <alan.hayward@arm.com>
9700 David Sherwood <david.sherwood@arm.com>
9702 * tree-vectorizer.h (vect_get_num_copies): New function.
9703 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9704 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9705 (vectorizable_induction): Likewise.
9706 (vectorizable_live_operation): Likewise.
9707 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9708 (vectorizable_bswap): Likewise.
9709 (vectorizable_call): Likewise.
9710 (vectorizable_conversion): Likewise.
9711 (vectorizable_assignment): Likewise.
9712 (vectorizable_shift): Likewise.
9713 (vectorizable_operation): Likewise.
9714 (vectorizable_store): Likewise.
9715 (vectorizable_load): Likewise.
9716 (vectorizable_condition): Likewise.
9717 (vectorizable_comparison): Likewise.
9718 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9720 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9721 Alan Hayward <alan.hayward@arm.com>
9722 David Sherwood <david.sherwood@arm.com>
9724 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9725 of vect_init_vector.
9727 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9728 Alan Hayward <alan.hayward@arm.com>
9729 David Sherwood <david.sherwood@arm.com>
9731 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9732 an inline wrapper that provides a location.
9733 (gimple_build_vector): Likewise.
9734 * gimple-fold.c (gimple_build_vector_from_val): New function.
9735 (gimple_build_vector): Likewise.
9736 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9737 functions to build the initial value. Always return a gimple value.
9738 (get_initial_defs_for_reduction): Likewise. Only compute
9740 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9741 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9742 (vectorizable_induction): Use gimple_build_vector rather than
9745 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9746 Alan Hayward <alan.hayward@arm.com>
9747 David Sherwood <david.sherwood@arm.com>
9749 * target.h (vec_perm_indices): New typedef.
9750 (auto_vec_perm_indices): Likewise.
9751 * optabs-query.h: Include target.h
9752 (can_vec_perm_p): Take a vec_perm_indices *.
9753 * optabs-query.c (can_vec_perm_p): Likewise.
9754 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9755 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9756 * tree-vect-generic.c (lower_vec_perm): Likewise.
9757 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9758 (vect_grouped_load_supported): Likewise.
9759 (vect_shift_permute_load_chain): Likewise.
9760 (vect_permute_store_chain): Use auto_vec_perm_indices.
9761 (vect_permute_load_chain): Likewise.
9762 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9763 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9764 Update uses of can_vec_perm_p.
9765 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9766 mode with a number of elements. Take a vec_perm_indices *.
9767 (vect_create_epilog_for_reduction): Update accordingly.
9768 Use auto_vec_perm_indices.
9769 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9770 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9771 (vect_transform_slp_perm_load): Likewise.
9772 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9773 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9774 (vect_gen_perm_mask_checked): Likewise.
9775 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9776 (vect_gen_perm_mask_checked): Likewise.
9777 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9778 (vectorizable_store): Likewise.
9779 (vectorizable_load): Likewise.
9780 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9781 (vectorizable_bswap): Likewise.
9783 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9784 Alan Hayward <alan.hayward@arm.com>
9785 David Sherwood <david.sherwood@arm.com>
9787 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9788 * tree.c (build_vector): Likewise.
9789 (build_vector_from_ctor): Update accordingly.
9790 (build_vector_from_val): Likewise.
9791 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9792 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9793 * tree-vect-generic.c (add_rshift): Likewise.
9794 (expand_vector_divmod): Likewise.
9795 (optimize_vector_constructor): Likewise.
9796 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9797 (vect_transform_slp_perm_load): Likewise.
9798 (vect_schedule_slp_instance): Likewise.
9799 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9800 (vectorizable_call): Likewise.
9801 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9802 * expmed.c (make_tree): Likewise.
9803 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9804 a vector passed to build_vector.
9805 (fold_convert_const): Likewise.
9806 (exact_inverse): Likewise.
9807 (fold_ternary_loc): Likewise.
9808 (fold_relational_const): Likewise.
9809 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9810 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9811 (const_unop): Likewise. Store the reduction accumulator in a
9812 variable rather than an array.
9813 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9814 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9815 the new vector, rather than constructing it after the input arrays.
9816 (native_interpret_vector): Use auto_vec<tree> when building
9817 a vector passed to build_vector. Add elements in order.
9818 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9819 auto_vec<tree> when building a vector passed to build_vector.
9820 (vect_create_epilog_for_reduction): Likewise.
9821 (vectorizable_induction): Likewise.
9822 (get_initial_def_for_reduction): Likewise. Fix indentation of
9824 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9826 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9827 passed to build_vector.
9829 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9830 Alan Hayward <alan.hayward@arm.com>
9831 David Sherwood <david.sherwood@arm.com>
9833 * tree-core.h (tree_base::u): Add an "nelts" field.
9834 (tree_vector): Use VECTOR_CST_NELTS as the length.
9835 * tree.c (tree_size): Likewise.
9836 (make_vector): Initialize VECTOR_CST_NELTS.
9837 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9838 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9839 TYPE_VECTOR_SUBPARTS.
9840 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9841 as the number of units, setting it from VECTOR_CST_NELTS.
9842 (const_vector_from_tree): Likewise.
9843 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9844 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9845 (fold_negate_expr_1): Likewise.
9846 (fold_convert_const): Likewise.
9847 (const_binop): Likewise. Differentiate the number of output and
9849 (const_unop): Likewise.
9850 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9851 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9852 in cases that did the opposite.
9854 2017-09-14 Richard Biener <rguenther@suse.de>
9856 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9859 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9861 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9863 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9866 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9867 if and where to split a bb, except for splitting before debug insn
9868 sequences followed by non-label real insn. Delete debug insns
9869 in between basic blocks.
9871 * combine.c (make_compound_operation_int): Formatting fixes.
9873 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9874 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9875 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9876 * config/sol2.h (LINK_EH_SPEC): Likewise.
9877 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9878 * config/s390/linux.h (LINK_SPEC): Likewise.
9879 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9880 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9881 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9882 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9883 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9884 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9885 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9886 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9888 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9890 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9892 (ENDFILE_LINUX_SPEC): Likewise.
9893 (LINK_EH_SPEC): Likewise.
9894 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9895 (LINK_OS_LINUX_SPEC32): Likewise.
9896 (LINK_OS_LINUX_SPEC64): Likewise.
9897 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9898 (LINK_OS_LINUX_SPEC): Likewise.
9900 2017-09-13 Martin Liska <mliska@suse.cz>
9903 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9904 CASE_HIGH is NULL_TREE.
9906 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9907 Alan Hayward <alan.hayward@arm.com>
9908 David Sherwood <david.sherwood@arm.com>
9910 * target.def (secondary_memory_needed): New hook.
9911 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9912 instead of SECONDARY_MEMORY_NEEDED.
9913 (secondary_memory_needed_mode): Likewise.
9914 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9915 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9916 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9917 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9918 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9919 * doc/tm.texi: Regenerate.
9920 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9921 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9922 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9923 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9924 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9925 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9926 mode argument first and change the reg_class arguments to reg_class_t.
9927 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9928 Make static. Update the call to inline_secondary_memory_needed.
9929 (ix86_register_move_cost): Update the call to
9930 inline_secondary_memory_needed.
9931 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9932 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9934 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9935 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9937 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9938 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9939 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9940 and match hook interface. Add comment from mips.h.
9941 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9942 * config/mmix/mmix.md (truncdfsf2): Refer to
9943 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9945 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9946 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9947 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9948 (pa_secondary_memory_needed): New function.
9949 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9950 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9951 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9952 (pdp11_secondary_memory_needed): Make static and match hook interface.
9953 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9954 * config/powerpcspe/powerpcspe-protos.h
9955 (rs6000_secondary_memory_needed_ptr): Delete.
9956 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9958 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9959 (rs6000_option_override_internal): Assign to
9960 targetm.secondary_memory_needed rather than
9961 rs6000_secondary_memory_needed_ptr.
9962 (rs6000_secondary_memory_needed): Match hook interface.
9963 (rs6000_debug_secondary_memory_needed): Likewise.
9964 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9965 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9966 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9967 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9968 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9969 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9971 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9972 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9973 (rs6000_option_override_internal): Assign to
9974 targetm.secondary_memory_needed rather than
9975 rs6000_secondary_memory_needed_ptr.
9976 (rs6000_secondary_memory_needed): Match hook interface.
9977 (rs6000_debug_secondary_memory_needed): Likewise.
9978 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9979 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9980 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9981 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9982 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9983 (sparc_secondary_memory_needed): New function.
9984 * lra-constraints.c (check_and_process_move): Refer to
9985 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9987 (curr_insn_transform): Likewise.
9988 (process_alt_operands): Use targetm.secondary_memory_needed
9989 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9990 (check_secondary_memory_needed_p): Likewise.
9991 (choose_split_class): Likewise.
9992 * reload.c: Unconditionally include code that was previously
9993 conditional on SECONDARY_MEMORY_NEEDED.
9994 (push_secondary_reload): Use targetm.secondary_memory_needed
9995 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9996 (push_reload): Likewise.
9997 * reload1.c: Unconditionally include code that was previously
9998 conditional on SECONDARY_MEMORY_NEEDED.
9999 (choose_reload_regs): Use targetm.secondary_memory_needed
10000 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10001 (gen_reload): Likewise.
10002 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10004 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10005 Alan Hayward <alan.hayward@arm.com>
10006 David Sherwood <david.sherwood@arm.com>
10008 * target.def (secondary_memory_needed_mode): New hook:
10009 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10010 * targhooks.h (default_secondary_memory_needed_mode): New function.
10011 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10012 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10013 * doc/tm.texi: Regenerate.
10014 * lra-constraints.c (check_and_process_move): Use
10015 targetm.secondary_memory_needed_mode instead of
10016 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10017 (curr_insn_transform): Likewise.
10018 * reload.c (get_secondary_mem): Likewise.
10019 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10020 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10022 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10023 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10024 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10025 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10026 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10028 * config/powerpcspe/powerpcspe-protos.h
10029 (rs6000_secondary_memory_needed_mode): Delete.
10030 * config/powerpcspe/powerpcspe.c
10031 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10032 (rs6000_secondary_memory_needed_mode): Make static.
10033 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10034 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10036 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10038 (rs6000_secondary_memory_needed_mode): Make static.
10039 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10040 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10041 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10042 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10043 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10045 (sparc_secondary_memory_needed_mode): New function.
10046 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10048 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10050 * config/aarch64/constraints.md (Umq): New constraint.
10051 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10053 (mov<mode>): Update condition.
10055 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10057 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10058 when bitposition is the same.
10060 2017-09-13 Richard Biener <rguenther@suse.de>
10062 * dwarf2out.c (output_die_symbol): Remove.
10063 (output_die): Do not output a DIEs symbol.
10065 2017-09-13 Richard Biener <rguenther@suse.de>
10067 PR middle-end/82128
10068 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10069 default-def to avoid breaking iterator update with the weird
10070 interaction with cgraph_update_edges_for_call_stmt_node.
10072 2017-09-13 Richard Biener <rguenther@suse.de>
10074 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10075 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10076 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10077 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10079 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10081 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10082 Disable pc relative literal load irrespective of
10083 TARGET_FIX_ERR_A53_84341 for default.
10085 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10087 * config/sparc/sparc.c (output_return): Output the source location of
10088 the insn in the delay slot, if any.
10089 (output_sibcall): Likewise.
10091 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10094 * common.opt (-static-pie): New alias.
10095 (shared): Negate static-pie.
10096 (-no-pie): Update help text.
10098 (static-pie): New option.
10099 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10100 -static-pie support.
10101 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10102 (LINK_EH_SPEC): Likewise.
10103 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10104 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10105 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10106 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10107 (init_gcc_specs): Likewise.
10108 (init_spec): Likewise.
10109 (display_help): Update help message for -pie.
10110 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10113 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10115 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10116 (movdi_aarch64): Likewise.
10117 (movti_aarch64): Likewise.
10119 2017-09-12 Simon Wright <simon@pushface.org>
10122 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10123 calculation of the minor version, always output as 0.
10125 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10128 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10129 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10130 on it early, rather than manual conversion late. For
10131 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10132 instead of performing manual conversion.
10134 2017-09-12 Carl Love <cel@us.ibm.com>
10136 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10137 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10139 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10140 VMULOSW): Add definitions.
10141 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10142 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10143 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10144 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10145 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10147 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10149 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10151 (movti_aarch64): Likewise.
10152 (movdf_aarch64): Likewise.
10153 (movtf_aarch64): Likewise.
10154 (load_pairdi): Likewise.
10155 (store_pairdi): Likewise.
10156 (load_pairdf): Likewise.
10157 (store_pairdf): Likewise.
10158 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10159 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10160 (ldr_got_small_<mode>): Likewise.
10161 (ldr_got_small_28k_<mode>): Likewise.
10162 (ldr_got_tiny): Likewise.
10163 * config/aarch64/iterators.md (ldst_sz): New.
10164 (ldpstp_sz): Likewise.
10165 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10167 (thunderx_load): Split load_8 to load_16.
10168 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10170 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10171 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10172 (xgene1_store_pair): Split store_8 to store_16.
10173 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10174 (falkor_st_0_st_sd): Split store_8 to store_16.
10176 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10178 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10179 and store1/2/3/4 to store_4/8/12/16.
10180 * config/aarch64/aarch64.md: Update for rename.
10181 * config/arm/arm.md: Likewise.: Likewise.
10182 * config/arm/arm.c: Likewise.
10183 * config/arm/thumb1.md: Likewise.
10184 * config/arm/thumb2.md: Likewise.
10185 * config/arm/vfp.md: Likewise.
10186 * config/arm/arm-generic.md: Likewise.
10187 * config/arm/arm1020e.md: Likewise.
10188 * config/arm/arm1026ejs.md: Likewise.
10189 * config/arm/arm1136jfs.md: Likewise.
10190 * config/arm/arm926ejs.md: Likewise.
10191 * config/arm/cortex-a15.md: Likewise.
10192 * config/arm/cortex-a17.md: Likewise.
10193 * config/arm/cortex-a5.md: Likewise.
10194 * config/arm/cortex-a53.md: Likewise.
10195 * config/arm/cortex-a57.md: Likewise.
10196 * config/arm/cortex-a7.md: Likewise.
10197 * config/arm/cortex-a8.md: Likewise.
10198 * config/arm/cortex-a9.md: Likewise.
10199 * config/arm/cortex-m4.md: Likewise.
10200 * config/arm/cortex-m7.md: Likewise.
10201 * config/arm/cortex-r4.md: Likewise.
10202 * config/arm/exynos-m1.md: Likewise.
10203 * config/arm/fa526.md: Likewise.
10204 * config/arm/fa606te.md: Likewise.
10205 * config/arm/fa626te.md: Likewise.
10206 * config/arm/fa726te.md: Likewise.
10207 * config/arm/fmp626.md: Likewise.
10208 * config/arm/iwmmxt.md: Likewise.
10209 * config/arm/ldmstm.md: Likewise.
10210 * config/arm/marvell-pj4.md: Likewise.
10211 * config/arm/xgene1.md: Likewise.
10212 * config/aarch64/thunderx.md: Likewise.
10213 * config/aarch64/thunderx2t99.md: Likewise.
10214 * config/aarch64/falkor.md: Likewise.
10216 2017-09-12 Martin Liska <mliska@suse.cz>
10218 * attribs.c (private_lookup_attribute): New function.
10219 * attribs.h (private_lookup_attribute): Declared here.
10220 (lookup_attribute): Called from this place.
10222 2017-09-12 Richard Biener <rguenther@suse.de>
10224 PR tree-optimization/82157
10225 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10226 stmts with side-effects.
10228 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10229 Alan Hayward <alan.hayward@arm.com>
10230 David Sherwood <david.sherwood@arm.com>
10232 * target.def (hard_regno_nregs): New hook.
10233 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10234 * targhooks.h (default_hard_regno_nregs): Declare.
10235 * targhooks.c (default_hard_regno_nregs): New function.
10236 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10237 (TARGET_HARD_REGNO_NREGS): ...this hook.
10238 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10239 (CLASS_MAX_NREGS): Likewise.
10240 * doc/tm.texi: Regenerate.
10241 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10242 instead of HARD_REGNO_NREGS.
10243 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10244 HARD_REGNO_NREGS in the comment.
10245 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10246 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10247 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10248 Return an unsigned int.
10249 (TARGET_HARD_REGNO_NREGS): Redefine.
10250 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10251 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10252 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10253 (arc_hard_regno_nregs): New function.
10254 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10255 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10256 (arm_hard_regno_nregs): New function.
10257 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10258 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10259 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10260 (TARGET_HARD_REGNO_NREGS): Redefine.
10261 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10262 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10263 (HARD_REGNO_NREGS): Delete.
10264 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10265 (cr16_hard_regno_nregs): New function.
10266 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10267 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10268 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10269 (cris_hard_regno_nregs): New function.
10270 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10271 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10272 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10273 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10274 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10275 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10276 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10277 (frv_hard_regno_nregs): Make static. Take and return an
10279 (frv_class_max_nregs): Remove outdated copy of documentation.
10280 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10281 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10282 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10283 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10284 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10285 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10286 (TARGET_HARD_REGNO_NREGS): Redefine.
10287 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10288 (CLASS_MAX_NREGS): Update comment.
10289 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10290 (ia64_hard_regno_nregs): New function.
10291 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10292 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10293 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10294 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10295 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10297 (m32c_hard_regno_nregs): Likewise. Make static.
10298 (TARGET_HARD_REGNO_NREGS): Redefine.
10299 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10300 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10301 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10302 (m68k_hard_regno_nregs): New function.
10303 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10304 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10305 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10306 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10307 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10308 Take and return an unsigned int.
10309 (TARGET_HARD_REGNO_NREGS): Redefine.
10310 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10311 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10312 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10313 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10314 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10315 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10316 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10317 (msp430_hard_regno_nregs): Make static. Take and return an
10319 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10320 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10321 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10322 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10323 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10324 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10325 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10326 (TARGET_HARD_REGNO_NREGS): Redefine.
10327 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10328 (PA_HARD_REGNO_NREGS): ...this.
10329 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10330 (PA_HARD_REGNO_NREGS): ...this.
10331 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10332 (pa_hard_regno_nregs): New function.
10333 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10334 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10335 (pdp11_hard_regno_nregs): New function.
10336 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10337 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10338 (rs6000_hard_regno_nregs_hook): New function.
10339 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10340 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10341 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10342 Take and return an unsigned int. Move earlier in file.
10343 (TARGET_HARD_REGNO_NREGS): Redefine.
10344 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10345 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10346 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10347 (rl78_hard_regno_nregs): Make static. Take and return an
10349 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10350 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10351 (rs6000_hard_regno_nregs_hook): New function.
10352 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10353 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10354 (TARGET_HARD_REGNO_NREGS): Redefine.
10355 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10356 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10357 instead of HARD_REGNO_NREGS.
10358 (s390_hard_regno_nregs): New function.
10359 (s390_hard_regno_mode_ok): Add comment from s390.h.
10360 (TARGET_HARD_REGNO_NREGS): Redefine.
10361 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10362 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10363 (sh_hard_regno_nregs): New function.
10364 (sh_pass_in_reg_p): Use it.
10365 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10366 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10367 (sparc_hard_regno_nregs): New function.
10368 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10369 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10370 (spu_function_arg_advance): Use it, supplying a valid register number.
10371 (TARGET_HARD_REGNO_NREGS): Redefine.
10372 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10373 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10374 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10375 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10376 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10377 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10378 (CLASS_MAX_NREGS): Remove copy of old documentation.
10379 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10380 (visium_hard_regno_nregs): New function.
10381 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10382 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10383 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10384 xtensa_hard_regno_nregs): New function.
10385 * system.h (HARD_REGNO_NREGS): Poison.
10387 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10389 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10390 hard_regno_nregs instead of HARD_REGNO_NREGS.
10391 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10392 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10393 (c6x_expand_epilogue): Likewise.
10394 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10395 (frv_read_iacc_argument): Likewise.
10396 * config/sh/sh.c: Include regs.h.
10397 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10398 (regs_used): Likewise.
10399 (output_stack_adjust): Likewise.
10400 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10401 * expmed.c: Include regs.h.
10402 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10403 * ree.c: Include regs.h.
10404 (combine_reaching_defs): Use hard_regno_nregs instead of
10406 (add_removable_extension): Likewise.
10408 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10410 * regs.h (hard_regno_nregs): Turn into a function.
10411 (end_hard_regno): Update accordingly.
10412 * caller-save.c (setup_save_areas): Likewise.
10413 (save_call_clobbered_regs): Likewise.
10414 (replace_reg_with_saved_mem): Likewise.
10415 (insert_restore): Likewise.
10416 (insert_save): Likewise.
10417 * combine.c (can_change_dest_mode): Likewise.
10418 (move_deaths): Likewise.
10419 (distribute_notes): Likewise.
10420 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10421 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10422 (rs6000_split_multireg_move): Likewise.
10423 (rs6000_register_move_cost): Likewise.
10424 (rs6000_memory_move_cost): Likewise.
10425 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10426 (rs6000_split_multireg_move): Likewise.
10427 (rs6000_register_move_cost): Likewise.
10428 (rs6000_memory_move_cost): Likewise.
10429 * cselib.c (cselib_reset_table): Likewise.
10430 (cselib_lookup_1): Likewise.
10431 * emit-rtl.c (set_mode_and_regno): Likewise.
10432 * function.c (aggregate_value_p): Likewise.
10433 * ira-color.c (setup_profitable_hard_regs): Likewise.
10434 (check_hard_reg_p): Likewise.
10435 (calculate_saved_nregs): Likewise.
10436 (assign_hard_reg): Likewise.
10437 (improve_allocation): Likewise.
10438 (calculate_spill_cost): Likewise.
10439 * ira-emit.c (modify_move_list): Likewise.
10440 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10441 (ira_hard_reg_in_set_p): Likewise.
10442 * ira.c (setup_reg_mode_hard_regset): Likewise.
10443 (clarify_prohibited_class_mode_regs): Likewise.
10444 (check_allocation): Likewise.
10445 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10446 (lra_setup_reg_renumber): Likewise.
10447 (setup_try_hard_regno_pseudos): Likewise.
10448 (spill_for): Likewise.
10449 (assign_hard_regno): Likewise.
10450 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10451 * lra-constraints.c (in_class_p): Likewise.
10452 (lra_constraint_offset): Likewise.
10453 (simplify_operand_subreg): Likewise.
10454 (lra_constraints): Likewise.
10455 (split_reg): Likewise.
10456 (split_if_necessary): Likewise.
10457 (invariant_p): Likewise.
10458 (inherit_in_ebb): Likewise.
10459 * lra-lives.c (process_bb_lives): Likewise.
10460 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
10461 (get_hard_regs): Likewise.
10462 (do_remat): Likewise.
10463 * lra-spills.c (assign_spill_hard_regs): Likewise.
10464 * mode-switching.c (create_pre_exit): Likewise.
10465 * postreload.c (reload_combine_recognize_pattern): Likewise.
10466 * recog.c (peep2_find_free_register): Likewise.
10467 * regcprop.c (kill_value_regno): Likewise.
10468 (set_value_regno): Likewise.
10469 (copy_value): Likewise.
10470 (maybe_mode_change): Likewise.
10471 (find_oldest_value_reg): Likewise.
10472 (copyprop_hardreg_forward_1): Likewise.
10473 * regrename.c (check_new_reg_p): Likewise.
10474 (regrename_do_replace): Likewise.
10475 * reload.c (push_reload): Likewise.
10476 (combine_reloads): Likewise.
10477 (find_dummy_reload): Likewise.
10478 (operands_match_p): Likewise.
10479 (find_reloads): Likewise.
10480 (find_equiv_reg): Likewise.
10481 (reload_adjust_reg_for_mode): Likewise.
10482 * reload1.c (count_pseudo): Likewise.
10483 (count_spilled_pseudo): Likewise.
10484 (find_reg): Likewise.
10485 (clear_reload_reg_in_use): Likewise.
10486 (free_for_value_p): Likewise.
10487 (allocate_reload_reg): Likewise.
10488 (choose_reload_regs): Likewise.
10489 (reload_adjust_reg_for_temp): Likewise.
10490 (emit_reload_insns): Likewise.
10491 (delete_output_reload): Likewise.
10492 * rtlanal.c (subreg_get_info): Likewise.
10493 * sched-deps.c (sched_analyze_reg): Likewise.
10494 * sel-sched.c (init_regs_for_mode): Likewise.
10495 (mark_unavailable_hard_regs): Likewise.
10496 (choose_best_reg_1): Likewise.
10497 (verify_target_availability): Likewise.
10498 * valtrack.c (dead_debug_insert_temp): Likewise.
10499 * var-tracking.c (track_loc_p): Likewise.
10500 (emit_note_insn_var_location): Likewise.
10501 * varasm.c (make_decl_rtl): Likewise.
10502 * reginfo.c (choose_hard_reg_mode): Likewise.
10503 (init_reg_modes_target): Refer directly to
10504 this_target_regs->x_hard_regno_nregs.
10506 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10508 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10509 instead of hard_regno_nregs.
10511 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10513 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10514 end_hard_regno instead of hard_regno_nregs.
10515 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10516 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10517 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10518 * ira-color.c (improve_allocation): Likewise.
10519 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10520 * lra-lives.c (mark_regno_live): Likewise.
10521 (mark_regno_dead): Likewise.
10522 * lra-remat.c (operand_to_remat): Likewise.
10523 * lra.c (collect_non_operand_hard_regs): Likewise.
10524 * postreload.c (reload_combine_note_store): Likewise.
10525 (move2add_valid_value_p): Likewise.
10526 * reload.c (regno_clobbered_p): Likewise.
10528 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10530 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10532 * config/v850/v850.c (v850_reorg): Likewise.
10533 * reload.c (refers_to_regno_for_reload_p): Likewise.
10534 (find_equiv_reg): Likewise.
10535 * reload1.c (reload_reg_reaches_end_p): Likewise.
10537 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10539 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10541 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10542 * config/arm/arm.c (output_move_neon): Likewise.
10543 (arm_attr_length_move_neon): Likewise.
10544 (neon_split_vcombine): Likewise.
10545 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10546 (c6x_mark_reg_written): Likewise.
10547 (c6x_dwarf_register_span): Likewise.
10548 * config/i386/i386.c (ix86_save_reg): Likewise.
10549 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10550 (rws_access_reg): Likewise.
10551 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10552 * mode-switching.c (create_pre_exit): Likewise.
10553 * ree.c (combine_reaching_defs): Likewise.
10554 (add_removable_extension): Likewise.
10555 * regcprop.c (find_oldest_value_reg): Likewise.
10556 (copyprop_hardreg_forward_1): Likewise.
10557 * reload.c (reload_inner_reg_of_subreg): Likewise.
10558 (push_reload): Likewise.
10559 (combine_reloads): Likewise.
10560 (find_dummy_reload): Likewise.
10561 (reload_adjust_reg_for_mode): Likewise.
10562 * reload1.c (find_reload_regs): Likewise.
10563 (forget_old_reloads_1): Likewise.
10564 (reload_reg_free_for_value_p): Likewise.
10565 (reload_adjust_reg_for_temp): Likewise.
10566 (emit_reload_insns): Likewise.
10567 (delete_output_reload): Likewise.
10568 * sel-sched.c (choose_best_reg_1): Likewise.
10569 (choose_best_pseudo_reg): Likewise.
10571 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10572 Alan Hayward <alan.hayward@arm.com>
10573 David Sherwood <david.sherwood@arm.com>
10575 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10576 * target.def (slow_unaligned_access): New hook.
10577 * targhooks.h (default_slow_unaligned_access): Declare.
10578 * targhooks.c (default_slow_unaligned_access): New function.
10579 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10580 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10581 * doc/tm.texi: Regenerate.
10582 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10583 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10584 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10586 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10587 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10589 (rs6000_slow_unaligned_access): New function.
10590 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10591 (expand_block_compare): Likewise.
10592 (expand_strn_compare): Likewise.
10593 (rs6000_rtx_costs): Likewise.
10594 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10595 (riscv_slow_unaligned_access): Likewise.
10596 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10597 (riscv_slow_unaligned_access_p): ...this and make static.
10598 (riscv_option_override): Update accordingly.
10599 (riscv_slow_unaligned_access): New function.
10600 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10601 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10602 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10603 (rs6000_slow_unaligned_access): New function.
10604 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10605 (rs6000_rtx_costs): Likewise.
10606 * config/rs6000/rs6000-string.c (expand_block_compare)
10607 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10608 of SLOW_UNALIGNED_ACCESS.
10609 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10610 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10611 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10612 of SLOW_UNALIGNED_ACCESS.
10613 * expmed.c (simple_mem_bitfield_p): Likewise.
10614 * expr.c (alignment_for_piecewise_move): Likewise.
10615 (emit_group_load_1): Likewise.
10616 (emit_group_store): Likewise.
10617 (copy_blkmode_from_reg): Likewise.
10618 (emit_push_insn): Likewise.
10619 (expand_assignment): Likewise.
10620 (store_field): Likewise.
10621 (expand_expr_real_1): Likewise.
10622 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10623 * lra-constraints.c (simplify_operand_subreg): Likewise.
10624 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10625 * gimple-ssa-store-merging.c: Likewise in block comment at start
10627 * tree-ssa-strlen.c: Include target.h.
10628 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10629 of SLOW_UNALIGNED_ACCESS.
10630 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10632 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10634 PR rtl-optimization/82185
10635 * expmed.c (emit_store_flag_int): Only test tem if it has been
10638 2017-09-12 Richard Biener <rguenther@suse.de>
10640 PR middle-end/82149
10641 * match.pd ((FTYPE) N CMP CST): Fix typo.
10643 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10645 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10647 (mips_near_type_p): Add 'short_call' attribute as a synonym
10649 * doc/extend.texi (short_call): Document new function attribute.
10651 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10654 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10655 assertion check that in the condition.
10656 (get_atomic_generic_size): Likewise. Before testing if parameter
10657 has pointer type, if it has array type, call for C++
10658 default_conversion to perform array-to-pointer conversion.
10660 2017-09-12 Richard Biener <rguenther@suse.de>
10662 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10663 for operations we cannot scalarize.
10665 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10667 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10668 vectors heap vectors. Clean up comments.
10669 Make visited_bbs a reference.
10670 (profitable_jump_thread_path): Make GC
10671 vectors heap vectors. Clean up comments.
10673 (convert_and_register_jump_thread_path): Make GC vectors heap
10675 (check_subpath_and_update_thread_path): Same. Clean up comments.
10676 Make visited_bbs a reference.
10677 (handle_phi): Abstract common code to to
10678 register_jump_thread_path_if_profitable.
10679 Rename VAR_BB to DEF_BB.
10681 Make GC vectors heap vectors.
10682 Make visited_bbs a reference.
10683 (handle_assignment): Same.
10684 (register_jump_thread_path_if_profitable): New.
10685 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10687 Make GC vectors heap vectors. Clean up comments.
10688 Make visited_bbs a reference.
10689 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10690 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10693 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10696 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10697 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10699 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10701 Revert r251800 and r251799.
10703 2017-09-11 Martin Jambor <mjambor@suse.cz>
10706 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10707 arguments in advance.
10708 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10709 use it to find predecessor edges.
10710 (naive_outof_ssa): Collect vector of predecessors.
10712 2017-09-08 Jason Merrill <jason@redhat.com>
10714 PR c++/70029 - ICE with ref-qualifier and -flto
10715 * langhooks.h (struct lang_hooks_for_types): Add
10716 copy_lang_qualifiers.
10717 * attribs.c (build_type_attribute_qual_variant): Use it.
10718 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10720 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10721 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10723 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10726 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10727 (*mulsi3_sp64): New instruction.
10728 (mulsi3): New expander.
10730 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10732 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10734 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10736 * sancov.c: Include memmodel.h.
10738 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10741 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10744 2017-09-07 Carl Love <cel@us.ibm.com>
10746 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10747 the sldi instruction.
10749 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10751 * sancov.c: Include tm_p.h.
10753 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10756 * output.h (switch_to_other_text_partition): New declaration.
10757 * varasm.c (switch_to_other_text_partition): New function.
10758 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10759 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10760 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10761 to the other text partition before emitting LCL label and switch back
10762 after emitting the word after it.
10764 2017-09-07 Richard Biener <rguenther@suse.de>
10766 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10767 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10768 critical edges here, after loop init.
10769 (pass_data_pre): Remove PROP_no_crit_edges flags.
10770 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10771 for valueization of call args to avoid leaking VN_TOP.
10772 (visit_use): Assert we do not visit default defs.
10773 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10774 Use error_mark_node to more easily detect leaking VN_TOP.
10775 All default-defs are varying, not VN_TOP. Mark them visited.
10776 (run_scc_vn): Make code match comment.
10778 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10780 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10781 OPTION_MASK_FLOAT128_KEYWORD.
10782 (POWERPC_MASKS): Likewise.
10783 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10784 support for the -mfloat128-type option, and make -mfloat128
10785 default on PowerPC Linux systems. Define or undefine
10786 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10787 Define __float128 to be __ieee128 if IEEE 128-bit support is
10788 enabled, or undefine it.
10789 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10790 Delete defining __FLOAT128_TYPE__.
10791 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10792 -mfloat128-type option and make -mfloat128 default on PowerPC
10794 (TARGET_FLOAT128_TYPE): Likewise.
10795 (-mfloat128-type): Likewise.
10796 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10797 Delete the -mfloat128-type option and make -mfloat128 default on
10798 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10799 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10800 128-bit floating point is enabled. Change tests from using
10801 -mfloat128-type to -mfloat128.
10802 (rs6000_mangle_type): Use the correct mangling for the __float128
10803 type even if normal long double is restricted to 64-bits.
10804 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10806 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10807 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10808 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10810 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10811 documentation for -mfloat128.
10813 2017-09-06 Olivier Hainque <hainque@adacore.com>
10815 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10817 2017-09-06 Wish Wu <wishwu007@gmail.com>
10818 Jakub Jelinek <jakub@redhat.com>
10820 * asan.c (initialize_sanitizer_builtins): Add
10821 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10822 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10823 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10824 BT_FN_VOID_UINT64_PTR variables.
10825 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10826 (BT_FN_VOID_UINT16_UINT16): Likewise.
10827 (BT_FN_VOID_UINT32_UINT32): Likewise.
10828 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10829 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10830 (BT_FN_VOID_UINT64_PTR): Likewise.
10831 * common.opt (flag_sanitize_coverage): New variable.
10832 (fsanitize-coverage=trace-pc): Remove.
10833 (fsanitize-coverage=): Add.
10834 * flag-types.h (enum sanitize_coverage_code): New enum.
10835 * fold-const.c (fold_range_test): Disable non-short-circuit
10836 optimization if flag_sanitize_coverage.
10837 (fold_truth_andor): Likewise.
10838 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10839 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10840 (coverage_sanitizer_opts): New array.
10841 (get_closest_sanitizer_option): Add OPTS argument, handle also
10842 OPT_fsanitize_coverage_.
10843 (parse_sanitizer_options): Adjusted to also handle
10844 OPT_fsanitize_coverage_.
10845 (common_handle_option): Add OPT_fsanitize_coverage_.
10846 * sancov.c (instrument_comparison, instrument_switch): New function.
10847 (sancov_pass): Add trace-cmp support.
10848 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10849 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10850 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10851 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10852 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10853 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10854 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10855 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10856 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10857 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10859 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10861 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10862 error. Only quit immediately if parsing is complete.
10863 (BEGIN): Initialize fatal_err and parse_done.
10864 (begin fpu, end fpu): Check number of arguments.
10865 (begin arch, end arch): Likewise.
10866 (begin cpu, end cpu): Likewise.
10867 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10868 (optalias): Likewise.
10870 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10872 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10873 * config/arm/arm-isa.h: Delete. Move definitions to ...
10874 * arm-cpus.in: ... here. Use new feature and fgroup values.
10875 * config/arm/arm.c (arm_option_override): Use lower case for feature
10877 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10878 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10879 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10881 (print_isa_bits_for): New function.
10882 (gen_isa): New function.
10883 (gen_comm_data): Use print_isa_bits_for.
10884 (define feature): New keyword.
10885 (define fgroup): New keyword.
10886 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10887 (arm-isa.h): Add rule to generate file.
10888 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10889 case for feature bit names.
10891 2017-09-06 Richard Biener <rguenther@suse.de>
10893 * tree-ssa-pre.c (NECESSARY): Remove.
10894 (create_expression_by_pieces): Do not touch pass-local flags.
10895 (insert_into_preds_of_block): Likewise.
10896 (do_pre_regular_insertion): Likewise.
10897 (eliminate_insert): Likewise.
10898 (eliminate_dom_walker::before_dom_children): Likewise.
10899 (fini_eliminate): Do not look at inserted_exprs.
10900 (mark_operand_necessary): Remove.
10901 (remove_dead_inserted_code): Replace with simple work-list
10902 algorithm based on inserted_exprs and SSA uses.
10903 (pass_pre::execute): Re-order fini_eliminate and
10904 remove_dead_inserted_code.
10906 2017-09-06 Olivier Hainque <hainque@adacore.com>
10908 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10909 for VxWorks 7. Adjust surrounding comments.
10911 2017-09-06 Richard Biener <rguenther@suse.de>
10913 * gimple-ssa-strength-reduction.c
10914 (find_candidates_dom_walker::before_dom_children): Also allow
10917 2017-09-06 Richard Biener <rguenther@suse.de>
10919 PR tree-optimization/82108
10920 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10921 for gap in the non-permutation SLP case.
10923 2017-09-06 Martin Jambor <mjambor@suse.cz>
10925 PR tree-optimization/82078
10926 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10927 add_access_to_work_queue...
10928 (build_accesses_from_assign): ...here.
10929 (propagate_all_subaccesses): Make sure racc is the group
10930 representative, if there is one.
10932 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10934 PR middle-end/82095
10935 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10938 2017-09-06 Richard Biener <rguenther@suse.de>
10940 * gimple-ssa-strength-reduction.c
10941 (find_candidates_dom_walker::before_doom_children): Use a
10942 type and not a mode check.
10944 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10947 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10949 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10951 2017-09-05 Jeff Law <law@redhat.com>
10953 PR tree-optimization/64910
10954 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10955 cases where we have 3 or more operands.
10957 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10959 PR middle-end/81768
10960 * omp-low.c (lower_omp_for): Recompute tree invariant if
10961 gimple_omp_for_initial/final is ADDR_EXPR.
10963 PR middle-end/81768
10964 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10965 into gimple val before gimplification fo the COND_EXPR.
10967 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10969 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10970 REGION_COPY argument.
10971 (thread_through_all_blocks): Remove unused argument to
10972 duplicate_thread_path.
10974 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10975 Alan Hayward <alan.hayward@arm.com>
10976 David Sherwood <david.sherwood@arm.com>
10978 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10979 Take a scalar_mode rather than a machine_mode.
10980 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10981 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10982 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10983 (aarch64_gen_adjusted_ldpstp): Likewise.
10984 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10986 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10987 Alan Hayward <alan.hayward@arm.com>
10988 David Sherwood <david.sherwood@arm.com>
10990 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10991 Take a scalar_int_mode instead of a machine_mode.
10992 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10993 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10994 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10995 (aarch64_simd_attr_length_rglist): Delete.
10996 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10997 a scalar_int_mode instead of a machine_mode.
10998 (aarch64_add_offset): Likewise.
10999 (aarch64_internal_mov_immediate): Likewise
11000 (aarch64_add_constant_internal): Likewise.
11001 (aarch64_add_constant): Likewise.
11002 (aarch64_movw_imm): Likewise.
11003 (aarch64_rtx_arith_op_extract_p): Likewise.
11004 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11005 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11006 Remove assert that the mode isn't a vector.
11007 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11008 (aarch64_expand_mov_immediate): Update calls after above changes.
11009 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11010 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11011 (aarch64_move_imm): Likewise.
11012 (aarch64_can_const_movi_rtx_p): Likewise.
11013 (aarch64_strip_extend): Likewise.
11014 (aarch64_extr_rtx_p): Likewise.
11015 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11016 a CONST_INT when the mode parameter is VOIDmode.
11017 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11021 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11022 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11023 (bitwise_type_for_mode): Update accordingly.
11025 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11027 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11028 * stor-layout.c (mode_for_size_tree): Likewise.
11029 (mode_for_array): Update accordingly.
11030 (layout_decl): Likewise.
11031 (compute_record_mode): Likewise. Only set the mode once.
11033 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11035 * target.def (get_mask_mode): Change return type to opt_mode.
11037 * doc/tm.texi: Regenerate.
11038 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11039 * targhooks.c (default_get_mask_mode): Likewise.
11040 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11041 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11042 targetm.get_mask_mode.
11043 * tree.c (build_truth_vector_type): Likewise.
11045 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11047 * machmode.h (mode_for_vector): Return an opt_mode.
11048 * stor-layout.c (mode_for_vector): Likewise.
11049 (mode_for_int_vector): Update accordingly.
11050 (layout_type): Likewise.
11051 * config/i386/i386.c (emit_memmov): Likewise.
11052 (ix86_expand_set_or_movmem): Likewise.
11053 (ix86_expand_vector_init): Likewise.
11054 (ix86_get_mask_mode): Likewise.
11055 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11057 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11058 * expmed.c (extract_bit_field_1): Likewise.
11059 * expr.c (expand_expr_real_2): Likewise.
11060 * optabs-query.c (can_vec_perm_p): Likewise.
11061 (can_vec_mask_load_store_p): Likewise.
11062 * optabs.c (expand_vec_perm): Likewise.
11063 * targhooks.c (default_get_mask_mode): Likewise.
11064 * tree-vect-stmts.c (vectorizable_store): Likewise.
11065 (vectorizable_load): Likewise.
11066 (get_vectype_for_scalar_type_and_size): Likewise.
11068 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11070 * machmode.h (mode_for_int_vector): New function.
11071 * stor-layout.c (mode_for_int_vector): Likewise.
11072 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11073 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11074 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11075 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11076 (s390_expand_vcond): Likewise.
11078 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11080 * machmode.h (opt_machine_mode): New type.
11081 (opt_mode<T>): Allow construction from anything that can be
11083 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11084 (mode_for_size): Return an opt_machine_mode.
11085 * stor-layout.c (mode_for_size): Likewise.
11086 (mode_for_size_tree): Update call accordingly.
11087 (bitwise_mode_for_mode): Likewise.
11088 (make_fract_type): Likewise.
11089 (make_accum_type): Likewise.
11090 * caller-save.c (replace_reg_with_saved_mem): Update call
11092 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11093 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11094 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11095 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11096 * expmed.c (extract_bit_field_1): Likewise.
11097 * reload.c (get_secondary_mem): Likewise.
11098 * varasm.c (assemble_integer): Likewise.
11099 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11104 * machmode.h (decimal_float_mode_for_size): New function.
11105 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11106 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11107 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11108 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11109 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11110 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11112 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11114 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11115 (get_builtin_sync_mode): Likewise.
11116 (expand_ifn_atomic_compare_exchange): Likewise.
11117 (expand_builtin_atomic_clear): Likewise.
11118 (expand_builtin_atomic_test_and_set): Likewise.
11119 (fold_builtin_atomic_always_lock_free): Likewise.
11120 * calls.c (compute_argument_addresses): Likewise.
11121 (emit_library_call_value_1): Likewise.
11122 (store_one_arg): Likewise.
11123 * combine.c (combine_instructions): Likewise.
11124 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11125 * config/arm/arm.c (arm_function_value): Likewise.
11126 (aapcs_allocate_return_reg): Likewise.
11127 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11128 * config/i386/i386.c (construct_container): Likewise.
11129 (ix86_gimplify_va_arg): Likewise.
11130 (ix86_expand_sse_cmp): Likewise.
11131 (emit_memmov): Likewise.
11132 (emit_memset): Likewise.
11133 (expand_small_movmem_or_setmem): Likewise.
11134 (ix86_expand_pextr): Likewise.
11135 (ix86_expand_pinsr): Likewise.
11136 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11137 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11139 * config/mips/mips.c (mips_function_value_1) Likewise.
11140 (mips_block_move_straight): Likewise.
11141 (mips_expand_ins_as_unaligned_store): Likewise.
11142 * config/powerpcspe/powerpcspe.c
11143 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11144 (rs6000_darwin64_record_arg_flush): Likewise.
11145 * config/rs6000/rs6000.c
11146 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11147 (rs6000_darwin64_record_arg_flush): Likewise.
11148 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11149 (sparc_function_value_1): Likewise.
11150 * config/spu/spu.c (adjust_operand): Likewise.
11151 (spu_emit_branch_or_set): Likewise.
11152 (arith_immediate_p): Likewise.
11153 * emit-rtl.c (gen_lowpart_common): Likewise.
11154 * expr.c (expand_expr_real_1): Likewise.
11155 * function.c (assign_parm_setup_block): Likewise.
11156 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11157 * reload1.c (alter_reg): Likewise.
11158 * stor-layout.c (mode_for_vector): Likewise.
11159 (layout_type): Likewise.
11161 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11163 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11164 (spu_convert_move): Likewise.
11165 * lower-subreg.c (resolve_simple_move): Likewise.
11167 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11170 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11171 define_insn to a define_expand.
11172 (altivec_vsum2sws_direct): New define_insn.
11173 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11175 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11177 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11180 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11185 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11186 realign stack if stack alignment needed is less than incoming
11189 2017-09-05 Marek Polacek <polacek@redhat.com>
11192 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11195 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11197 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11199 2017-09-05 Richard Biener <rguenther@suse.de>
11201 PR tree-optimization/82084
11202 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11204 2017-09-05 Richard Biener <rguenther@suse.de>
11206 PR tree-optimization/82102
11207 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11209 2017-09-05 Martin Liska <mliska@suse.cz>
11211 PR tree-optimization/82032
11212 * tree-cfg.c (generate_range_test): New function.
11213 * tree-cfg.h (generate_range_test): Declared here.
11214 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11215 (cleanup_control_expr_graph): Use it.
11216 * tree-switch-conversion.c (try_switch_expansion): Remove
11218 (emit_case_nodes): Use generate_range_test.
11220 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11223 * config/i386/i386.md (*<btsc><mode>_mask): Add
11224 TARGET_USE_BT to insn constraint.
11225 (*btr<mode>_mask): Ditto.
11227 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11229 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11230 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11232 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11235 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11236 TARGET_NEON and TARGET_IWMMXT.
11237 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11238 TARGET_NEON and TARGET_IWMMXT.
11239 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11241 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11243 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11244 (ix86_rewrite_tls_address): Ditto.
11245 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11246 (ix86_rewrite_tls_address_1): Ditto.
11247 (ix86_rewrite_tls_address): Ditto.
11248 * config/i386/predicates.md (tls_address_pattern): New predicate.
11249 * config/i386/i386.md (TLS address splitter): New splitter.
11251 2017-09-04 Richard Biener <rguenther@suse.de>
11253 PR tree-optimization/82084
11254 * fold-const.h (can_native_encode_string_p): Declare.
11255 * fold-const.c (can_native_encode_string_p): Factor out from ...
11256 (native_encode_string): ... here.
11257 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11258 vectorizing stores from constants we later cannot handle.
11260 2017-09-04 Marek Polacek <polacek@redhat.com>
11263 * doc/invoke.texi: Update -Wtautological-compare documentation.
11265 2017-09-04 Jeff Law <law@redhat.com>
11267 PR tree-optimization/64910
11268 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11269 swap the first and last operand if the last is a constant.
11271 2017-09-04 Marek Polacek <polacek@redhat.com>
11274 * convert.c (do_narrow): When sanitizing signed integer overflows,
11275 bail out for signed types.
11276 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11278 2017-09-04 Richard Biener <rguenther@suse.de>
11280 PR tree-optimization/82060
11281 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11282 Move devirtualization after stmt folding and before EH/AB/noreturn
11283 cleanup to get the stmt refs canonicalized. Use a bool instead
11284 of gimple_modified_p since that doesn't work for NOPs. Schedule
11285 NOPs generated by folding for removal.
11287 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11288 Alan Hayward <alan.hayward@arm.com>
11289 David Sherwood <david.sherwood@arm.com>
11291 * coretypes.h (pad_direction): New enum.
11292 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11293 (FUNCTION_ARG_PADDING): Likewise.
11294 * target.def (function_arg_padding): New hook.
11295 * targhooks.h (default_function_arg_padding): Declare.
11296 * targhooks.c (default_function_arg_padding): New function.
11297 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11298 (TARGET_FUNCTION_ARG_PADDING): ...this.
11299 * doc/tm.texi: Regenerate.
11300 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11301 instead of direction.
11302 (compute_argument_addresses): Likewise.
11303 (load_register_parameters): Likewise.
11304 (emit_library_call_value_1): Likewise.
11305 (store_one_arg): Use targetm.calls.function_arg_padding instead
11306 of FUNCTION_ARG_PADDING.
11307 (must_pass_in_stack_var_size_or_pad): Likewise.
11308 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11309 (emit_group_store): Likewise.
11310 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11311 instead of FUNCTION_ARG_PADDING.
11312 (emit_push_insn): Likewise, and propagate enum change throughout
11314 * function.h (direction): Delete.
11315 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11317 * function.c (assign_parm_find_stack_rtl): Likewise.
11318 (assign_parm_setup_block_p): Likewise.
11319 (assign_parm_setup_block): Likewise.
11320 (gimplify_parameters): Likewise.
11321 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11322 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11324 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11325 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11326 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11327 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11328 (aarch64_function_arg_padding): ...this new function.
11329 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11330 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11331 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11332 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11333 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11334 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11335 (arm_pad_arg_upward): Replace with...
11336 (arm_function_arg_padding): ...this new function.
11337 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11339 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11340 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11341 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11342 (ia64_hpux_function_arg_padding): Replace with...
11343 (ia64_function_arg_padding): ...this new function. Use pad_direction
11344 instead of direction. Check for TARGET_HPUX.
11345 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11346 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11347 (iq2000_function_arg_padding): New function.
11348 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11349 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11350 (mips_function_arg_padding): ...this new function.
11351 (mips_pad_reg_upward): Update accordingly.
11352 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11353 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11354 targetm.calls.function_arg_padding.
11355 (FUNCTION_ARG_PADDING): Delete.
11356 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11357 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11358 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11359 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11360 (nios2_block_reg_padding): Return pad_direction instead of direction.
11361 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11362 instead of direction.
11363 (nios2_function_arg_padding): Likewise. Make static.
11364 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11365 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11366 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11367 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11368 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11369 (pa_function_arg_padding): Make static. Return pad_direction instead
11371 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11372 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11373 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11374 instead of direction. Use targetm.calls.function_arg_padding.
11375 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11376 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11377 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11378 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11379 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11381 (function_arg_padding): Rename to...
11382 (rs6000_function_arg_padding): ...this. Make static. Return
11383 pad_direction instead of direction.
11384 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11385 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11386 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11387 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11388 instead of direction. Use targetm.calls.function_arg_padding.
11389 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11390 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11391 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11392 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11393 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11394 (function_arg_padding): Rename to...
11395 (rs6000_function_arg_padding): ...this. Make static. Return
11396 pad_direction instead of direction.
11397 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11398 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11399 * config/s390/s390.c (s390_function_arg_padding): New function.
11400 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11401 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11402 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11403 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11404 (function_arg_padding): Rename to...
11405 (sparc_function_arg_padding): ...this. Make static. Return
11406 pad_direction instead of direction.
11407 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11408 * config/spu/spu.c (spu_function_arg_padding): New function.
11409 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11410 * system.h (FUNCTION_ARG_PADDING): Poison.
11412 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11413 Alan Hayward <alan.hayward@arm.com>
11414 David Sherwood <david.sherwood@arm.com>
11416 * target.def (modes_tieable_p): New hook.
11417 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11418 (TARGET_MODES_TIEABLE_P): ...this.
11419 * doc/tm.texi.in: Regenerate.
11420 * hooks.h (hook_bool_mode_mode_true): Declare.
11421 * hooks.c (hook_bool_mode_mode_true): New function.
11422 * combine.c (subst): Use targetm.modes_tieable_p instead of
11424 * dse.c (find_shift_sequence): Likewise.
11425 * expmed.c (extract_low_bits): Likewise.
11426 * lower-subreg.c: Include target.h.
11427 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11429 * rtlanal.c (rtx_cost): Likewise.
11430 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11431 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11432 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11433 (TARGET_MODES_TIEABLE_P): Redefine.
11434 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11435 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11436 (TARGET_MODES_TIEABLE_P): Redefine.
11437 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11438 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11439 (arc_modes_tieable_p): New function.
11440 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11441 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11442 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11443 (arm_modes_tieable_p): Make static.
11444 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11445 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11446 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11447 (TARGET_MODES_TIEABLE_P): Redefine.
11448 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11449 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11450 (TARGET_MODES_TIEABLE_P): Redefine.
11451 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11452 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11453 (cr16_modes_tieable_p): New function.
11454 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11455 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11456 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11457 (TRULY_NOOP_TRUNCATION): Update comment.
11458 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
11459 (TRULY_NOOP_TRUNCATION): Update comment.
11460 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
11461 (frv_modes_tieable_p): New function.
11462 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
11463 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
11464 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
11465 (TARGET_MODES_TIEABLE_P): Redefine.
11466 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
11467 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
11468 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
11469 (TARGET_MODES_TIEABLE_P): Redefine.
11470 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
11471 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
11472 (ia64_modes_tieable_p): New function.
11473 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
11474 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
11475 (iq2000_modes_tieable_p): New function.
11476 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
11477 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
11478 (lm32_modes_tieable_p): New function.
11479 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
11480 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
11481 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
11482 (TARGET_MODES_TIEABLE_P): Redefine.
11483 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
11484 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
11485 (m32r_modes_tieable_p): New function.
11486 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
11487 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
11488 (m68k_modes_tieable_p): New function.
11489 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
11490 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
11491 (mcore_modes_tieable_p): New function.
11492 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11493 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11495 (TARGET_MODES_TIEABLE_P): Redefine.
11496 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11497 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11498 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11499 (TARGET_MODES_TIEABLE_P): Redefine.
11500 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11501 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11502 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11503 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11504 (mn10300_modes_tieable_p): ...this and make static.
11505 (TARGET_MODES_TIEABLE_P): Redefine.
11506 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11507 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11508 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11509 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11510 (msp430_modes_tieable_p): Make static.
11511 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11512 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11513 (TARGET_MODES_TIEABLE_P): Redefine.
11514 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11515 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11516 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11517 (TARGET_MODES_TIEABLE_P): Redefine.
11518 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11519 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11520 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11521 (TARGET_MODES_TIEABLE_P): Redefine.
11522 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11523 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11524 (pdp11_modes_tieable_p): New function.
11525 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11526 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11527 (rs6000_modes_tieable_p): New function.
11528 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11529 * config/powerpcspe/powerpcspe.md: Update comment.
11530 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11531 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11532 (TARGET_MODES_TIEABLE_P): Redefine.
11533 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11534 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11535 (rl78_modes_tieable_p): New function.
11536 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11537 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11538 (rs6000_modes_tieable_p): New function.
11539 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11540 * config/rs6000/rs6000.md: Update comment.
11541 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11542 * config/rx/rx.c (rx_modes_tieable_p): New function.
11543 (TARGET_MODES_TIEABLE_P): Redefine.
11544 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11545 * config/s390/s390.c (s390_modes_tieable_p): New function.
11546 (TARGET_MODES_TIEABLE_P): Redefine.
11547 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11548 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11549 (sh_modes_tieable_p): New function.
11550 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11551 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11552 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11553 (sparc_modes_tieable_p): Make static.
11554 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11555 * config/spu/spu.c (spu_modes_tieable_p): New function.
11556 (TARGET_MODES_TIEABLE_P): Redefine.
11557 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11558 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11559 (TARGET_MODES_TIEABLE_P): Redefine.
11560 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11561 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11562 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11563 * config/v850/v850.c (v850_modes_tieable_p): New function.
11564 (TARGET_MODES_TIEABLE_P): Redefine.
11565 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11566 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11567 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11568 (visium_modes_tieable_p): New function.
11569 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11570 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11571 (xtensa_modes_tieable_p): New function.
11572 * system.h (MODES_TIEABLE_P): Poison.
11574 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11575 Alan Hayward <alan.hayward@arm.com>
11576 David Sherwood <david.sherwood@arm.com>
11578 * target.def (hard_regno_mode_ok): New hook.
11579 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11580 (TARGET_HARD_REGNO_MODE_OK): ...this.
11581 * doc/tm.texi.in: Regenerate.
11582 * hooks.h (hook_bool_uint_mode_true): Declare.
11583 * hooks.c (hook_bool_uint_mode_true): New function.
11584 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11585 HARD_REGNO_MODE_OK.
11586 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11587 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11588 instead of HARD_REGNO_MODE_OK.
11589 * caller-save.c: Include target.h.
11590 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11591 HARD_REGNO_MODE_OK.
11592 * combine.c (can_combine_p): Likewise.
11593 (combinable_i3pat): Likewise.
11594 (can_change_dest_mode): Likewise.
11595 * expr.c (init_expr_target): Likewise.
11596 (convert_move): Likewise.
11597 (convert_modes): Likewise.
11598 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11599 (setup_prohibited_mode_move_regs): Likewise.
11600 * ira.h (target_ira): Likewise.
11601 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11602 * lra-constraints.c (process_alt_operands): Likewise.
11603 (split_reg): Likewise.
11604 * recog.c (peep2_find_free_register): Likewise.
11605 * ree.c (combine_reaching_defs): Likewise.
11606 * regcprop.c (maybe_mode_change): Likewise.
11607 * reginfo.c (init_reg_sets_1): Likewise.
11608 (choose_hard_reg_mode): Likewise.
11609 (simplifiable_subregs): Likewise.
11610 * regrename.c (check_new_reg_p): Likewise.
11611 * reload.c (find_valid_class): Likewise.
11612 (find_valid_class_1): Likewise.
11613 (reload_inner_reg_of_subreg): Likewise.
11614 (push_reload): Likewise.
11615 (combine_reloads): Likewise.
11616 (find_dummy_reload): Likewise.
11617 (find_reloads): Likewise.
11618 * reload1.c (find_reg): Likewise.
11619 (set_reload_reg): Likewise.
11620 (allocate_reload_reg): Likewise.
11621 (choose_reload_regs): Likewise.
11622 (reload_adjust_reg_for_temp): Likewise.
11623 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11624 (simplify_subreg_regno): Likewise.
11625 * sel-sched.c (init_regs_for_mode): Likewise.
11626 * varasm.c (make_decl_rtl): Likewise.
11627 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11628 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11629 HARD_REGNO_MODE_OK.
11630 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11631 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11632 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11633 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11634 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11635 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11636 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11637 (arc_mode_class): Delete.
11638 (HARD_REGNO_MODE_OK): Delete.
11639 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11640 (arc_hard_regno_mode_ok): Rename old array to...
11641 (arc_hard_regno_mode_ok_modes): ...this.
11642 (arc_conditional_register_usage): Update accordingly.
11643 (arc_mode_class): Make static.
11644 (arc_hard_regno_mode_ok): New function.
11645 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11646 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11647 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11648 (arm_hard_regno_mode_ok): Make static.
11649 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11650 HARD_REGNO_MODE_OK.
11651 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11652 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11653 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11655 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11656 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11657 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11658 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11659 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11661 * config/bfin/predicates.md (valid_reg_operand): Use
11662 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11663 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11664 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11665 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11666 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11667 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11668 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11669 (cr16_hard_regno_mode_ok): Make static and return a bool.
11670 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11671 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11672 (cris_hard_regno_mode_ok): New function.
11673 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11674 (epiphany_mode_class): Delete.
11675 (HARD_REGNO_MODE_OK): Delete.
11676 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11677 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11678 (hard_regno_mode_ok): Rename to...
11679 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11680 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11681 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11682 HARD_REGNO_MODE_OK.
11683 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11684 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11685 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11686 (frv_hard_regno_mode_ok): Make static and return a bool.
11687 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11688 HARD_REGNO_MODE_OK.
11689 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11690 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11691 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11692 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11695 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11696 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11697 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11700 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11701 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11702 (ia64_hard_regno_mode_ok): New function.
11703 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11704 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11705 (iq2000_hard_regno_mode_ok): New function.
11706 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11707 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11708 (lm32_hard_regno_mode_ok): New function.
11709 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11710 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11711 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11712 instead of HARD_REGNO_MODE_OK.
11713 (m32c_hard_regno_ok): Rename to...
11714 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11715 (m32c_cannot_change_mode_class): Update accordingly.
11716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11717 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11718 (m32r_mode_class): Delete.
11719 (HARD_REGNO_MODE_OK): Delete.
11720 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11721 (m32r_hard_regno_mode_ok): Rename to...
11722 (m32r_hard_regno_modes): ...this.
11723 (m32r_mode_class): Make static.
11724 (m32r_hard_regno_mode_ok): New function.
11725 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11726 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11727 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11728 (m68k_hard_regno_mode_ok): Make static.
11729 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11730 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11731 (mcore_hard_regno_mode_ok): New function.
11732 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11733 (HARD_REGNO_MODE_OK): Delete.
11734 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11736 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11737 (microblaze_hard_regno_mode_ok): New function.
11738 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11739 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11740 (mips_hard_regno_mode_ok): Delete.
11741 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11742 (mips_hard_regno_mode_ok_p): ...this and make static.
11743 (mips_hard_regno_mode_ok_p): Rename to...
11744 (mips_hard_regno_mode_ok_uncached): ...this.
11745 (mips_hard_regno_mode_ok): New function.
11746 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11747 of HARD_REGNO_MODE_OK.
11748 (mips_option_override): Update after above name changes.
11749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11750 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11751 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11752 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11753 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11754 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11755 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11756 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11757 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11758 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11759 (msp430_hard_regno_mode_ok): Make static and return a bool.
11760 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11761 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11762 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11764 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11765 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11766 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11767 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11768 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11769 (PA_HARD_REGNO_MODE_OK): ...this
11770 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11771 (PA_HARD_REGNO_MODE_OK): ...this.
11772 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11773 (pa_hard_regno_mode_ok): New function.
11774 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11775 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11776 (pdp11_hard_regno_mode_ok): New function.
11777 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11778 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11780 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11783 (rs6000_hard_regno_mode_ok): Rename to...
11784 (rs6000_hard_regno_mode_ok_uncached): ...this.
11785 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11786 (rs6000_hard_regno_mode_ok): New function.
11787 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11788 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11789 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11790 (riscv_hard_regno_mode_ok): ...this and make static.
11791 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11792 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11793 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11794 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11795 (rl78_hard_regno_mode_ok): Make static and return bool.
11796 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11797 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11799 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11801 (rs6000_hard_regno_mode_ok): Rename to...
11802 (rs6000_hard_regno_mode_ok_uncached): ...this.
11803 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11804 (rs6000_hard_regno_mode_ok): New function.
11805 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11806 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11807 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11808 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11809 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11810 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11812 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11813 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11814 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11815 (sh_hard_regno_mode_ok): Make static.
11816 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11817 instead of HARD_REGNO_MODE_OK.
11818 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11819 (sparc_mode_class): Delete.
11820 (HARD_REGNO_MODE_OK): Delete.
11821 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11822 (hard_regno_mode_classes): Make static.
11823 (sparc_mode_class): Likewise.
11824 (sparc_hard_regno_mode_ok): New function.
11825 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11826 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11827 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11829 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11830 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11831 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11832 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11833 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11834 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11835 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11836 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11837 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11838 (visium_hard_regno_mode_ok): New function.
11839 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11840 instead of HARD_REGNO_MODE_OK.
11841 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11842 (HARD_REGNO_MODE_OK): Delete.
11843 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11844 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11845 (xtensa_option_override): Update accordingly.
11846 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11847 (xtensa_hard_regno_mode_ok): New function.
11848 * system.h (HARD_REGNO_MODE_OK): Poison.
11850 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11851 Alan Hayward <alan.hayward@arm.com>
11852 David Sherwood <david.sherwood@arm.com>
11854 * target.def (hard_regno_call_part_clobbered): New hook.
11855 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11856 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11857 * doc/tm.texi: Regenerate.
11858 * hooks.h (hook_bool_uint_mode_false): Declare.
11859 * hooks.c (hook_bool_uint_mode_false): New function.
11860 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11861 * cselib.c (cselib_process_insn): Use
11862 targetm.hard_regno_call_part_clobbered instead of
11863 HARD_REGNO_CALL_PART_CLOBBERED.
11864 * ira-conflicts.c (ira_build_conflicts): Likewise.
11865 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11866 * lra-constraints.c (need_for_call_save_p): Likewise.
11867 * lra-lives.c: Include target.h.
11868 (check_pseudos_live_through_calls): Use
11869 targetm.hard_regno_call_part_clobbered instead of
11870 HARD_REGNO_CALL_PART_CLOBBERED.
11871 * regcprop.c: Include target.h.
11872 (copyprop_hardreg_forward_1): Use
11873 targetm.hard_regno_call_part_clobbered instead of
11874 HARD_REGNO_CALL_PART_CLOBBERED.
11875 * reginfo.c (choose_hard_reg_mode): Likewise.
11876 * regrename.c (check_new_reg_p): Likewise.
11877 * reload.c (find_equiv_reg): Likewise.
11878 * reload1.c (emit_reload_insns): Likewise.
11879 * sched-deps.c (deps_analyze_insn): Likewise.
11880 * sel-sched.c (init_regs_for_mode): Likewise.
11881 (mark_unavailable_hard_regs): Likewise.
11882 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11883 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11884 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11886 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11887 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11888 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11890 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11892 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11893 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11894 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11896 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11897 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11898 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11900 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11901 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11903 * config/powerpcspe/powerpcspe.c
11904 (rs6000_hard_regno_call_part_clobbered): New function.
11905 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11906 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11907 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11909 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11910 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11911 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11913 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11914 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11915 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11917 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * rtl.h (subreg_memory_offset): Declare.
11922 * emit-rtl.c (subreg_memory_offset): New function.
11923 * expmed.c (store_bit_field_1): Use it.
11924 * expr.c (undefined_operand_subword_p): Likewise.
11925 * simplify-rtx.c (simplify_subreg): Likewise.
11927 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11929 PR rtl-optimization/57448
11932 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11933 using atomic_load pattern.
11934 (expand_atomic_store): Likewise.
11936 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11939 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11940 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11941 handling. Use replace_call_with_value with NULL instead of
11942 gsi_replace, unlink_stmt_vdef and release_defs.
11944 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11947 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11949 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11952 * rtl.h (emit_library_call_value_1): Declare.
11953 (emit_library_call): Replace declaration with a series of overloads.
11954 Remove the parameter count argument.
11955 (emit_library_call_value): Likewise.
11956 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11957 with an "rtx_mode_t *".
11958 (emit_library_call_value): Delete.
11959 (emit_library_call): Likewise.
11960 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11961 (asan_emit_allocas_unpoison): Likewise.
11962 * builtins.c (expand_builtin_powi): Likewise.
11963 (expand_asan_emit_allocas_unpoison): Likewise.
11964 * cfgexpand.c (expand_main_function): Likewise.
11965 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11966 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11967 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11968 * config/arm/arm.c (arm_trampoline_init): Likewise.
11969 (arm_call_tls_get_addr): Likewise.
11970 (arm_expand_divmod_libfunc): Likewise.
11971 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11972 (smulsi3_highpart): Likewise.
11973 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11974 (c6x_expand_compare): Likewise.
11975 (c6x_expand_movmem): Likewise.
11976 * config/frv/frv.c (frv_trampoline_init): Likewise.
11977 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11978 (ix86_expand_divmod_libfunc): Likewise.
11979 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11980 (ia64_expand_compare): Likewise.
11981 (ia64_profile_hook): Likewise.
11982 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11983 (nonlocal_goto): Likewise.
11984 (restore_stack_nonlocal): Likewise.
11985 * config/m32r/m32r.c (block_move_call): Likewise.
11986 (m32r_trampoline_init): Likewise.
11987 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11988 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11989 (m68k_call_m68k_read_tp): Likewise.
11990 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11991 (microblaze_expand_divide): Likewise.
11992 * config/mips/mips.h (mips_args): Likewise.
11993 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11994 (MIPS_ICACHE_SYNC): Likewise.
11995 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11996 (nios2_trampoline_init): Likewise.
11997 * config/pa/pa.c (hppa_tls_call): Likewise.
11998 (pa_trampoline_init): Likewise.
11999 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12000 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12001 (expand_strn_compare): Likewise.
12002 (rs6000_generate_compare): Likewise.
12003 (rs6000_expand_float128_convert): Likewise.
12004 (output_profile_hook): Likewise.
12005 (rs6000_trampoline_init): Likewise.
12006 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12007 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12008 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12009 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12010 (rs6000_generate_compare): Likewise.
12011 (rs6000_expand_float128_convert): Likewise.
12012 (output_profile_hook): Likewise.
12013 (rs6000_trampoline_init): Likewise.
12014 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12015 * config/sh/sh.c (sh_trampoline_init): Likewise.
12016 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12017 (sparc_emit_float_lib_cmp): Likewise.
12018 (sparc32_initialize_trampoline): Likewise.
12019 (sparc64_initialize_trampoline): Likewise.
12020 (sparc_profile_hook): Likewise.
12021 * config/spu/spu.c (ea_load_store): Likewise.
12022 * config/spu/spu.md (floatunssidf2): Likewise.
12023 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12024 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12025 * config/visium/visium.c (expand_block_move_4): Likewise.
12026 (expand_block_move_2): Likewise.
12027 (expand_block_move_1): Likewise.
12028 (expand_block_set_4): Likewise.
12029 (expand_block_set_2): Likewise.
12030 (expand_block_set_1): Likewise.
12031 (visium_trampoline_init): Likewise.
12032 (visium_profile_hook): Likewise.
12033 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12034 (xtensa_setup_frame_addresses): Likewise.
12035 (xtensa_trampoline_init): Likewise.
12036 * except.c (sjlj_emit_function_enter): Likewise.
12037 (sjlj_emit_function_exit): Likewise.
12038 * explow.c (allocate_dynamic_stack_space): Likewise.
12039 (probe_stack_range): Likewise.
12040 * expr.c (convert_mode_scalar): Likewise.
12041 * optabs.c (expand_binop): Likewise.
12042 (expand_twoval_binop_libfunc): Likewise.
12043 (expand_unop): Likewise.
12044 (prepare_cmp_insn): Likewise.
12045 (prepare_float_lib_cmp): Likewise.
12046 (expand_float): Likewise.
12047 (expand_fix): Likewise.
12048 (expand_fixed_convert): Likewise.
12049 (maybe_emit_sync_lock_test_and_set): Likewise.
12050 (expand_atomic_compare_and_swap): Likewise.
12051 (expand_mem_thread_fence): Likewise.
12052 (expand_atomic_fetch_op): Likewise.
12054 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12056 * doc/generic.texi (OpenACC): Adjust URL.
12057 * doc/invoke.texi (C Dialect Options): Ditto.
12059 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12061 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12062 predicate for operand 1. Add (m,<S>) constraint.
12063 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12064 Prevent memory operand 1 with register operand 2.
12066 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12068 PR rtl-optimization/82024
12069 * combine.c (try_combine): If the combination result is a PARALLEL,
12070 and we only need to retain the SET in there that would be placed
12071 at I2, check that we can place that at I3 instead, before doing so.
12073 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12076 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12078 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12079 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12082 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12083 Jeff Law <law@redhat.com>
12085 * varasm.c (bss_initializer_p): Do not put constants into .bss
12086 (categorize_decl_for_section): Handle bss_initializer_p returning
12087 false when DECL_INITIAL is NULL.
12089 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12092 * config/s390/s390.c (s390_can_inline_p): New function.
12094 2017-09-01 Jeff Law <law@redhat.com>
12096 PR tree-optimization/82052
12097 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12098 Always initialize the returned slot after a hash table miss
12099 when INSERT is true.
12101 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12103 * config/s390/s390.md (mem_signal_fence): Remove.
12104 * doc/md.texi (mem_signal_fence): Remove.
12105 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12107 * target-insns.def (mem_signal_fence): Remove.
12109 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12112 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12115 * asan.c (create_odr_indicator): Strip name encoding from assembler
12116 name before appending it after __odr_asan_.
12118 2017-09-01 Martin Liska <mliska@suse.cz>
12120 PR tree-optimization/82059
12121 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12122 frequency only when an edge is redirected.
12124 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12126 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12127 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12128 (arc_conditional_register_usage): Remove ARC600 lp_count
12130 (arc_file_start): Emit Tag_ARC_CPU_variation.
12131 (arc_can_use_doloop_p): New conditions to use ZOLs.
12132 (hwloop_fail): New function.
12133 (hwloop_optimize): Likewise.
12134 (hwloop_pattern_reg): Likewise.
12135 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12136 (arc_reorg_loops): New function, calls reorg_loops.
12137 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12138 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12140 (arc_loop_hazard): Remove function, functionality moved into
12142 (arc_hazard): Remove arc_loop_hazard call.
12143 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12144 into hwloop_optimize.
12145 (arc_label_align): Remove ZOL handling.
12146 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12147 * config/arc/arc.md (doloop_begin): Remove pattern.
12148 (doloop_begin_i): Likewise.
12149 (doloop_end_i): Likewise.
12150 (doloop_fallback): Likewise.
12151 (doloop_fallback_m): Likewise.
12152 (doloop_end): Reimplement expand.
12153 (arc_lp): New pattern for LP instruction.
12154 (loop_end): New pattern.
12155 (loop_fail): Likewise.
12156 (decrement_and_branch_until_zero): Likewise.
12157 * config/arc/arc.opt (mlpc-width): New option.
12158 * doc/invoke.texi (mlpc-width): Document option.
12160 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12162 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12163 (arc_ccfsm_advance): Fix checking for delay slots.
12164 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12166 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12168 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12170 (movhi_insn): Update store instruction constraint which are saving
12171 6-bit short immediates.
12172 (movsi_insn): Consider also short scaled load operations.
12173 (zero_extendhisi2_i): Use Usd constraint instead of T.
12174 (extendhisi2_i): Add q constraint.
12175 (arc_clzsi2): Add type and length attributes.
12176 (arc_ctzsi2): Likewise.
12177 * config/arc/constraints.md (Usc): Update constraint, the
12178 assembler can parse two relocations for a single instruction.
12180 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12182 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12183 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12185 2017-08-31 Olivier Hainque <hainque@adacore.com>
12187 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12188 match as powerpc-wrs-vxworks*.
12190 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12192 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12193 register constraint for by-element operand.
12194 (aarch64_mls_elt_merge<mode>): Likewise.
12196 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12198 * config/arc/arc.c (arc_can_follow_jump): Check for short
12201 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12203 * config.gcc: Use g.opt for arc.
12204 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12205 functionality moved to ...
12206 (legitimate_scaled_address_p): New function, ...here.
12207 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12208 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12209 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12211 (arc_override_options): Handle G option.
12212 (arc_output_pic_addr_const): Correct function definition.
12213 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12214 (arc_decl_anon_ns_mem_p): Delete.
12215 (arc_in_small_data_p): Overhaul this function to take into
12216 consideration the value given via G option.
12217 (arc_rewrite_small_data_1): Renamed and corrected old
12218 arc_rewrite_small_data function.
12219 (arc_rewrite_small_data): New function.
12220 (small_data_pattern): Don't use pic_offset_table_rtx.
12221 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12222 * config/arc/simdext.md (movmisalignv2hi): Use
12223 prepare_move_operands function.
12225 (movmisalign*): Likewise.
12226 * doc/invoke.texi (ARC options): Document -G option.
12228 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12230 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12232 * config/arc/arc.c (arc_print_operand): Output scalled address for
12233 sdata whenever is possible.
12234 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12235 load/stores are available.
12236 (compact_sda_memory_operand): Check for the alignment required by
12237 code density instructions.
12238 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12240 * config/arc/constraints.md (Usd): Update constraint.
12241 (Us0): New constraint.
12242 (Usc): Update constraint.
12244 2017-08-31 Richard Biener <rguenther@suse.de>
12246 PR middle-end/82054
12247 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12248 function only once.
12250 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12252 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12253 Resize type_signature.
12255 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12256 Alan Hayward <alan.hayward@arm.com>
12257 David Sherwood <david.sherwood@arm.com>
12259 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12260 subregs whose inner modes can be stored in GPRs.
12261 (aarch64_classify_index): Likewise.
12263 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12268 (V_INT_EQUIV): ...this.
12269 (v_cmp_result): Rename to...
12270 (v_int_equiv): ...this.
12271 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12272 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12273 (copysign<mode>3): Likewise.
12274 (aarch64_simd_bsl<mode>_internal): Likewise.
12275 (aarch64_simd_bsl<mode>): Likewise.
12276 (vec_cmp<mode><mode>): Likewise.
12277 (vcond<mode><mode>): Likewise.
12278 (vcond<v_cmp_mixed><mode>): Likewise.
12279 (vcondu<mode><v_cmp_mixed>): Likewise.
12280 (aarch64_cm<optab><mode>): Likewise.
12281 (aarch64_cmtst<mode>): Likewise.
12282 (aarch64_fac<optab><mode>): Likewise.
12283 (vec_perm_const<mode>): Likewise.
12284 (vcond_mask_<mode><v_cmp_result>): Rename to...
12285 (vcond_mask_<mode><v_int_equiv>): ...this.
12286 (vec_cmp<mode><v_cmp_result>): Rename to...
12287 (vec_cmp<mode><v_int_equiv>): ...this.
12289 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12290 Alan Hayward <alan.hayward@arm.com>
12291 David Sherwood <david.sherwood@arm.com>
12293 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12295 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12296 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12297 (UNSPEC_LD4_DREG): New unspecs.
12298 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12299 (aarch64_ld2<mode>_dreg_be): Replace with...
12300 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12302 (aarch64_ld3<mode>_dreg_le)
12303 (aarch64_ld3<mode>_dreg_be): Replace with...
12304 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12306 (aarch64_ld4<mode>_dreg_le)
12307 (aarch64_ld4<mode>_dreg_be): Replace with...
12308 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12311 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12313 PR tree-optimization/81987
12314 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12315 insert an initializer in a location not dominated by the stride
12318 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12320 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12321 on the entire header of the finally block in the fallthru case.
12323 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12325 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12327 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12329 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12330 rs6000_emit_move_from_cr and call renamed function.
12331 (rs6000_emit_prologue): Call renamed functions.
12332 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12333 movesi_from_cr, remove volatile CRs.
12335 2017-08-30 Jon Beniston <jon@beniston.com>
12336 Richard Biener <rguenther@suse.de>
12338 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12339 of VECTOR_MODE_P check.
12340 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12341 element vector types.
12343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12345 * df.h (df_read_modify_subreg_p): Remove in favor of...
12346 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12347 const_rtx instead of an rtx.
12348 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12349 * df-problems.c (df_word_lr_mark_ref): Likewise.
12350 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12351 (mark_pseudo_reg_dead): Likewise.
12352 (mark_ref_dead): Likewise.
12353 * reginfo.c (init_subregs_of_mode): Likewise.
12354 * sched-deps.c (sched_analyze_1): Likewise.
12355 * df-scan.c (df_def_record_1): Likewise.
12356 (df_uses_record): Likewise.
12357 (df_read_modify_subreg_p): Remove in favor of...
12358 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12359 const_rtx instead of an rtx.
12361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12362 Alan Hayward <alan.hayward@arm.com>
12363 David Sherwood <david.sherwood@arm.com>
12365 * rtl.h (partial_subreg_p): New function.
12366 * caller-save.c (save_call_clobbered_regs): Use it.
12367 * calls.c (expand_call): Likewise.
12368 * combine.c (combinable_i3pat): Likewise.
12369 (simplify_set): Likewise.
12370 (make_extraction): Likewise.
12371 (make_compound_operation_int): Likewise.
12372 (gen_lowpart_or_truncate): Likewise.
12373 (force_to_mode): Likewise.
12374 (make_field_assignment): Likewise.
12375 (reg_truncated_to_mode): Likewise.
12376 (record_truncated_value): Likewise.
12377 (move_deaths): Likewise.
12378 * cse.c (record_jump_cond): Likewise.
12379 (cse_insn): Likewise.
12380 * cselib.c (cselib_lookup_1): Likewise.
12381 * expmed.c (extract_bit_field_using_extv): Likewise.
12382 * function.c (assign_parm_setup_reg): Likewise.
12383 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12384 * ira-build.c (create_insn_allocnos): Likewise.
12385 * lra-coalesce.c (merge_pseudos): Likewise.
12386 * lra-constraints.c (match_reload): Likewise.
12387 (simplify_operand_subreg): Likewise.
12388 (curr_insn_transform): Likewise.
12389 * lra-lives.c (process_bb_lives): Likewise.
12390 * lra.c (new_insn_reg): Likewise.
12391 (lra_substitute_pseudo): Likewise.
12392 * regcprop.c (mode_change_ok): Likewise.
12393 (maybe_mode_change): Likewise.
12394 (copyprop_hardreg_forward_1): Likewise.
12395 * reload.c (push_reload): Likewise.
12396 (find_reloads): Likewise.
12397 (find_reloads_subreg_address): Likewise.
12398 * reload1.c (alter_reg): Likewise.
12399 (eliminate_regs_1): Likewise.
12400 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12402 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12404 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12405 back to if statements, including unpack.
12407 2017-08-30 Martin Liska <mliska@suse.cz>
12409 PR inline-asm/82001
12410 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12412 (func_checker::compare_asm_inputs_outputs): ... this function.
12413 (func_checker::compare_gimple_asm): Use the function to compare
12414 also ASM constrains.
12415 * ipa-icf-gimple.h: Rename the function.
12417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12418 Alan Hayward <alan.hayward@arm.com>
12419 David Sherwood <david.sherwood@arm.com>
12421 * coretypes.h (complex_mode): New type.
12422 * gdbhooks.py (build_pretty_printer): Handle it.
12423 * machmode.h (complex_mode): New class.
12424 (complex_mode::includes_p): New function.
12425 (is_complex_int_mode): Likewise.
12426 (is_complex_float_mode): Likewise.
12427 * genmodes.c (get_mode_class): Handle complex mode classes.
12428 * function.c (expand_function_end): Use is_complex_int_mode.
12430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12431 Alan Hayward <alan.hayward@arm.com>
12432 David Sherwood <david.sherwood@arm.com>
12434 * coretypes.h (scalar_mode_pod): New typedef.
12435 * gdbhooks.py (build_pretty_printer): Handle it.
12436 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12437 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12438 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12439 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12440 as_a <scalar_mode>.
12442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12443 Alan Hayward <alan.hayward@arm.com>
12444 David Sherwood <david.sherwood@arm.com>
12446 * machmode.h (mode_for_vector): Take a scalar_mode instead
12448 * stor-layout.c (mode_for_vector): Likewise.
12449 * explow.c (promote_mode): Use as_a <scalar_mode>.
12450 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12453 Alan Hayward <alan.hayward@arm.com>
12454 David Sherwood <david.sherwood@arm.com>
12456 * target.def (preferred_simd_mode): Take a scalar_mode
12457 instead of a machine_mode.
12458 * targhooks.h (default_preferred_simd_mode): Likewise.
12459 * targhooks.c (default_preferred_simd_mode): Likewise.
12460 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
12461 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
12462 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
12463 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
12464 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
12465 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
12466 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
12467 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
12469 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
12470 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
12471 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
12472 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
12473 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
12474 * doc/tm.texi: Regenerate.
12475 * optabs-query.c (can_vec_mask_load_store_p): Return false for
12478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12479 Alan Hayward <alan.hayward@arm.com>
12480 David Sherwood <david.sherwood@arm.com>
12482 * target.def (scalar_mode_supported_p): Take a scalar_mode
12483 instead of a machine_mode.
12484 * targhooks.h (default_scalar_mode_supported_p): Likewise.
12485 * targhooks.c (default_scalar_mode_supported_p): Likewise.
12486 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
12487 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12488 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
12489 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
12490 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
12491 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12492 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12493 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12494 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12495 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12496 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12497 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12499 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12500 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12501 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12502 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12503 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12505 * doc/tm.texi: Regenerate.
12507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12508 Alan Hayward <alan.hayward@arm.com>
12509 David Sherwood <david.sherwood@arm.com>
12511 * coretypes.h (opt_scalar_mode): New typedef.
12512 * gdbhooks.py (build_pretty_printers): Handle it.
12513 * machmode.h (mode_iterator::get_2xwider): Add overload for
12515 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12517 * expr.c (convert_mode_scalar): Likewise.
12518 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12519 * optabs.c (expand_float): Likewise.
12520 (expand_fix): Likewise.
12521 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12524 Alan Hayward <alan.hayward@arm.com>
12525 David Sherwood <david.sherwood@arm.com>
12527 * optabs.c (expand_float): Explicitly check for scalars before
12528 using a branching expansion.
12529 (expand_fix): Likewise.
12531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12532 Alan Hayward <alan.hayward@arm.com>
12533 David Sherwood <david.sherwood@arm.com>
12535 * expr.c (convert_mode): Split scalar handling out into...
12536 (convert_mode_scalar): ...this new function. Treat the modes
12539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12540 Alan Hayward <alan.hayward@arm.com>
12541 David Sherwood <david.sherwood@arm.com>
12543 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12545 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12548 Alan Hayward <alan.hayward@arm.com>
12549 David Sherwood <david.sherwood@arm.com>
12551 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12552 rather than a machine_mode.
12553 (fixed_from_string): Likewise.
12554 (fixed_convert): Likewise.
12555 (fixed_convert_from_int): Likewise.
12556 (fixed_convert_from_real): Likewise.
12557 (real_convert_from_fixed): Likewise.
12558 * fixed-value.c (fixed_from_double_int): Likewise.
12559 (fixed_from_string): Likewise.
12560 (fixed_convert): Likewise.
12561 (fixed_convert_from_int): Likewise.
12562 (fixed_convert_from_real): Likewise.
12563 (real_convert_from_fixed): Likewise.
12564 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12567 Alan Hayward <alan.hayward@arm.com>
12568 David Sherwood <david.sherwood@arm.com>
12570 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12571 of separate mode class checks. Do not allow vector modes here.
12572 (immed_wide_int_const): Use as_a <scalar_mode>.
12573 * explow.c (trunc_int_for_mode): Likewise.
12574 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12575 (wi::shwi): Likewise.
12576 (wi::min_value): Likewise.
12577 (wi::max_value): Likewise.
12578 * dwarf2out.c (loc_descriptor): Likewise.
12579 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12580 for CONST_WIDE_INT.
12582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12583 Alan Hayward <alan.hayward@arm.com>
12584 David Sherwood <david.sherwood@arm.com>
12586 * tree.h (SCALAR_TYPE_MODE): New macro.
12587 * expr.c (expand_expr_addr_expr_1): Use it.
12588 (expand_expr_real_2): Likewise.
12589 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12590 (fold_convert_const_fixed_from_int): Likewise.
12591 (fold_convert_const_fixed_from_real): Likewise.
12592 (native_encode_fixed): Likewise
12593 (native_encode_complex): Likewise
12594 (native_encode_vector): Likewise.
12595 (native_interpret_fixed): Likewise.
12596 (native_interpret_real): Likewise.
12597 (native_interpret_complex): Likewise.
12598 (native_interpret_vector): Likewise.
12599 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12600 (simd_clone_adjust_argument_types): Likewise.
12601 (simd_clone_init_simd_arrays): Likewise.
12602 (simd_clone_adjust): Likewise.
12603 * stor-layout.c (layout_type): Likewise.
12604 * tree.c (build_minus_one_cst): Likewise.
12605 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12606 * tree-inline.c (estimate_move_cost): Likewise.
12607 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12608 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12609 (vectorizable_reduction): Likewise.
12610 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12611 (vect_recog_mixed_size_cond_pattern): Likewise.
12612 (check_bool_pattern): Likewise.
12613 (adjust_bool_pattern): Likewise.
12614 (search_type_for_mask_1): Likewise.
12615 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12616 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12617 (vectorizable_load): Likewise.
12618 (vectorizable_store): Likewise.
12619 * ubsan.c (ubsan_encode_value): Likewise.
12620 * varasm.c (output_constant): Likewise.
12622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12623 Alan Hayward <alan.hayward@arm.com>
12624 David Sherwood <david.sherwood@arm.com>
12626 * coretypes.h (scalar_mode): New class.
12627 * machmode.h (scalar_mode): Likewise.
12628 (scalar_mode::includes_p): New function.
12629 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12630 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12631 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12632 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12633 * expmed.c (store_bit_field_1): Likewise.
12634 (extract_bit_field_1): Likewise.
12635 * expr.c (write_complex_part): Likewise.
12636 (read_complex_part): Likewise.
12637 (emit_move_complex_push): Likewise.
12638 (expand_expr_real_2): Likewise.
12639 * function.c (assign_parm_setup_reg): Likewise.
12640 (assign_parms_unsplit_complex): Likewise.
12641 * optabs.c (expand_binop): Likewise.
12642 * rtlanal.c (subreg_get_info): Likewise.
12643 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12644 * varasm.c (output_constant_pool_2): Likewise.
12646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12647 Alan Hayward <alan.hayward@arm.com>
12648 David Sherwood <david.sherwood@arm.com>
12650 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12652 (expmed_mult_highpart_optab): Likewise.
12653 (expmed_mult_highpart): Likewise.
12655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12656 Alan Hayward <alan.hayward@arm.com>
12657 David Sherwood <david.sherwood@arm.com>
12659 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12660 instead of a machine_mode.
12661 (builtin_memset_read_str): Likewise.
12662 * builtins.c (c_readstr): Likewise.
12663 (builtin_memcpy_read_str): Likewise.
12664 (builtin_strncpy_read_str): Likewise.
12665 (builtin_memset_read_str): Likewise.
12666 (builtin_memset_gen_str): Likewise.
12667 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12668 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12669 instead of a machine_mode.
12670 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12672 (make_extraction): Likewise.
12673 (try_widen_shift_mode): Take and return scalar_int_modes instead
12675 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12676 a scalar_int_mode instead of a machine_mode.
12677 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12678 (avr_addr_space_pointer_mode): Likewise.
12679 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12680 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12681 (msp430_unwind_word_mode): Likewise.
12682 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12683 (spu_addr_space_pointer_mode): Likewise.
12684 (spu_addr_space_address_mode): Likewise.
12685 (spu_libgcc_cmp_return_mode): Likewise.
12686 (spu_libgcc_shift_count_mode): Likewise.
12687 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12688 (rl78_addr_space_pointer_mode): Likewise.
12689 (fl78_unwind_word_mode): Likewise.
12690 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12692 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12693 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12694 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12695 (mips_valid_pointer_mode): Likewise.
12696 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12697 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12698 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12700 (ft32_addr_space_address_mode): Likewise.
12701 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12702 scalar_int_mode instead of a machine_mode.
12703 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12705 (m32c_addr_space_address_mode): Likewise.
12706 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12707 (rs6000_eh_return_filter_mode): Likewise.
12708 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12709 (rs6000_eh_return_filter_mode): Likewise.
12710 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12711 (s390_libgcc_shift_count_mode): Likewise.
12712 (s390_unwind_word_mode): Likewise.
12713 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12715 * target.def (mode_rep_extended): Likewise.
12716 (valid_pointer_mode): Likewise.
12717 (addr_space.valid_pointer_mode): Likewise.
12718 (eh_return_filter_mode): Return a scalar_int_mode rather than
12720 (libgcc_cmp_return_mode): Likewise.
12721 (libgcc_shift_count_mode): Likewise.
12722 (unwind_word_mode): Likewise.
12723 (addr_space.pointer_mode): Likewise.
12724 (addr_space.address_mode): Likewise.
12725 * doc/tm.texi: Regenerate.
12726 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12728 (do_jump): Use scalar_int_mode for local variables.
12729 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12730 rather than a machine_mode.
12731 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12732 (scompare_loc_descriptor_wide): Likewise.
12733 (scompare_loc_descriptor_narrow): Likewise.
12734 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12736 * except.c (sjlj_emit_dispatch_table): Likewise.
12737 (expand_builtin_eh_copy_values): Likewise.
12738 * explow.c (convert_memory_address_addr_space_1): Likewise.
12739 Take a scalar_int_mode rather than a machine_mode.
12740 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12741 than a machine_mode.
12742 (memory_address_addr_space): Use scalar_int_mode for local variables.
12743 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12744 rather than a machine_mode.
12745 * expmed.c (mask_rtx): Likewise.
12746 (init_expmed_one_conv): Likewise.
12747 (expand_mult_highpart_adjust): Likewise.
12748 (extract_high_half): Likewise.
12749 (expmed_mult_highpart_optab): Likewise.
12750 (expmed_mult_highpart): Likewise.
12751 (expand_smod_pow2): Likewise.
12752 (expand_sdiv_pow2): Likewise.
12753 (emit_store_flag_int): Likewise.
12754 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12756 (extract_low_bits): Likewise.
12757 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12759 * expr.c (pieces_addr::adjust): Likewise.
12760 (can_store_by_pieces): Likewise.
12761 (store_by_pieces): Likewise.
12762 (clear_by_pieces_1): Likewise.
12763 (expand_expr_addr_expr_1): Likewise.
12764 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12765 (expand_expr_real_1): Likewise.
12766 (try_casesi): Likewise.
12767 * final.c (shorten_branches): Likewise.
12768 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12769 type of "mode" to machine_mode.
12770 * internal-fn.c (expand_arith_overflow_result_store): Take a
12771 scalar_int_mode rather than a machine_mode.
12772 (expand_mul_overflow): Use scalar_int_mode for local variables.
12773 * loop-doloop.c (doloop_modify): Likewise.
12774 (doloop_optimize): Likewise.
12775 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12776 than a machine_mode.
12777 (expand_doubleword_shift_condmove): Likewise.
12778 (expand_doubleword_shift): Likewise.
12779 (expand_doubleword_clz): Likewise.
12780 (expand_doubleword_popcount): Likewise.
12781 (expand_doubleword_parity): Likewise.
12782 (expand_absneg_bit): Use scalar_int_mode for local variables.
12783 (prepare_float_lib_cmp): Likewise.
12784 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12785 rather than a machine_mode.
12786 (convert_memory_address_addr_space): Likewise.
12787 (get_mode_bounds): Likewise.
12788 (get_address_mode): Return a scalar_int_mode rather than a
12790 * rtlanal.c (get_address_mode): Likewise.
12791 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12792 than a machine_mode.
12793 * targhooks.c (default_mode_rep_extended): Likewise.
12794 (default_valid_pointer_mode): Likewise.
12795 (default_addr_space_valid_pointer_mode): Likewise.
12796 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12797 than a machine_mode.
12798 (default_libgcc_cmp_return_mode): Likewise.
12799 (default_libgcc_shift_count_mode): Likewise.
12800 (default_unwind_word_mode): Likewise.
12801 (default_addr_space_pointer_mode): Likewise.
12802 (default_addr_space_address_mode): Likewise.
12803 * targhooks.h (default_eh_return_filter_mode): Likewise.
12804 (default_libgcc_cmp_return_mode): Likewise.
12805 (default_libgcc_shift_count_mode): Likewise.
12806 (default_unwind_word_mode): Likewise.
12807 (default_addr_space_pointer_mode): Likewise.
12808 (default_addr_space_address_mode): Likewise.
12809 (default_mode_rep_extended): Take a scalar_int_mode rather than
12811 (default_valid_pointer_mode): Likewise.
12812 (default_addr_space_valid_pointer_mode): Likewise.
12813 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12815 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12816 rather than a machine_mode.
12817 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12818 for local variables.
12819 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12820 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12821 than a machine_mode.
12823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12824 Alan Hayward <alan.hayward@arm.com>
12825 David Sherwood <david.sherwood@arm.com>
12827 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12828 the mode argument to scalar_int_mode.
12829 (do_jump_by_parts_zero_rtx): Likewise.
12830 (do_jump_by_parts_equality_rtx): Likewise.
12831 (do_jump_by_parts_greater): Take a mode argument.
12832 (do_jump_by_parts_equality): Likewise.
12833 (do_jump_1): Update calls accordingly.
12835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12836 Alan Hayward <alan.hayward@arm.com>
12837 David Sherwood <david.sherwood@arm.com>
12839 * is-a.h (safe_dyn_cast): New function.
12840 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12841 (jump_table_for_label): Likewise.
12842 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12843 instead of an rtx_insn *.
12844 (shorten_branches): Use dyn_cast instead of LABEL_P and
12845 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12846 rtx_jump_table_data::get_data_mode.
12847 (final_scan_insn): Likewise.
12849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12850 Alan Hayward <alan.hayward@arm.com>
12851 David Sherwood <david.sherwood@arm.com>
12853 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12854 trying to combine a full-register integer set with a subreg
12857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12858 Alan Hayward <alan.hayward@arm.com>
12859 David Sherwood <david.sherwood@arm.com>
12861 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12862 that is always either address_mode or pointer_mode.
12864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12865 Alan Hayward <alan.hayward@arm.com>
12866 David Sherwood <david.sherwood@arm.com>
12868 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12869 when the two are known to be equal.
12871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12872 Alan Hayward <alan.hayward@arm.com>
12873 David Sherwood <david.sherwood@arm.com>
12875 * simplify-rtx.c (simplify_const_unary_operation): Use
12876 is_a <scalar_int_mode> instead of checking for a nonzero
12877 precision. Forcibly convert op_mode to a scalar_int_mode
12878 in that case. More clearly differentiate the operand and
12879 result modes and use the former when deciding what the value
12880 of a count-bits operation should be. Use is_int_mode instead
12881 of checking for a MODE_INT. Remove redundant check for whether
12882 this mode has a zero precision.
12884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12885 Alan Hayward <alan.hayward@arm.com>
12886 David Sherwood <david.sherwood@arm.com>
12888 * optabs.c (widen_leading): Change the type of the mode argument
12889 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12890 (widen_bswap): Likewise.
12891 (expand_parity): Likewise.
12892 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12893 (expand_ffs): Likewise.
12894 (epand_unop): Check for scalar integer modes before calling the
12897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12898 Alan Hayward <alan.hayward@arm.com>
12899 David Sherwood <david.sherwood@arm.com>
12901 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12903 (expand_expr_real_1): Update call accordingly.
12905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12906 Alan Hayward <alan.hayward@arm.com>
12907 David Sherwood <david.sherwood@arm.com>
12909 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12910 value_mode arguments. Use scalar_int_mode internally.
12911 (store_bit_field_1): Rename the new integer mode from imode
12912 to op0_mode and use it instead of GET_MODE (op0). Update calls
12913 to store_split_bit_field, store_bit_field_using_insv and
12914 store_fixed_bit_field.
12915 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12916 Use scalar_int_mode internally. Use a bit count rather than a mode
12917 when calculating the largest bit size for get_best_mode.
12918 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12919 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12920 Remove assertion that OP0 has a scalar integer mode.
12921 (store_split_bit_field): Add op0_mode and value_mode arguments.
12922 Update calls to extract_fixed_bit_field.
12923 (extract_bit_field_using_extv): Add an op0_mode argument.
12924 Use scalar_int_mode internally.
12925 (extract_bit_field_1): Rename the new integer mode from imode to
12926 op0_mode and use it instead of GET_MODE (op0). Update calls to
12927 extract_split_bit_field, extract_bit_field_using_extv and
12928 extract_fixed_bit_field.
12929 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12930 to extract_split_bit_field and extract_fixed_bit_field_1.
12931 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12932 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12933 on the target mode.
12934 (extract_split_bit_field): Add an op0_mode argument. Update call
12935 to extract_fixed_bit_field.
12937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12938 Alan Hayward <alan.hayward@arm.com>
12939 David Sherwood <david.sherwood@arm.com>
12941 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12942 * explow.c (hard_function_value): Likewise.
12943 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12944 convert_to_mode call outside the loop.
12945 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12946 for the mode iterator. Require the mode specified by max_pieces
12948 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12950 (copy_blkmode_to_reg): Likewise.
12951 (set_storage_via_setmem): Likewise.
12952 * optabs.c (prepare_cmp_insn): Likewise.
12953 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12954 * stor-layout.c (finish_bitfield_representative): Likewise.
12956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12957 Alan Hayward <alan.hayward@arm.com>
12958 David Sherwood <david.sherwood@arm.com>
12960 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12961 * expr.c (convert_move): Use them.
12962 (convert_modes): Likewise.
12963 (store_expr_with_bounds): Likewise.
12965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12966 Alan Hayward <alan.hayward@arm.com>
12967 David Sherwood <david.sherwood@arm.com>
12969 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12970 parameter for the mode of "x". Remove the "known_x", "known_mode"
12971 and "known_ret" arguments. Change the type of the mode argument
12972 to scalar_int_mode.
12973 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12974 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12975 (reg_num_sign_bit_copies_for_combine): Likewise.
12976 * rtlanal.c (nonzero_bits1): Likewise.
12977 (num_sign_bit_copies1): Likewise.
12978 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12979 (reg_num_sign_bit_copies_general): Likewise.
12980 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12981 (reg_nonzero_bits_general): Likewise.
12983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12984 Alan Hayward <alan.hayward@arm.com>
12985 David Sherwood <david.sherwood@arm.com>
12987 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12988 than in subroutines. Return 1 for non-integer modes.
12989 (cached_num_sign_bit_copies): Change the type of the mode parameter
12990 to scalar_int_mode.
12991 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12992 classes. Handle CONST_INT_P first and then check whether X also
12993 has a scalar integer mode. Check the same thing for inner registers
12994 of a SUBREG and for values that are being extended or truncated.
12996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12997 Alan Hayward <alan.hayward@arm.com>
12998 David Sherwood <david.sherwood@arm.com>
13000 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13001 in subroutines. Return the mode mask for non-integer modes.
13002 (cached_nonzero_bits): Change the type of the mode parameter
13003 to scalar_int_mode.
13004 (nonzero_bits1): Likewise. Remove early exit for other mode
13005 classes. Handle CONST_INT_P first and then check whether X
13006 also has a scalar integer mode.
13008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13009 Alan Hayward <alan.hayward@arm.com>
13010 David Sherwood <david.sherwood@arm.com>
13012 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13013 Return a scalar_int_mode and assert that the size is greater than
13015 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13016 (op_by_pieces_d::op_by_pieces_d): Likewise.
13017 (op_by_pieces_d::run): Likewise.
13018 (can_store_by_pieces): Likewise.
13020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13021 Alan Hayward <alan.hayward@arm.com>
13022 David Sherwood <david.sherwood@arm.com>
13024 * combine.c (extract_left_shift): Add a mode argument and update
13026 (make_compound_operation_int): Change the type of the mode parameter
13027 to scalar_int_mode and update the call to extract_left_shift.
13029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13030 Alan Hayward <alan.hayward@arm.com>
13031 David Sherwood <david.sherwood@arm.com>
13033 * combine.c (simplify_and_const_int): Change the type of the mode
13034 parameter to scalar_int_mode.
13035 (simplify_and_const_int_1): Likewise. Update recursive call.
13037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13038 Alan Hayward <alan.hayward@arm.com>
13039 David Sherwood <david.sherwood@arm.com>
13041 * combine.c (simplify_compare_const): Check that the mode is a
13042 scalar_int_mode (rather than VOIDmode) before testing its
13044 (simplify_comparison): Move COMPARISON_P handling out of the
13045 loop and restrict the latter part of the loop to scalar_int_modes.
13046 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13047 and when considering SUBREG_REGs. Use is_int_mode instead of
13048 checking GET_MODE_CLASS against MODE_INT.
13050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13051 Alan Hayward <alan.hayward@arm.com>
13052 David Sherwood <david.sherwood@arm.com>
13054 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13055 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13056 shift_unit_mode and for modes involved in scalar shifts.
13058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13059 Alan Hayward <alan.hayward@arm.com>
13060 David Sherwood <david.sherwood@arm.com>
13062 * combine.c (force_int_to_mode): New function, split out from...
13063 (force_to_mode): ...here. Keep xmode up-to-date and use it
13064 instead of GET_MODE (x).
13066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13067 Alan Hayward <alan.hayward@arm.com>
13068 David Sherwood <david.sherwood@arm.com>
13070 * optabs-query.h (extraction_insn::struct_mode): Change type to
13071 opt_scalar_int_mode and update comment.
13072 (extraction_insn::field_mode): Change type to scalar_int_mode.
13073 (extraction_insn::pos_mode): Likewise.
13074 * combine.c (make_extraction): Update accordingly.
13075 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13076 (get_optab_extraction_insn): Likewise.
13077 * recog.c (simplify_while_replacing): Likewise.
13078 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13079 parameter to opt_scalar_int_mode.
13081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13082 Alan Hayward <alan.hayward@arm.com>
13083 David Sherwood <david.sherwood@arm.com>
13085 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13086 to a scalar_int_mode instead of a machine_mode.
13087 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13088 (get_best_mode): Return a boolean and use a pointer argument to store
13089 the selected mode. Replace the limit mode parameter with a bit limit.
13090 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13091 for the values returned by bit_field_mode_iterator::next_mode.
13092 (store_bit_field): Update call to get_best_mode.
13093 (store_fixed_bit_field): Likewise.
13094 (extract_fixed_bit_field): Likewise.
13095 * expr.c (optimize_bitfield_assignment_op): Likewise.
13096 * fold-const.c (optimize_bit_field_compare): Likewise.
13097 (fold_truth_andor_1): Likewise.
13098 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13099 Update for new type of m_mode.
13100 (get_best_mode): As above.
13102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13103 Alan Hayward <alan.hayward@arm.com>
13104 David Sherwood <david.sherwood@arm.com>
13106 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13107 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13108 (store_bit_field): Check is_a <scalar_int_mode> before calling
13109 strict_volatile_bitfield_p.
13110 (extract_bit_field): Likewise.
13112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13113 Alan Hayward <alan.hayward@arm.com>
13114 David Sherwood <david.sherwood@arm.com>
13116 * target.def (cstore_mode): Return a scalar_int_mode.
13117 * doc/tm.texi: Regenerate.
13118 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13119 * targhooks.h (default_cstore_mode): Likewise.
13120 * targhooks.c (default_cstore_mode): Likewise, using a forced
13122 * expmed.c (emit_cstore): Expect the target of the cstore to be
13125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13126 Alan Hayward <alan.hayward@arm.com>
13127 David Sherwood <david.sherwood@arm.com>
13129 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13131 (niter_desc): Likewise mode.
13132 (iv_analyze): Add a mode parameter.
13134 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13135 and change its type to scalar_int_mode.
13136 * loop-iv.c: Update commentary at head of file.
13137 (iv_constant): Pass the mode paraeter before the rtx it describes
13138 and change its type to scalar_int_mode. Remove VOIDmode handling.
13139 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13140 (iv_extend): Likewise.
13141 (shorten_into_mode): Likewise.
13142 (iv_add): Use scalar_int_mode.
13143 (iv_mult): Likewise.
13144 (iv_shift): Likewise.
13145 (canonicalize_iv_subregs): Likewise.
13146 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13147 it describes and change its mode to scalar_int_mode. Also change
13148 the type of the returned inner_mode to scalar_int_mode.
13149 (get_biv_step): Likewise, turning outer_mode from a pointer
13150 into a direct parameter. Update call to get_biv_step_1.
13151 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13152 iv_constant and get_biv_step.
13153 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13154 and change its type to scalar_int_mode. Don't initialise iv->mode
13155 to VOIDmode and remove later checks for its still being VOIDmode.
13156 Update calls to iv_analyze_op and iv_analyze_expr. Check
13157 is_a <scalar_int_mode> when changing the mode under consideration.
13158 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13159 Update call to iv_analyze_expr.
13160 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13161 inner register is not also a scalar_int_mode. Update call to
13163 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13164 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13165 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13166 separate mode class checks. Update calls to iv_analyze. Remove
13167 fix-up of VOIDmodes after iv_analyze_biv.
13168 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13169 don't have a scalar_int_mode. Update call to biv_p.
13171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13172 Alan Hayward <alan.hayward@arm.com>
13173 David Sherwood <david.sherwood@arm.com>
13175 * cfgexpand.c (convert_debug_memory_address): Use
13176 as_a <scalar_int_mode>.
13177 * combine.c (expand_compound_operation): Likewise.
13178 (make_extraction): Likewise.
13179 (change_zero_ext): Likewise.
13180 (simplify_comparison): Likewise.
13181 * cse.c (cse_insn): Likewise.
13182 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13183 (mem_loc_descriptor): Likewise.
13184 (loc_descriptor): Likewise.
13185 * expmed.c (init_expmed_one_mode): Likewise.
13186 (synth_mult): Likewise.
13187 (emit_store_flag_1): Likewise.
13188 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13189 of a comparison with size.
13190 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13191 (reduce_to_bit_field_precision): Likewise.
13192 * function.c (expand_function_end): Likewise.
13193 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13194 * loop-doloop.c (doloop_modify): Likewise.
13195 * optabs.c (expand_binop): Likewise.
13196 (expand_unop): Likewise.
13197 (expand_copysign_absneg): Likewise.
13198 (prepare_cmp_insn): Likewise.
13199 (maybe_legitimize_operand): Likewise.
13200 * recog.c (const_scalar_int_operand): Likewise.
13201 * rtlanal.c (get_address_mode): Likewise.
13202 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13203 (simplify_cond_clz_ctz): Likewise.
13204 * tree-nested.c (get_nl_goto_field): Likewise.
13205 * tree.c (build_vector_type_for_mode): Likewise.
13206 * var-tracking.c (use_narrower_mode): Likewise.
13208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13209 Alan Hayward <alan.hayward@arm.com>
13210 David Sherwood <david.sherwood@arm.com>
13212 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13213 * builtins.c (expand_builtin_signbit): Use it.
13214 * cfgexpand.c (expand_debug_expr): Likewise.
13215 * dojump.c (do_jump): Likewise.
13216 (do_compare_and_jump): Likewise.
13217 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13218 * expmed.c (make_tree): Likewise.
13219 * expr.c (expand_expr_real_2): Likewise.
13220 (expand_expr_real_1): Likewise.
13221 (try_casesi): Likewise.
13222 * fold-const-call.c (fold_const_call_ss): Likewise.
13223 * fold-const.c (unextend): Likewise.
13224 (extract_muldiv_1): Likewise.
13225 (fold_single_bit_test): Likewise.
13226 (native_encode_int): Likewise.
13227 (native_encode_string): Likewise.
13228 (native_interpret_int): Likewise.
13229 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13230 * internal-fn.c (expand_addsub_overflow): Likewise.
13231 (expand_neg_overflow): Likewise.
13232 (expand_mul_overflow): Likewise.
13233 (expand_arith_overflow): Likewise.
13234 * match.pd: Likewise.
13235 * stor-layout.c (layout_type): Likewise.
13236 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13237 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13238 * tree-ssanames.c (get_range_info): Likewise.
13239 * tree-switch-conversion.c (array_value_type) Likewise.
13240 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13241 (vect_recog_divmod_pattern): Likewise.
13242 (vect_recog_mixed_size_cond_pattern): Likewise.
13243 * tree-vrp.c (extract_range_basic): Likewise.
13244 (simplify_float_conversion_using_ranges): Likewise.
13245 * tree.c (int_fits_type_p): Likewise.
13246 * ubsan.c (instrument_bool_enum_load): Likewise.
13247 * varasm.c (mergeable_string_section): Likewise.
13248 (narrowing_initializer_constant_valid_p): Likewise.
13249 (output_constant): Likewise.
13251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13252 Alan Hayward <alan.hayward@arm.com>
13253 David Sherwood <david.sherwood@arm.com>
13255 * machmode.h (NARROWEST_INT_MODE): New macro.
13256 * expr.c (alignment_for_piecewise_move): Use it instead of
13257 GET_CLASS_NARROWEST_MODE (MODE_INT).
13258 (push_block): Likewise.
13259 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13261 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13264 Alan Hayward <alan.hayward@arm.com>
13265 David Sherwood <david.sherwood@arm.com>
13267 * postreload.c (move2add_valid_value_p): Change the type of the
13268 mode parameter to scalar_int_mode.
13269 (move2add_use_add2_insn): Add a mode parameter and use it instead
13271 (move2add_use_add3_insn): Likewise.
13272 (reload_cse_move2add): Update accordingly.
13274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13275 Alan Hayward <alan.hayward@arm.com>
13276 David Sherwood <david.sherwood@arm.com>
13278 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13280 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13281 * optabs.c (expand_unop): Likewise.
13283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13284 Alan Hayward <alan.hayward@arm.com>
13285 David Sherwood <david.sherwood@arm.com>
13287 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13288 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13289 (popcount_loc_descriptor): Likewise.
13290 (bswap_loc_descriptor): Likewise.
13291 (rotate_loc_descriptor): Likewise.
13292 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13293 calling the functions above.
13295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13296 Alan Hayward <alan.hayward@arm.com>
13297 David Sherwood <david.sherwood@arm.com>
13299 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13301 (try_combine): Likewise.
13302 (simplify_if_then_else): Likewise.
13303 * cse.c (cse_insn): Likewise.
13304 * dwarf2out.c (mem_loc_descriptor): Likewise.
13305 * emit-rtl.c (gen_lowpart_common): Likewise.
13306 * simplify-rtx.c (simplify_truncation): Likewise.
13307 (simplify_binary_operation_1): Likewise.
13308 (simplify_const_relational_operation): Likewise.
13309 (simplify_ternary_operation): Likewise.
13310 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13313 Alan Hayward <alan.hayward@arm.com>
13314 David Sherwood <david.sherwood@arm.com>
13316 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13317 * reload.c (push_reload): Likewise.
13318 (find_reloads): Likewise.
13320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13321 Alan Hayward <alan.hayward@arm.com>
13322 David Sherwood <david.sherwood@arm.com>
13324 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13325 (make_compound_operation_int): Likewise.
13326 (change_zero_ext): Likewise.
13327 * expr.c (convert_move): Likewise.
13328 (convert_modes): Likewise.
13329 * fwprop.c (forward_propagate_subreg): Likewise.
13330 * loop-iv.c (get_biv_step_1): Likewise.
13331 * optabs.c (widen_operand): Likewise.
13332 * postreload.c (move2add_valid_value_p): Likewise.
13333 * recog.c (simplify_while_replacing): Likewise.
13334 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13335 (simplify_binary_operation_1): Likewise. Remove redundant
13336 mode equality check.
13338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13339 Alan Hayward <alan.hayward@arm.com>
13340 David Sherwood <david.sherwood@arm.com>
13342 * combine.c (combine_simplify_rtx): Add checks for
13343 is_a <scalar_int_mode>.
13344 (simplify_if_then_else): Likewise.
13345 (make_field_assignment): Likewise.
13346 (simplify_comparison): Likewise.
13347 * ifcvt.c (noce_try_bitop): Likewise.
13348 * loop-invariant.c (canonicalize_address_mult): Likewise.
13349 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13352 Alan Hayward <alan.hayward@arm.com>
13353 David Sherwood <david.sherwood@arm.com>
13355 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13356 is_a <scalar_int_mode> instead of != BLKmode.
13358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13359 Alan Hayward <alan.hayward@arm.com>
13360 David Sherwood <david.sherwood@arm.com>
13362 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13363 instead of != VOIDmode.
13364 * combine.c (if_then_else_cond): Likewise.
13365 (change_zero_ext): Likewise.
13366 * dwarf2out.c (mem_loc_descriptor): Likewise.
13367 (loc_descriptor): Likewise.
13368 * rtlanal.c (canonicalize_condition): Likewise.
13369 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13372 Alan Hayward <alan.hayward@arm.com>
13373 David Sherwood <david.sherwood@arm.com>
13375 * simplify-rtx.c (simplify_binary_operation_1): Use
13376 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13379 Alan Hayward <alan.hayward@arm.com>
13380 David Sherwood <david.sherwood@arm.com>
13382 * wide-int.h (int_traits<unsigned char>) New class.
13383 (int_traits<unsigned short>) Likewise.
13384 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13385 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13387 * combine.c (set_nonzero_bits_and_sign_copies): Use
13388 is_a <scalar_int_mode>.
13389 (find_split_point): Likewise.
13390 (combine_simplify_rtx): Likewise.
13391 (simplify_logical): Likewise.
13392 (expand_compound_operation): Likewise.
13393 (expand_field_assignment): Likewise.
13394 (make_compound_operation): Likewise.
13395 (extended_count): Likewise.
13396 (change_zero_ext): Likewise.
13397 (simplify_comparison): Likewise.
13398 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13399 (ucompare_loc_descriptor): Likewise.
13400 (minmax_loc_descriptor): Likewise.
13401 (mem_loc_descriptor): Likewise.
13402 (loc_descriptor): Likewise.
13403 * expmed.c (init_expmed_one_mode): Likewise.
13404 * lra-constraints.c (lra_constraint_offset): Likewise.
13405 * optabs.c (prepare_libcall_arg): Likewise.
13406 * postreload.c (move2add_note_store): Likewise.
13407 * reload.c (operands_match_p): Likewise.
13408 * rtl.h (load_extend_op): Likewise.
13409 * rtlhooks.c (gen_lowpart_general): Likewise.
13410 * simplify-rtx.c (simplify_truncation): Likewise.
13411 (simplify_unary_operation_1): Likewise.
13412 (simplify_binary_operation_1): Likewise.
13413 (simplify_const_binary_operation): Likewise.
13414 (simplify_const_relational_operation): Likewise.
13415 (simplify_subreg): Likewise.
13416 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13417 * var-tracking.c (adjust_mems): Likewise.
13418 (prepare_call_arguments): Likewise.
13420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13421 Alan Hayward <alan.hayward@arm.com>
13422 David Sherwood <david.sherwood@arm.com>
13424 * machmode.h (is_int_mode): New fuction.
13425 * combine.c (find_split_point): Use it.
13426 (combine_simplify_rtx): Likewise.
13427 (simplify_if_then_else): Likewise.
13428 (simplify_set): Likewise.
13429 (simplify_shift_const_1): Likewise.
13430 (simplify_comparison): Likewise.
13431 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13432 * cse.c (notreg_cost): Likewise.
13433 (cse_insn): Likewise.
13434 * cselib.c (cselib_lookup_1): Likewise.
13435 * dojump.c (do_jump_1): Likewise.
13436 (do_compare_rtx_and_jump): Likewise.
13437 * dse.c (get_call_args): Likewise.
13438 * dwarf2out.c (rtl_for_decl_init): Likewise.
13439 (native_encode_initializer): Likewise.
13440 * expmed.c (emit_store_flag_1): Likewise.
13441 (emit_store_flag): Likewise.
13442 * expr.c (convert_modes): Likewise.
13443 (store_field): Likewise.
13444 (expand_expr_real_1): Likewise.
13445 * fold-const.c (fold_read_from_constant_string): Likewise.
13446 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13447 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13448 * optabs.c (expand_binop): Likewise.
13449 (expand_unop): Likewise.
13450 (expand_abs_nojump): Likewise.
13451 (expand_one_cmpl_abs_nojump): Likewise.
13452 * simplify-rtx.c (mode_signbit_p): Likewise.
13453 (val_signbit_p): Likewise.
13454 (val_signbit_known_set_p): Likewise.
13455 (val_signbit_known_clear_p): Likewise.
13456 (simplify_relational_operation_1): Likewise.
13457 * tree.c (vector_type_mode): Likewise.
13459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13460 Alan Hayward <alan.hayward@arm.com>
13461 David Sherwood <david.sherwood@arm.com>
13463 * machmode.h (smallest_mode_for_size): Fix formatting.
13464 (smallest_int_mode_for_size): New function.
13465 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
13466 instead of smallest_mode_for_size.
13467 * combine.c (make_extraction): Likewise.
13468 * config/arc/arc.c (arc_expand_movmem): Likewise.
13469 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
13470 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13471 * config/s390/s390.c (s390_expand_insv): Likewise.
13472 * config/sparc/sparc.c (assign_int_registers): Likewise.
13473 * config/spu/spu.c (spu_function_value): Likewise.
13474 (spu_function_arg): Likewise.
13475 * coverage.c (get_gcov_type): Likewise.
13476 (get_gcov_unsigned_t): Likewise.
13477 * dse.c (find_shift_sequence): Likewise.
13478 * expmed.c (store_bit_field_1): Likewise.
13479 * expr.c (convert_move): Likewise.
13480 (store_field): Likewise.
13481 * internal-fn.c (expand_arith_overflow): Likewise.
13482 * optabs-query.c (get_best_extraction_insn): Likewise.
13483 * optabs.c (expand_twoval_binop_libfunc): Likewise.
13484 * stor-layout.c (layout_type): Likewise.
13485 (initialize_sizetypes): Likewise.
13486 * targhooks.c (default_get_mask_mode): Likewise.
13487 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
13489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13490 Alan Hayward <alan.hayward@arm.com>
13491 David Sherwood <david.sherwood@arm.com>
13493 * machmode.h (opt_mode::else_blk): New function.
13494 (int_mode_for_mode): Declare.
13495 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13496 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13498 * cfgexpand.c (expand_debug_expr): Likewise.
13499 * combine.c (gen_lowpart_or_truncate): Likewise.
13500 (gen_lowpart_for_combine): Likewise.
13501 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13502 * config/avr/avr.c (avr_to_int_mode): Likewise.
13503 (avr_out_plus_1): Likewise.
13504 (avr_out_plus): Likewise.
13505 (avr_out_round): Likewise.
13506 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13507 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13508 (s390_expand_vcond): Likewise.
13509 * config/spu/spu.c (spu_split_immediate): Likewise.
13510 (spu_expand_mov): Likewise.
13511 * dse.c (get_stored_val): Likewise.
13512 * expmed.c (store_bit_field_1): Likewise.
13513 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13515 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13516 (extract_low_bits): Likewise.
13517 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13518 handling rather than repeating the check.
13519 (emit_group_store): Likewise.
13520 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13521 * optabs.c (expand_absneg_bit): Likewise.
13522 (expand_copysign_absneg): Likewise.
13523 (expand_copysign_bit): Likewise.
13524 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13525 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13526 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13527 * var-tracking.c (prepare_call_arguments): Likewise.
13528 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13529 int_mode_for_mode instead of mode_for_size.
13530 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13533 Alan Hayward <alan.hayward@arm.com>
13534 David Sherwood <david.sherwood@arm.com>
13536 * machmode.h (int_mode_for_size): New function.
13537 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13538 instead of mode_for_size.
13539 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13541 * combine.c (expand_field_assignment): Use int_mode_for_size
13542 instead of mode_for_size.
13543 (make_extraction): Likewise.
13544 (simplify_shift_const_1): Likewise.
13545 (simplify_comparison): Likewise.
13546 * dojump.c (do_jump): Likewise.
13547 * dwarf2out.c (mem_loc_descriptor): Likewise.
13548 * emit-rtl.c (init_derived_machine_modes): Likewise.
13549 * expmed.c (flip_storage_order): Likewise.
13550 (convert_extracted_bit_field): Likewise.
13551 * expr.c (copy_blkmode_from_reg): Likewise.
13552 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13553 * internal-fn.c (expand_mul_overflow): Likewise.
13554 * lower-subreg.c (simple_move): Likewise.
13555 * optabs-libfuncs.c (init_optabs): Likewise.
13556 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13557 * tree.c (vector_type_mode): Likewise.
13558 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13559 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13560 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13561 * tree-vect-stmts.c (vectorizable_load): Likewise.
13562 (vectorizable_store): Likewise.
13564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13565 Alan Hayward <alan.hayward@arm.com>
13566 David Sherwood <david.sherwood@arm.com>
13568 * coretypes.h (pod_mode): New type.
13569 (scalar_int_mode_pod): New typedef.
13570 * machmode.h (pod_mode): New class.
13571 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13572 * genmodes.c (emit_mode_int_n): Update accordingly.
13573 * lower-subreg.h (target_lower_subreg): Change type to
13574 scalar_int_mode_pod.
13575 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13576 scalar_int_mode_pod.
13578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13579 Alan Hayward <alan.hayward@arm.com>
13580 David Sherwood <david.sherwood@arm.com>
13582 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13583 machine_mode to scalar_int_mode.
13584 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13585 (rs6000_option_override_internal): Remove cast to int.
13586 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13587 machine_mode to scalar_int_mode.
13588 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13589 (rs6000_option_override_internal): Remove cast to int.
13590 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13591 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13593 * config/s390/s390.c (s390_expand_builtin): Likewise.
13594 * coretypes.h (scalar_int_mode): New type.
13595 (opt_scalar_int_mode): New typedef.
13596 * machmode.h (scalar_int_mode): New class.
13597 (scalar_int_mode::includes_p): New function.
13598 (byte_mode): Change type to scalar_int_mode.
13599 (word_mode): Likewise.
13600 (ptr_mode): Likewise.
13601 * emit-rtl.c (byte_mode): Likewise.
13602 (word_mode): Likewise.
13603 (ptr_mode): Likewise.
13604 (init_derived_machine_modes): Update accordingly.
13605 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13606 and MODE_PARTIAL_INT.
13607 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13608 opt_scalar_int_mode.
13610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13611 Alan Hayward <alan.hayward@arm.com>
13612 David Sherwood <david.sherwood@arm.com>
13614 * target.def (libgcc_floating_mode_supported_p): Take a
13616 * doc/tm.texi: Regenerate.
13617 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13619 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13620 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13624 Alan Hayward <alan.hayward@arm.com>
13625 David Sherwood <david.sherwood@arm.com>
13627 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13628 * doc/tm.texi: Regenerate.
13629 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13630 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13631 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13632 * targhooks.h (default_floatn_mode): Likewise.
13633 * targhooks.c (default_floatn_mode): Likewise.
13634 * tree.c (build_common_tree_nodes): Update accordingly.
13636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13637 Alan Hayward <alan.hayward@arm.com>
13638 David Sherwood <david.sherwood@arm.com>
13640 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13641 (mode_iterator::iterate_p): Likewise.
13642 (mode_iterator::get_wider): Likewise.
13643 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13646 Alan Hayward <alan.hayward@arm.com>
13647 David Sherwood <david.sherwood@arm.com>
13649 * coretypes.h (opt_scalar_float_mode): New typedef.
13650 * machmode.h (float_mode_for_size): New function.
13651 * emit-rtl.c (double_mode): Delete.
13652 (init_emit_once): Use float_mode_for_size.
13653 * stor-layout.c (layout_type): Likewise.
13654 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13657 Alan Hayward <alan.hayward@arm.com>
13658 David Sherwood <david.sherwood@arm.com>
13660 * output.h (assemble_real): Take a scalar_float_mode.
13661 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13662 * config/arm/arm.md (consttable_4): Likewise.
13663 (consttable_8): Likewise.
13664 (consttable_16): Likewise.
13665 * config/mips/mips.md (consttable_float): Likewise.
13666 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13667 * varasm.c (assemble_real): Take a scalar_float_mode.
13668 (output_constant_pool_2): Update accordingly.
13669 (output_constant): Likewise.
13671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13672 Alan Hayward <alan.hayward@arm.com>
13673 David Sherwood <david.sherwood@arm.com>
13675 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13676 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13677 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13678 (native_encode_real): Likewise.
13679 (native_interpret_real): Likewise.
13680 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13681 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13684 Alan Hayward <alan.hayward@arm.com>
13685 David Sherwood <david.sherwood@arm.com>
13687 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13688 <scalar_float_mode>. Simplify.
13689 (gen_extend_conv_libfunc): Likewise.
13691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13692 Alan Hayward <alan.hayward@arm.com>
13693 David Sherwood <david.sherwood@arm.com>
13695 * coretypes.h (scalar_float_mode): New type.
13696 * machmode.h (mode_traits::from_int): Use machine_mode if
13697 USE_ENUM_MODES is defined.
13698 (is_a): New function.
13700 (dyn_cast): Likewise.
13701 (scalar_float_mode): New class.
13702 (scalar_float_mode::includes_p): New function.
13703 (is_float_mode): Likewise.
13704 * gdbhooks.py (MachineModePrinter): New class.
13705 (build_pretty_printer): Use it for scalar_float_mode.
13706 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13707 (format_helper::format_helper): Turn into a template.
13708 * genmodes.c (get_mode_class): New function.
13709 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13710 or machine_mode if none.
13711 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13712 as_a <scalar_float_mode>.
13713 * dwarf2out.c (mem_loc_descriptor): Likewise.
13714 (insert_float): Likewise.
13715 (add_const_value_attribute): Likewise.
13716 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13717 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13718 (expand_unop): Update accordingly.
13719 (expand_abs_nojump): Likewise.
13720 (expand_copysign_absneg): Take a scalar_float_mode.
13721 (expand_copysign_bit): Likewise.
13722 (expand_copysign): Update accordingly.
13724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13725 Alan Hayward <alan.hayward@arm.com>
13726 David Sherwood <david.sherwood@arm.com>
13728 * coretypes.h (opt_mode): New class.
13729 * machmode.h (opt_mode): Likewise.
13730 (opt_mode::else_void): New function.
13731 (opt_mode::require): Likewise.
13732 (opt_mode::exists): Likewise.
13733 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13734 (GET_MODE_2XWIDER_MODE): Likewise.
13735 (mode_iterator::get_wider): Update accordingly.
13736 (mode_iterator::get_2xwider): Likewise.
13737 (mode_iterator::get_known_wider): Likewise, turning into a template.
13738 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13739 forcing a wider mode to exist.
13740 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13741 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13742 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13743 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13744 * lower-subreg.c (init_lower_subreg): Likewise.
13745 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13746 on the final iteration.
13747 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13748 a wider mode exists before asking for a move pattern.
13749 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13750 forcing a wider mode to exist.
13751 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13752 returning false if no such mode exists.
13753 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13754 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13755 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13756 Avoid checking for a MODE_INT if we already know the mode is not a
13758 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13759 forcing a wider mode to exist.
13760 (expmed_mult_highpart_optab): Likewise.
13761 (expmed_mult_highpart): Likewise.
13762 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13764 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13765 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13766 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13767 * internal-fn.c (expand_mul_overflow): Update use of
13768 GET_MODE_2XWIDER_MODE.
13769 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13770 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13771 GET_MODE_WIDER_MODE.
13772 (convert_plusminus_to_widen): Likewise.
13773 * tree-switch-conversion.c (array_value_type): Likewise.
13774 * var-tracking.c (emit_note_insn_var_location): Likewise.
13775 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13776 Return false inside rather than outside the loop if no wider mode
13778 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13779 and GET_MODE_2XWIDER_MODE
13780 (can_compare_p): Use else_void.
13781 * gdbhooks.py (OptMachineModePrinter): New class.
13782 (build_pretty_printer): Use it for opt_mode.
13784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13785 Alan Hayward <alan.hayward@arm.com>
13786 David Sherwood <david.sherwood@arm.com>
13788 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13789 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13792 Alan Hayward <alan.hayward@arm.com>
13793 David Sherwood <david.sherwood@arm.com>
13795 * machmode.h (mode_traits): New structure.
13796 (get_narrowest_mode): New function.
13797 (mode_iterator::start): Likewise.
13798 (mode_iterator::iterate_p): Likewise.
13799 (mode_iterator::get_wider): Likewise.
13800 (mode_iterator::get_known_wider): Likewise.
13801 (mode_iterator::get_2xwider): Likewise.
13802 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13803 (FOR_EACH_MODE): Likewise.
13804 (FOR_EACH_MODE_FROM): Likewise.
13805 (FOR_EACH_MODE_UNTIL): Likewise.
13806 (FOR_EACH_WIDER_MODE): Likewise.
13807 (FOR_EACH_2XWIDER_MODE): Likewise.
13808 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13809 * combine.c (simplify_comparison): Likewise
13810 * config/i386/i386.c (type_natural_mode): Likewise.
13811 * cse.c (cse_insn): Likewise.
13812 * dse.c (find_shift_sequence): Likewise.
13813 * emit-rtl.c (init_derived_machine_modes): Likewise.
13814 (init_emit_once): Likewise.
13815 * explow.c (hard_function_value): Likewise.
13816 * expmed.c (extract_fixed_bit_field_1): Likewise.
13817 (extract_bit_field_1): Likewise.
13818 (expand_divmod): Likewise.
13819 (emit_store_flag_1): Likewise.
13820 * expr.c (init_expr_target): Likewise.
13821 (convert_move): Likewise.
13822 (alignment_for_piecewise_move): Likewise.
13823 (widest_int_mode_for_size): Likewise.
13824 (emit_block_move_via_movmem): Likewise.
13825 (copy_blkmode_to_reg): Likewise.
13826 (set_storage_via_setmem): Likewise.
13827 (compress_float_constant): Likewise.
13828 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13829 * optabs-query.c (get_best_extraction_insn): Likewise.
13830 * optabs.c (expand_binop): Likewise.
13831 (expand_twoval_unop): Likewise.
13832 (expand_twoval_binop): Likewise.
13833 (widen_leading): Likewise.
13834 (widen_bswap): Likewise.
13835 (expand_parity): Likewise.
13836 (expand_unop): Likewise.
13837 (prepare_cmp_insn): Likewise.
13838 (prepare_float_lib_cmp): Likewise.
13839 (expand_float): Likewise.
13840 (expand_fix): Likewise.
13841 (expand_sfix_optab): Likewise.
13842 * postreload.c (move2add_use_add2_insn): Likewise.
13843 * reg-stack.c (reg_to_stack): Likewise.
13844 * reginfo.c (choose_hard_reg_mode): Likewise.
13845 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13846 * stor-layout.c (mode_for_size): Likewise.
13847 (smallest_mode_for_size): Likewise.
13848 (mode_for_vector): Likewise.
13849 (finish_bitfield_representative): Likewise.
13850 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13851 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13852 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13853 * var-tracking.c (prepare_call_arguments): Likewise.
13855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13856 Alan Hayward <alan.hayward@arm.com>
13857 David Sherwood <david.sherwood@arm.com>
13859 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13860 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13861 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13862 * machmode.h (mode_size): Move earlier in file.
13863 (mode_precision): Likewise.
13864 (mode_inner): Likewise.
13865 (mode_nunits): Likewise.
13866 (mode_unit_size): Likewise.
13867 (unit_unit_precision): Likewise.
13868 (mode_wider): Likewise.
13869 (mode_2xwider): Likewise.
13870 (machine_mode): New class.
13871 (mode_to_bytes): New function.
13872 (mode_to_bits): Likewise.
13873 (mode_to_precision): Likewise.
13874 (mode_to_inner): Likewise.
13875 (mode_to_unit_size): Likewise.
13876 (mode_to_unit_precision): Likewise.
13877 (mode_to_nunits): Likewise.
13878 (GET_MODE_SIZE): Use mode_to_bytes.
13879 (GET_MODE_BITSIZE): Use mode_to_bits.
13880 (GET_MODE_PRECISION): Use mode_to_precision.
13881 (GET_MODE_INNER): Use mode_to_inner.
13882 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13883 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13884 (GET_MODE_NUNITS): Use mode_to_nunits.
13885 * system.h (ALWAYS_INLINE): New macro.
13886 * config/powerpcspe/powerpcspe-c.c
13887 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13888 int for arg1_mode and arg2_mode.
13890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13891 Alan Hayward <alan.hayward@arm.com>
13892 David Sherwood <david.sherwood@arm.com>
13894 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13895 Prefix mode names with E_ in case statements.
13896 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13897 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13898 (aarch64_split_simd_move): Likewise.
13899 (aarch64_gen_storewb_pair): Likewise.
13900 (aarch64_gen_loadwb_pair): Likewise.
13901 (aarch64_gen_store_pair): Likewise.
13902 (aarch64_gen_load_pair): Likewise.
13903 (aarch64_get_condition_code_1): Likewise.
13904 (aarch64_constant_pool_reload_icode): Likewise.
13905 (get_rsqrte_type): Likewise.
13906 (get_rsqrts_type): Likewise.
13907 (get_recpe_type): Likewise.
13908 (get_recps_type): Likewise.
13909 (aarch64_gimplify_va_arg_expr): Likewise.
13910 (aarch64_simd_container_mode): Likewise.
13911 (aarch64_emit_load_exclusive): Likewise.
13912 (aarch64_emit_store_exclusive): Likewise.
13913 (aarch64_expand_compare_and_swap): Likewise.
13914 (aarch64_gen_atomic_cas): Likewise.
13915 (aarch64_emit_bic): Likewise.
13916 (aarch64_emit_atomic_swap): Likewise.
13917 (aarch64_emit_atomic_load_op): Likewise.
13918 (aarch64_evpc_trn): Likewise.
13919 (aarch64_evpc_uzp): Likewise.
13920 (aarch64_evpc_zip): Likewise.
13921 (aarch64_evpc_ext): Likewise.
13922 (aarch64_evpc_rev): Likewise.
13923 (aarch64_evpc_dup): Likewise.
13924 (aarch64_gen_ccmp_first): Likewise.
13925 (aarch64_gen_ccmp_next): Likewise.
13926 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13927 (alpha_emit_xfloating_libcall): Likewise.
13928 (emit_insxl): Likewise.
13929 (alpha_arg_type): Likewise.
13930 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13931 (arc_preferred_simd_mode): Likewise.
13932 (arc_secondary_reload): Likewise.
13933 (get_arc_condition_code): Likewise.
13934 (arc_print_operand): Likewise.
13935 (arc_legitimate_constant_p): Likewise.
13936 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13937 * config/arc/arc.md (casesi_load): Likewise.
13938 (casesi_compact_jump): Likewise.
13939 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13940 (cc_use_register): Likewise.
13941 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13942 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13943 (arm_init_iwmmxt_builtins): Likewise.
13944 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13945 (neon_expand_vector_init): Likewise.
13946 (arm_attr_length_move_neon): Likewise.
13947 (maybe_get_arm_condition_code): Likewise.
13948 (arm_emit_vector_const): Likewise.
13949 (arm_preferred_simd_mode): Likewise.
13950 (arm_output_iwmmxt_tinsr): Likewise.
13951 (thumb1_output_casesi): Likewise.
13952 (thumb2_output_casesi): Likewise.
13953 (arm_emit_load_exclusive): Likewise.
13954 (arm_emit_store_exclusive): Likewise.
13955 (arm_expand_compare_and_swap): Likewise.
13956 (arm_evpc_neon_vuzp): Likewise.
13957 (arm_evpc_neon_vzip): Likewise.
13958 (arm_evpc_neon_vrev): Likewise.
13959 (arm_evpc_neon_vtrn): Likewise.
13960 (arm_evpc_neon_vext): Likewise.
13961 (arm_validize_comparison): Likewise.
13962 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13963 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13964 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13965 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13966 (c6x_preferred_simd_mode): Likewise.
13967 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13968 (epiphany_rtx_costs): Likewise.
13969 * config/epiphany/predicates.md (proper_comparison_operator):
13971 * config/frv/frv.c (condexec_memory_operand): Likewise.
13972 (frv_emit_move): Likewise.
13973 (output_move_single): Likewise.
13974 (output_condmove_single): Likewise.
13975 (frv_hard_regno_mode_ok): Likewise.
13976 (frv_matching_accg_mode): Likewise.
13977 * config/h8300/h8300.c (split_adds_subs): Likewise.
13978 (h8300_rtx_costs): Likewise.
13979 (h8300_print_operand): Likewise.
13980 (compute_mov_length): Likewise.
13981 (output_logical_op): Likewise.
13982 (compute_logical_op_length): Likewise.
13983 (compute_logical_op_cc): Likewise.
13984 (h8300_shift_needs_scratch_p): Likewise.
13985 (output_a_shift): Likewise.
13986 (compute_a_shift_length): Likewise.
13987 (compute_a_shift_cc): Likewise.
13988 (expand_a_rotate): Likewise.
13989 (output_a_rotate): Likewise.
13990 * config/i386/i386.c (classify_argument): Likewise.
13991 (function_arg_advance_32): Likewise.
13992 (function_arg_32): Likewise.
13993 (function_arg_64): Likewise.
13994 (function_value_64): Likewise.
13995 (ix86_gimplify_va_arg): Likewise.
13996 (ix86_legitimate_constant_p): Likewise.
13997 (put_condition_code): Likewise.
13998 (split_double_mode): Likewise.
13999 (ix86_avx256_split_vector_move_misalign): Likewise.
14000 (ix86_expand_vector_logical_operator): Likewise.
14001 (ix86_split_idivmod): Likewise.
14002 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14003 (ix86_build_const_vector): Likewise.
14004 (ix86_build_signbit_mask): Likewise.
14005 (ix86_match_ccmode): Likewise.
14006 (ix86_cc_modes_compatible): Likewise.
14007 (ix86_expand_branch): Likewise.
14008 (ix86_expand_sse_cmp): Likewise.
14009 (ix86_expand_sse_movcc): Likewise.
14010 (ix86_expand_int_sse_cmp): Likewise.
14011 (ix86_expand_vec_perm_vpermi2): Likewise.
14012 (ix86_expand_vec_perm): Likewise.
14013 (ix86_expand_sse_unpack): Likewise.
14014 (ix86_expand_int_addcc): Likewise.
14015 (ix86_split_to_parts): Likewise.
14016 (ix86_vectorize_builtin_gather): Likewise.
14017 (ix86_vectorize_builtin_scatter): Likewise.
14018 (avx_vpermilp_parallel): Likewise.
14019 (inline_memory_move_cost): Likewise.
14020 (ix86_tieable_integer_mode_p): Likewise.
14021 (x86_maybe_negate_const_int): Likewise.
14022 (ix86_expand_vector_init_duplicate): Likewise.
14023 (ix86_expand_vector_init_one_nonzero): Likewise.
14024 (ix86_expand_vector_init_one_var): Likewise.
14025 (ix86_expand_vector_init_concat): Likewise.
14026 (ix86_expand_vector_init_interleave): Likewise.
14027 (ix86_expand_vector_init_general): Likewise.
14028 (ix86_expand_vector_set): Likewise.
14029 (ix86_expand_vector_extract): Likewise.
14030 (emit_reduc_half): Likewise.
14031 (ix86_emit_i387_round): Likewise.
14032 (ix86_mangle_type): Likewise.
14033 (ix86_expand_round_sse4): Likewise.
14034 (expand_vec_perm_blend): Likewise.
14035 (canonicalize_vector_int_perm): Likewise.
14036 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14037 (expand_vec_perm_1): Likewise.
14038 (expand_vec_perm_interleave3): Likewise.
14039 (expand_vec_perm_even_odd_pack): Likewise.
14040 (expand_vec_perm_even_odd_1): Likewise.
14041 (expand_vec_perm_broadcast_1): Likewise.
14042 (ix86_vectorize_vec_perm_const_ok): Likewise.
14043 (ix86_expand_vecop_qihi): Likewise.
14044 (ix86_expand_mul_widen_hilo): Likewise.
14045 (ix86_expand_sse2_abs): Likewise.
14046 (ix86_expand_pextr): Likewise.
14047 (ix86_expand_pinsr): Likewise.
14048 (ix86_preferred_simd_mode): Likewise.
14049 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14050 * config/i386/sse.md (*andnot<mode>3): Likewise.
14051 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14052 (*<code><mode>3): Likewise.
14053 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14054 (ia64_expand_atomic_op): Likewise.
14055 (ia64_arg_type): Likewise.
14056 (ia64_mode_to_int): Likewise.
14057 (ia64_scalar_mode_supported_p): Likewise.
14058 (ia64_vector_mode_supported_p): Likewise.
14059 (expand_vec_perm_broadcast): Likewise.
14060 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14061 (iq2000_function_arg_advance): Likewise.
14062 (iq2000_function_arg): Likewise.
14063 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14064 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14065 (m68k_libcall_value): Likewise.
14066 (m68k_function_value): Likewise.
14067 (sched_attr_op_type): Likewise.
14068 * config/mcore/mcore.c (mcore_output_move): Likewise.
14069 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14071 (microblaze_function_arg): Likewise.
14072 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14073 (mips_print_operand): Likewise.
14074 (mips_mode_ok_for_mov_fmt_p): Likewise.
14075 (mips_vector_mode_supported_p): Likewise.
14076 (mips_preferred_simd_mode): Likewise.
14077 (mips_expand_vpc_loongson_even_odd): Likewise.
14078 (mips_expand_vec_unpack): Likewise.
14079 (mips_expand_vi_broadcast): Likewise.
14080 (mips_expand_vector_init): Likewise.
14081 (mips_expand_vec_reduc): Likewise.
14082 (mips_expand_msa_cmp): Likewise.
14083 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14084 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14085 (cc_flags_for_mode): Likewise.
14086 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14087 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14088 (nds32_output_casesi_pc_relative): Likewise.
14089 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14090 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14091 (nvptx_gen_unpack): Likewise.
14092 (nvptx_gen_pack): Likewise.
14093 (nvptx_gen_shuffle): Likewise.
14094 (nvptx_gen_wcast): Likewise.
14095 (nvptx_preferred_simd_mode): Likewise.
14096 * config/pa/pa.c (pa_secondary_reload): Likewise.
14097 * config/pa/predicates.md (base14_operand): Likewise.
14098 * config/powerpcspe/powerpcspe-c.c
14099 (altivec_resolve_overloaded_builtin): Likewise.
14100 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14102 (rs6000_preferred_simd_mode): Likewise.
14103 (output_vec_const_move): Likewise.
14104 (rs6000_expand_vector_extract): Likewise.
14105 (rs6000_split_vec_extract_var): Likewise.
14106 (reg_offset_addressing_ok_p): Likewise.
14107 (rs6000_legitimate_offset_address_p): Likewise.
14108 (rs6000_legitimize_address): Likewise.
14109 (rs6000_emit_set_const): Likewise.
14110 (rs6000_const_vec): Likewise.
14111 (rs6000_emit_move): Likewise.
14112 (spe_build_register_parallel): Likewise.
14113 (rs6000_darwin64_record_arg_recurse): Likewise.
14114 (swap_selector_for_mode): Likewise.
14115 (spe_init_builtins): Likewise.
14116 (paired_init_builtins): Likewise.
14117 (altivec_init_builtins): Likewise.
14118 (do_load_for_compare): Likewise.
14119 (rs6000_generate_compare): Likewise.
14120 (rs6000_expand_float128_convert): Likewise.
14121 (emit_load_locked): Likewise.
14122 (emit_store_conditional): Likewise.
14123 (rs6000_output_function_epilogue): Likewise.
14124 (rs6000_handle_altivec_attribute): Likewise.
14125 (rs6000_function_value): Likewise.
14126 (emit_fusion_gpr_load): Likewise.
14127 (emit_fusion_p9_load): Likewise.
14128 (emit_fusion_p9_store): Likewise.
14129 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14130 (fusion_gpr_mem_load): Likewise.
14131 (fusion_addis_mem_combo_load): Likewise.
14132 (fusion_addis_mem_combo_store): Likewise.
14133 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14134 (fusion_gpr_mem_load): Likewise.
14135 (fusion_addis_mem_combo_load): Likewise.
14136 (fusion_addis_mem_combo_store): Likewise.
14137 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14139 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14140 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14141 (rs6000_preferred_simd_mode): Likewise.
14142 (output_vec_const_move): Likewise.
14143 (rs6000_expand_vector_extract): Likewise.
14144 (rs6000_split_vec_extract_var): Likewise.
14145 (reg_offset_addressing_ok_p): Likewise.
14146 (rs6000_legitimate_offset_address_p): Likewise.
14147 (rs6000_legitimize_address): Likewise.
14148 (rs6000_emit_set_const): Likewise.
14149 (rs6000_const_vec): Likewise.
14150 (rs6000_emit_move): Likewise.
14151 (rs6000_darwin64_record_arg_recurse): Likewise.
14152 (swap_selector_for_mode): Likewise.
14153 (paired_init_builtins): Likewise.
14154 (altivec_init_builtins): Likewise.
14155 (rs6000_expand_float128_convert): Likewise.
14156 (emit_load_locked): Likewise.
14157 (emit_store_conditional): Likewise.
14158 (rs6000_output_function_epilogue): Likewise.
14159 (rs6000_handle_altivec_attribute): Likewise.
14160 (rs6000_function_value): Likewise.
14161 (emit_fusion_gpr_load): Likewise.
14162 (emit_fusion_p9_load): Likewise.
14163 (emit_fusion_p9_store): Likewise.
14164 * config/rx/rx.c (rx_gen_move_template): Likewise.
14165 (flags_from_mode): Likewise.
14166 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14167 (s390_slb_comparison): Likewise.
14168 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14169 (s390_vector_mode_supported_p): Likewise.
14170 (s390_cc_modes_compatible): Likewise.
14171 (s390_match_ccmode_set): Likewise.
14172 (s390_canonicalize_comparison): Likewise.
14173 (s390_emit_compare_and_swap): Likewise.
14174 (s390_branch_condition_mask): Likewise.
14175 (s390_rtx_costs): Likewise.
14176 (s390_secondary_reload): Likewise.
14177 (__SECONDARY_RELOAD_CASE): Likewise.
14178 (s390_expand_cs): Likewise.
14179 (s390_preferred_simd_mode): Likewise.
14180 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14181 * config/sh/sh.c (sh_print_operand): Likewise.
14182 (dump_table): Likewise.
14183 (sh_secondary_reload): Likewise.
14184 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14185 * config/sh/sh.md (casesi_worker_1): Likewise.
14186 (casesi_worker_2): Likewise.
14187 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14188 (fcc_comparison_operator): Likewise.
14189 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14190 (emit_soft_tfmode_cvt): Likewise.
14191 (sparc_preferred_simd_mode): Likewise.
14192 (output_cbranch): Likewise.
14193 (sparc_print_operand): Likewise.
14194 (sparc_expand_vec_perm_bmask): Likewise.
14195 (vector_init_bshuffle): Likewise.
14196 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14197 (spu_vector_mode_supported_p): Likewise.
14198 (spu_expand_insv): Likewise.
14199 (spu_emit_branch_or_set): Likewise.
14200 (spu_handle_vector_attribute): Likewise.
14201 (spu_builtin_splats): Likewise.
14202 (spu_builtin_extract): Likewise.
14203 (spu_builtin_promote): Likewise.
14204 (spu_expand_sign_extend): Likewise.
14205 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14206 (tilegx_simd_int): Likewise.
14207 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14208 (tilepro_simd_int): Likewise.
14209 * config/v850/v850.c (const_double_split): Likewise.
14210 (v850_print_operand): Likewise.
14211 (ep_memory_offset): Likewise.
14212 * config/vax/vax.c (vax_rtx_costs): Likewise.
14213 (vax_output_int_move): Likewise.
14214 (vax_output_int_add): Likewise.
14215 (vax_output_int_subtract): Likewise.
14216 * config/visium/predicates.md (visium_branch_operator): Likewise.
14217 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14218 (visium_print_operand_address): Likewise.
14219 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14220 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14221 (xtensa_expand_conditional_branch): Likewise.
14222 (xtensa_copy_incoming_a7): Likewise.
14223 (xtensa_output_literal): Likewise.
14224 * dfp.c (decimal_real_maxval): Likewise.
14225 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14228 Alan Hayward <alan.hayward@arm.com>
14229 David Sherwood <david.sherwood@arm.com>
14231 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14232 (mode_nunits_inline): Likewise.
14233 (mode_inner_inline): Likewise.
14234 (mode_unit_size_inline): Likewise.
14235 (mode_unit_precision_inline): Likewise.
14236 (emit_insn_modes_h): Likewise. Also emit a #define of the
14238 (emit_mode_wider): Add an E_ prefix to mode names.
14239 (emit_mode_complex): Likewise.
14240 (emit_mode_inner): Likewise.
14241 (emit_mode_adjustments): Likewise.
14242 (emit_mode_int_n): Likewise.
14243 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14244 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14245 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14246 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14247 (CRC32_BUILTIN, ENTRY): Likewise.
14248 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14249 (aarch64_pop_regs): Likewise.
14250 (aarch64_process_components): Likewise.
14251 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14252 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14253 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14254 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14255 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14256 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14258 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14259 (pcmpistr): Likewise.
14260 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14261 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14262 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14264 * config/rl78/rl78.c (mduc_regs): Likewise.
14265 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14266 (htm_expand_builtin): Likewise.
14267 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14268 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14269 * config/xtensa/xtensa.c (print_operand): Likewise.
14270 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14271 (NUM_MODE_VECTOR_INT): Likewise.
14272 * genoutput.c (null_operand): Likewise.
14273 (output_operand_data): Likewise.
14274 * genrecog.c (print_parameter_value): Likewise.
14275 * lra.c (debug_operand_data): Likewise.
14277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14279 * dwarf2out.c (scompare_loc_descriptor_wide)
14280 (scompare_loc_descriptor_narrow): New functions, split out from...
14281 (scompare_loc_descriptor): ...here.
14282 * expmed.c (emit_store_flag_int): New function, split out from...
14283 (emit_store_flag): ...here.
14285 2017-08-30 Richard Biener <rguenther@suse.de>
14287 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14288 (dwarf2out_early_finish): Move setting of AT_pubnames from
14289 early debug output to early finish.
14291 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14293 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14294 and -mdata-region to the assembler.
14296 2017-08-30 Richard Biener <rguenther@suse.de>
14298 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14300 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14301 (dwarf2out_early_global_decl): Only generate a DIE for the
14302 abstract origin if it doesn't already exist or is a declaration DIE.
14303 (resolve_addr): Do not add the linkage name twice when
14304 generating a stub DIE for the DW_TAG_GNU_call_site target.
14306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14308 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14309 Use machine_mode rather than int for arg1_mode.
14311 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14314 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14315 that the second argument of the built-in functions to unpack
14316 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14317 switch statement instead a lot of if statements.
14318 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14319 Allow 64-bit values to be in Altivec registers as well as
14320 traditional floating point registers.
14321 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14323 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14325 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14326 MAX_REGS_PER_ADDRESS == 1.
14328 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14330 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14331 * config/i386/i386.c (ix86_option_override_internal): Simplify
14332 setting of opts->x_flag_entry.
14334 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14335 Jakub Jelinek <jakub@redhat.com>
14336 Richard Biener <rguenther@suse.de>
14338 PR tree-optimization/81503
14339 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14340 folded constant fits in the target type; reorder tests for clarity.
14342 2017-08-29 Martin Liska <mliska@suse.cz>
14344 * passes.def: Include pass_lower_switch.
14345 * stmt.c (dump_case_nodes): Remove and move to
14346 tree-switch-conversion.
14347 (case_values_threshold): Likewise.
14348 (expand_switch_as_decision_tree_p): Likewise.
14349 (emit_case_decision_tree): Likewise.
14350 (expand_case): Likewise.
14351 (balance_case_nodes): Likewise.
14352 (node_has_low_bound): Likewise.
14353 (node_has_high_bound): Likewise.
14354 (node_is_bounded): Likewise.
14355 (emit_case_nodes): Likewise.
14356 (struct simple_case_node): New struct.
14357 (add_case_node): Remove.
14358 (emit_case_dispatch_table): Use vector instead of case_list.
14359 (reset_out_edges_aux): Remove.
14360 (compute_cases_per_edge): Likewise.
14361 (expand_case): Build list of simple_case_node.
14362 (expand_sjlj_dispatch_table): Use it.
14363 * tree-switch-conversion.c (struct case_node): Moved from
14364 stmt.c and adjusted.
14365 (emit_case_nodes): Likewise.
14366 (node_has_low_bound): Likewise.
14367 (node_has_high_bound): Likewise.
14368 (node_is_bounded): Likewise.
14369 (case_values_threshold): Likewise.
14370 (reset_out_edges_aux): Likewise.
14371 (compute_cases_per_edge): Likewise.
14372 (add_case_node): Likewise.
14373 (dump_case_nodes): Likewise.
14374 (balance_case_nodes): Likewise.
14375 (expand_switch_as_decision_tree_p): Likewise.
14376 (emit_jump): Likewise.
14377 (emit_case_decision_tree): Likewise.
14378 (try_switch_expansion): Likewise.
14379 (do_jump_if_equal): Likewise.
14380 (emit_cmp_and_jump_insns): Likewise.
14381 (fix_phi_operands_for_edge): New function.
14382 (record_phi_operand_mapping): Likewise.
14383 (class pass_lower_switch): New pass.
14384 (pass_lower_switch::execute): New function.
14385 (make_pass_lower_switch): Likewise.
14386 (conditional_probability):
14387 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14388 * tree-pass.h: Add make_pass_lower_switch.
14390 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14393 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14396 2017-08-29 Richard Biener <rguenther@suse.de>
14398 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14399 we do not add a DW_AT_inline attribute twice.
14400 (gen_subprogram_die): Remove code setting DW_AT_inline on
14401 DECL_ABSTRACT_P nodes.
14403 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14405 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14406 calls to internal functions.
14407 (gimplify_modify_expr): Likewise.
14408 * tree-call-cdce.c (use_internal_fn): Likewise.
14409 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14410 (convert_to_divmod): Set the nothrow flag.
14411 * tree-if-conv.c (predicate_mem_writes): Likewise.
14412 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14413 (vectorizable_call): Likewise.
14414 (vectorizable_store): Likewise.
14415 (vectorizable_load): Likewise.
14416 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14417 (vect_recog_mask_conversion_pattern): Likewise.
14419 2017-08-29 Martin Liska <mliska@suse.cz>
14422 * gcc.c (driver_handle_option): Add new argument.
14423 * opts-common.c (handle_option): Pass
14424 target_option_override_hook.
14425 * opts-global.c (lang_handle_option): Add new option.
14426 (set_default_handlers): Add new argument.
14427 (decode_options): Likewise.
14428 * opts.c (target_handle_option): Likewise.
14429 (common_handle_option): Call target_option_override_hook.
14430 * opts.h (struct cl_option_handler_func): Add hook for
14431 target option override.
14432 (struct cl_option_handlers): Likewise.
14433 (set_default_handlers): Add new argument.
14434 (decode_options): Likewise.
14435 (common_handle_option): Likewise.
14436 (target_handle_option): Likewise.
14437 * toplev.c (toplev::main): Pass targetm.target_option.override
14440 2017-08-29 Richard Biener <rguenther@suse.de>
14441 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14443 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14444 life to the active subtree.
14446 2017-08-28 Jeff Law <law@redhat.com>
14448 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14449 derive_equivalences.
14450 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14451 Code moved into....
14452 (edge_info::derive_equivalences): New private member function
14454 * tree-ssa-dom.c (class edge_info): Changed from a struct
14455 to a class. Add ctor/dtor, methods and data members.
14456 (edge_info::edge_info): Renamed from allocate_edge_info.
14457 Initialize additional members.
14458 (edge_info::~edge_info): New.
14459 (free_dom_edge_info): Delete the edge info.
14460 (record_edge_info): Use new class & associated member functions.
14461 Tighten forms for testing for edge equivalences.
14462 (record_temporary_equivalences): Iterate over the simple
14463 equivalences rather than assuming there's only one per edge.
14464 (cprop_into_successor_phis): Iterate over the simple
14465 equivalences rather than assuming there's only one per edge.
14466 (optimize_stmt): Use operand_equal_p rather than pointer
14467 equality for mini-DSE code.
14469 2017-08-28 Nathan Sidwell <nathan@acm.org>
14471 * gcc.c (execute): Fold SIGPIPE handling into switch
14472 statement. Adjust internal error message.
14474 2017-08-28 Richard Biener <rguenther@suse.de>
14477 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
14478 Do nothing for removed DIEs.
14480 2017-08-28 Richard Biener <rguenther@suse.de>
14482 PR tree-optimization/81977
14483 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
14486 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
14489 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
14490 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
14491 using targetm.gen_mem_thread_fence.
14493 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14496 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14497 predicate to register_operand. Reorder operands.
14498 (*btr<mode>): Ditto.
14499 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14500 (*btr<mode>_mask): Ditto.
14502 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14504 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14505 * config/rs6000/xmmintrin.h: New file.
14506 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14508 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14511 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14512 parameter and_insn and return it.
14513 (recombine_lvx_pattern): Insert a copy to ensure availability of
14514 the base register of the copied masking operation at the point of
14515 the instruction replacement.
14516 (recombine_stvx_pattern): Likewise.
14518 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14520 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14521 undocumented switches.
14522 (-mpower9-dform-vector): Likewise.
14523 (-mpower9-dform): Likewise.
14524 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14525 comments to delete references to -mpower9-dform* switches.
14526 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14527 Delete reference to -mpower9-dform* switches, test for
14528 -mpower9-vector instead.
14529 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14530 (OTHER_P9_VECTOR_MASKS): Likewise.
14531 (POWERPC_MASKS): Likewise.
14532 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14533 tests against -mpower9-dform* to -mpower9-vector. Delete code
14534 that checked for -mpower9-dform* consistancy with other options.
14535 Add test for -mpower9-misc to enable other power9 switches.
14536 (rs6000_init_hard_regno_mode_ok): Likewise.
14537 (rs6000_option_override_internal): Likewise.
14538 (rs6000_emit_prologue): Likewise.
14539 (rs6000_emit_epilogue): Likewise.
14540 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14541 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14542 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14544 (emit_fusion_p9_store): Likewise.
14545 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14546 resetting these macros if the assembler does not support ISA 3.0
14548 (TARGET_P9_DFORM_VECTOR): Likewise.
14549 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14550 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14552 2017-08-25 Alan Modra <amodra@gmail.com>
14554 PR rtl-optimization/81747
14555 * cse.c (cse_extended_basic_block): Don't attempt to record
14556 equivalences for degenerate conditional branches that branch
14557 to their fall-through.
14559 2017-08-24 Martin Sebor <msebor@redhat.com>
14561 PR middle-end/81908
14562 * gimple-fold.c (size_must_be_zero_p): New function.
14563 (gimple_fold_builtin_memory_op): Call it.
14565 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14567 * config/rs6000/mm_malloc.h: New file.
14569 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14571 PR tree-optimization/81913
14572 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14573 analysis when either IVs in condition can wrap.
14575 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14577 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14578 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14580 2017-08-24 Richard Biener <rguenther@suse.de>
14583 * targhooks.c (default_target_can_inline_p): Properly
14584 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14585 is present and always compare.
14586 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14587 infer -mfpmath=sse from TARGET_SSE_P.
14588 (ix86_can_inline_p): Properly use target_option_default_node when
14589 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14591 2017-08-24 Richard Biener <rguenther@suse.de>
14594 * dwarf2out.c (output_die): Handle flag_generate_offload like
14596 (output_comp_unit): Likewise.
14597 (gen_array_type_die): Likewise.
14598 (dwarf2out_early_finish): Likewise.
14599 (note_variable_value_in_expr): Likewise.
14600 (dwarf2out_finish): Likewise. Adjust assert.
14601 * cgraphunit.c (symbol_table::compile): Move setting of
14602 flag_generate_offload earlier ...
14603 (symbol_table::finalize_compilation_unit): ... here, before
14604 early debug finalization.
14606 2017-08-24 Richard Biener <rguenther@suse.de>
14608 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14609 and ipa-fnsummary.h.
14610 (ix86_can_inline_p): When ix86_fpmath flags do not match
14611 check whether the callee uses FP math at all.
14613 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14615 PR middle-end/81931
14616 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14617 instead of TYPE_PRECISION.
14619 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14620 Alan Hayward <alan.hayward@arm.com>
14621 David Sherwood <david.sherwood@arm.com>
14623 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14625 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14627 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14629 (ix86_frame::stack_realign_allocate): New field.
14630 (struct machine_frame_state): Modify comments.
14631 (machine_frame_state::sp_realigned_fp_end): New field.
14632 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14633 layout calculation.
14634 (sp_valid_at): Add assertion to assure no attempt to access invalid
14635 offset of a realigned stack.
14636 (fp_valid_at): Likewise.
14637 (choose_baseaddr): Modify comments.
14638 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14639 ix86_expand_prologue.
14640 (ix86_expand_prologue): Modify stack realignment and allocation.
14641 (ix86_expand_epilogue): Modify comments.
14642 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14643 avx2_runtime, avx512f, and avx512f_runtime.
14645 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14647 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14648 (mstackrealign): Do not init to -1.
14649 * config/i386/i386.c (ix86_option_override_internal):
14650 Check opts_set, not opts when setting default value of
14651 opts->x_ix86_force_align_arg_pointer.
14653 2017-08-23 Richard Biener <rguenther@suse.de>
14655 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14656 lang_hooks.decl_printable_name.
14657 * print-rtl-function.c (print_rtx_function): Likewise.
14658 * tree-pretty-print.c (dump_function_header): Likewise.
14660 2017-08-23 Richard Biener <rguenther@suse.de>
14663 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14664 -g0 at compile-time.
14666 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14668 PR middle-end/19706
14669 * doc/sourcebuild.texi (Other hardware attributes):
14672 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14674 PR middle-end/19706
14675 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14676 Fix single-use check.
14678 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14680 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14682 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14684 * doc/install.texi: Modify to add more details on running selected
14687 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14689 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14690 is combined with -mabi=ms.
14691 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14694 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14696 PR tree-optimization/81488
14697 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14698 and cached_basis fields.
14699 (MAX_SPREAD): New constant.
14700 (alloc_cand_and_find_basis): Initialize new fields.
14701 (clear_visited): New function.
14702 (create_phi_basis_1): Rename from create_phi_basis, set visited
14703 and cached_basis fields.
14704 (create_phi_basis): New wrapper function.
14705 (phi_add_costs_1): Rename from phi_add_costs, add spread
14706 parameter, set visited field, short-circuit when limits reached.
14707 (phi_add_costs): New wrapper function.
14708 (record_phi_increments_1): Rename from record_phi_increments, set
14710 (record_phi_increments): New wrapper function.
14711 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14712 (phi_incr_cost): New wrapper function.
14713 (all_phi_incrs_profitable_1): Rename from
14714 all_phi_incrs_profitable, set visited field.
14715 (all_phi_incrs_profitable): New wrapper function.
14717 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14718 Alan Hayward <alan.hayward@arm.com>
14719 David Sherwood <david.sherwood@arm.com>
14721 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14722 that takes the outer and inner modes.
14723 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14724 comparison as the canonical test for a paradoxical subreg.
14725 * combine.c (simplify_set): Use paradoxical_subreg_p.
14726 (make_extraction): Likewise.
14727 (force_to_mode): Likewise.
14728 (rtx_equal_for_field_assignment_p): Likewise.
14729 (gen_lowpart_for_combine): Likewise.
14730 (simplify_comparison): Likewise.
14731 * cse.c (equiv_constant): Likewise.
14732 * expmed.c (store_bit_field_1): Likewise.
14733 * final.c (alter_subreg): Likewise.
14734 * fwprop.c (propagate_rtx): Likewise.
14735 (forward_propagate_subreg): Likewise.
14736 * ira-conflicts.c (ira_build_conflicts): Likewise.
14737 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14738 * lra-constraints.c (curr_insn_transform): Likewise.
14739 (split_reg): Likewise.
14740 * lra-eliminations.c (move_plus_up): Likewise.
14741 (lra_eliminate_regs_1): Likewise.
14742 * recog.c (general_operand): Likewise.
14743 * ree.c (combine_reaching_defs): Likewise.
14744 * reload.c (push_reload): Likewise.
14745 (find_reloads): Likewise.
14746 * reload1.c (elimination_effects): Likewise.
14747 (compute_reload_subreg_offset): Likewise.
14748 (choose_reload_regs): Likewise.
14749 * rtlanal.c (subreg_lsb_1): Likewise.
14750 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14751 (simplify_subreg): Likewise.
14752 * var-tracking.c (track_loc_p): Likewise.
14753 * emit-rtl.c (byte_lowpart_offset): Likewise.
14754 (paradoxical_subreg_p): Delete out-of-line definition.
14756 2017-08-22 Jeff Law <law@redhat.com>
14758 PR tree-optimization/81741
14759 PR tree-optimization/71947
14760 * tree-ssa-dom.c: Include tree-inline.h.
14761 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14762 equivalences if one is more expensive to compute than the other.
14763 * tree-ssa-scopedtables.h (class const_or_copies): Make
14764 record_const_or_copy_raw method private.
14765 (class avail_exprs_stack): New method simplify_binary_operation.
14766 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14767 avail_exprs_stack::simplify_binary_operation as needed.
14768 (avail_exprs_stack::simplify_binary_operation): New function.
14770 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14772 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14773 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14774 (DOT_SYMBOLS): Likewise.
14775 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14776 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14777 (RS6000_ABI_NAME): Likewise.
14778 (TARGET_CMODEL): Likewise.
14779 (TOC_SECTION_ASM_OP): Likewise.
14780 (SET_CMODEL): New macro.
14781 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14783 2017-08-22 Richard Biener <rguenther@suse.de>
14785 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14786 to allow for free-lang-data replacements similar to verify_type_variant.
14788 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14789 Alan Hayward <alan.hayward@arm.com>
14790 David Sherwood <david.sherwood@arm.com>
14792 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14793 VOIDmode for the LABEL_REF.
14795 2017-08-22 Richard Biener <rguenther@suse.de>
14797 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14799 2017-08-22 Richard Biener <rguenther@suse.de>
14801 * common.opt (feliminate-dwarf2-dups): Ignore.
14802 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14803 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14804 same_die_p_wrap, compute_section_prefix,
14805 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14806 (comdat_symbol_id, comdat_symbol_number): Likewise.
14807 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14809 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14810 (output_die): Mark unreachable path unreachable.
14811 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14812 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14813 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14814 (dwarf2out_early_finish): Likewise.
14816 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14818 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14820 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14823 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14824 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14825 (avr_attribute_table) <io, io_low, address>: Initialize
14826 .decl_required with true.
14828 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14830 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14831 undocumented debugging options.
14832 (-mvsx-scalar-double): Likewise.
14833 (-mallow-df-permute): Likewise.
14834 (-mvectorize-builtins): Likewise.
14835 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14836 (rs6000_builtin_vectorized_function): Likewise.
14837 (rs6000_builtin_md_vectorized_function): Likewise.
14838 (rs6000_opt_vars): Likewise.
14840 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14843 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14844 (*btrq_imm): Rename from *btrq.
14845 (*btcq_imm): Rename from *btcq.
14846 (btsc): New code attribute.
14847 (*<btsc><mode>): New insn pattern.
14848 (*btr<mode>): Ditto.
14849 (*<btsc><mode>_mask): New insn_and_split pattern.
14850 (*btr<mode>_mask): Ditto.
14852 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14853 Alan Hayward <alan.hayward@arm.com>
14854 David Sherwood <david.sherwood@arm.com>
14856 * function.c (pad_below): Simplify padding calculation.
14858 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14859 Alan Hayward <alan.hayward@arm.com>
14860 David Sherwood <david.sherwood@arm.com>
14862 * target.def (function_prologue): Remove frame size argument.
14863 (function_epilogue): Likewise.
14864 * doc/tm.texi: Regenerate.
14865 * final.c (final_start_function): Update call to function_prologue.
14866 (final_end_function): Update call to function_epilogue.
14867 (default_function_pro_epilogue): Remove frame size argument.
14868 * output.h (default_function_pro_epilogue): Likewise.
14869 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14870 (arm_output_function_prologue): Likewise.
14871 * config/frv/frv.c (frv_function_prologue): Likewise.
14872 (frv_function_epilogue): Likewise.
14873 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14874 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14875 (ia64_output_function_epilogue): Likewise.
14876 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14877 (m32r_output_function_epilogue): Likewise.
14878 * config/microblaze/microblaze.c (microblaze_function_prologue)
14879 (microblaze_function_epilogue): Likewise.
14880 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14881 (mips_output_function_epilogue): Likewise.
14882 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14883 (mmix_target_asm_function_epilogue): Likewise.
14884 * config/msp430/msp430.c (msp430_start_function): Likewise.
14885 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14886 (nds32_asm_function_epilogue): Likewise.
14887 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14888 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14889 (pa_output_function_epilogue): Likewise.
14890 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14891 (rs6000_output_function_epilogue): Likewise.
14892 * config/rl78/rl78.c (rl78_start_function): Likewise.
14893 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14894 (rs6000_output_function_epilogue): Likewise.
14895 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14896 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14897 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14898 (sparc_asm_function_epilogue): Likewise.
14900 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14902 * tree.h (type_has_mode_precision_p): New function.
14903 * convert.c (convert_to_integer_1): Use it.
14904 * expr.c (expand_expr_real_2): Likewise.
14905 (expand_expr_real_1): Likewise.
14906 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14907 * match.pd: Likewise.
14908 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14909 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14910 * tree-tailcall.c (process_assignment): Likewise.
14911 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14912 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14913 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14914 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14915 (vectorizable_assignment): Likewise.
14916 (vectorizable_shift): Likewise.
14917 (vectorizable_operation): Likewise.
14918 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14920 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14922 * match.pd: Add pow (C, x) simplification.
14924 2017-08-21 Richard Biener <rguenther@suse.de>
14926 PR tree-optimization/81900
14927 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14928 for blocks with abnormal predecessors.
14929 (compute_antic): Do not set visited flag prematurely.
14931 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14934 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14938 * stor-layout.h (vector_type_mode): Move to...
14939 * tree.h (vector_type_mode): ...here.
14940 * stor-layout.c (vector_type_mode): Move to...
14941 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14943 2017-08-21 Richard Biener <rguenther@suse.de>
14945 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14946 register_external_die hooks.
14947 (debug_false_tree_charstarstar_uhwistar): Declare.
14948 (debug_nothing_tree_charstar_uhwi): Likewise.
14949 * debug.c (do_nothing_debug_hooks): Adjust.
14950 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14951 (debug_nothing_tree_charstar_uhwi): Likewise.
14952 * dbxout.c (dbx_debug_hooks): Adjust.
14953 (xcoff_debug_hooks): Likewise.
14954 * sdbout.c (sdb_debug_hooks): Likewise.
14955 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14956 * dwarf2out.c (macinfo_label_base): New global.
14957 (dwarf2out_register_external_die): New function for the
14958 register_external_die hook.
14959 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14960 (dwarf2_debug_hooks): Use them.
14961 (dwarf2_lineno_debug_hooks): Adjust.
14962 (struct die_struct): Add with_offset flag.
14963 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14964 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14965 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14966 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14967 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14968 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14969 defining section names for the early LTO debug variants.
14970 (reset_indirect_string): New helper.
14971 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14972 (print_dw_val): Add support for offsetted symbol references.
14973 (get_ultimate_context): Split out from is_cxx.
14974 (is_cxx): Use get_ultimate_context.
14975 (is_fortran): Add decl overload.
14976 (compute_comp_unit_symbol): Split out worker from
14977 compute_section_prefix.
14978 (compute_section_prefix): Call compute_comp_unit_symbol and
14979 set comdat_type_p here.
14980 (output_die): Skip DIE symbol output for the LTO added one.
14981 Handle DIE symbol references with offset.
14982 (output_comp_unit): Guard section name mangling properly.
14983 For LTO debug sections emit a symbol at the section beginning
14984 which we use to refer to its DIEs.
14985 (add_abstract_origin_attribute): For DIEs registered via
14986 dwarf2out_register_external_die directly refer to the early
14987 DIE rather than indirectly through the shadow one we created.
14988 Remove obsolete call to dwarf2out_abstract_function for
14989 non-function/block origins.
14990 (gen_array_type_die): When generating early LTO debug do
14991 not emit DW_AT_string_length.
14992 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14993 late when in LTO. As suggested place a gcc_unreachable for
14994 the DECL_ABSTRACT_P case.
14995 (gen_subprogram_die): Avoid another specification DIE
14996 for early built declarations/definitions for the late LTO case.
14997 (gen_variable_die): Add type references for late duplicated VLA dies
14999 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15000 we have the abstract instance already.
15001 (process_scope_var): Adjust decl DIE contexts in LTO which
15002 first puts them in limbo.
15003 (gen_decl_die): Do not generate type DIEs late apart from
15004 types for VLAs or for decls we do not yet have a DIE. Do not
15005 call dwarf2out_abstract_function late.
15006 (dwarf2out_early_global_decl): Make sure to create DIEs
15007 for abstract instances of a decl first.
15008 (dwarf2out_late_global_decl): Adjust comment.
15009 (output_macinfo_op): With multiple macro sections use
15010 macinfo_label_base to distinguish labels.
15011 (output_macinfo): Likewise. Update macinfo_label_base.
15012 Pass in the line info label.
15013 (note_variable_value_in_expr): When generating LTO resolve
15014 all variable values here by generating DIEs as needed.
15015 (init_sections_and_labels): Add early LTO debug flag parameter
15016 and generate different sections and names if set. Add generation
15017 counter for the labels so we can have multiple of them.
15018 (reset_dies): Helper to allow DIEs to be output multiple times.
15019 (dwarf2out_finish): When outputting DIEs to the fat part of an
15020 LTO object first reset DIEs.
15021 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15022 (modified_type_die): Check for decl_ultimate_origin being self
15024 (gen_type_die_with_usage): Likewise.
15025 (gen_typedef_die): Allow decl_ultimate_origin being self.
15026 (set_decl_abstract_flags): Remove.
15027 (set_block_abstract_flags): Likewise.
15028 (dwarf2out_abstract_function): Treat the early generated DIEs
15029 as the abstract copy and only add DW_AT_inline and
15030 DW_AT_artificial here and call set_decl_origin_self.
15031 If the DIE has an abstract origin don't do anything.
15032 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15033 if we have none yet (Go fails to build one, PR78628).
15034 (variably_modified_type_p): Prevent endless recursion for Ada
15035 cyclic pointer types.
15036 * lto-streamer-in.c: Include debug.h.
15037 (dref_queue): New global.
15038 (lto_read_tree_1): Stream in DIE references.
15039 (lto_input_tree): Register DIE references.
15040 (input_function): Stream DECL_DEBUG_ARGS.
15041 * lto-streamer-out.c: Include debug.h.
15042 (lto_write_tree_1): Output DIE references.
15043 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15044 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15045 (output_function): Stream DECL_DEBUG_ARGS.
15046 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15047 Stream DECL_ABSTRACT_ORIGIN.
15048 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15049 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15050 DECL_CONTEXT for file-scope decls.
15051 * lto-streamer.h (struct dref_entry): Declare.
15052 (dref_queue): Likewise.
15053 * cfgexpand.c (pass_expand::execute): Do not call the
15054 outlining_inline_function hook here.
15055 * lto-wrapper.c (debug_obj): New global.
15056 (tool_cleanup): Unlink it if required.
15057 (debug_objcopy): New function.
15058 (run_gcc): Handle early debug sections in the IL files by
15059 extracting them to separate files, partially linkin them and
15060 feeding the result back as result to the linker.
15061 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15062 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15063 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15064 sections into a separate segment.
15065 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15067 (darwin_asm_dwarf_section): Likewise.
15068 (darwin_asm_output_dwarf_offset): Likewise.
15069 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15071 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15072 Alan Hayward <alan.hayward@arm.com>
15073 David Sherwood <david.sherwood@arm.com>
15075 * read-md.h (md_reader::record_potential_iterator_use): Replace
15076 pointer argument with an rtx and an index.
15077 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15078 (apply_mode_iterator): Likewise.
15079 (apply_code_iterator): Likewise.
15080 (apply_int_iterator): Likewise.
15081 (apply_subst_iterator): Likewise.
15082 (record_iterator_use): Likewise.
15083 (record_attribute_use): Likewise.
15084 (md_reader::record_potential_iterator_use): Likewise. Update calls
15085 to record_iterator_use and apply_iterator.
15086 (iterator_use): Replace ptr with x and index.
15087 (attribute_use): Likewise.
15088 (apply_attribute_uses): Update calls to apply_iterator.
15089 (apply_iterators): Likewise. Update initialization of iterator_use.
15090 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15091 and record_potential_iterator_use.
15092 (rtx_reader::read_rtx_operand): Likewise.
15094 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15095 Alan Hayward <alan.hayward@arm.com>
15096 David Sherwood <david.sherwood@arm.com>
15098 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15101 2017-08-21 Richard Biener <rguenther@suse.de>
15103 PR middle-end/81884
15104 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15105 at struct end conservatively when comparing common bases.
15107 2017-08-21 Richard Biener <rguenther@suse.de>
15109 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15110 (mem_ref_in_stmt): Remove.
15111 (determine_max_movement): Use ref index to get at the reference.
15112 (invariantness_dom_walker::before_dom_children): Deal with
15113 lim data already initialized.
15114 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15116 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15118 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15119 (TARGET_ISA_ROUND): Ditto.
15120 (TARGET_ROUND): Ditto.
15121 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15122 * config/i386/i386.md: Ditto.
15123 * config/i386/sse.md: Ditto.
15124 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15125 with OPTION_MASK_ISA_SSE4_1.
15127 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15130 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15131 __builtin_ia32_lzcnt_u16.
15133 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15136 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15137 (rs6000_set_current_function): Rewrite function to use it.
15139 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15142 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15143 and TYPE_WARN_IF_NOT_ALIGN.
15144 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15145 (handle_warn_if_not_align): New.
15146 (place_union_field): Call handle_warn_if_not_align.
15147 (place_field): Call handle_warn_if_not_align.
15148 Copy TYPE_WARN_IF_NOT_ALIGN.
15149 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15150 (layout_type): Likewise.
15151 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15153 (tree_decl_common): Add warn_if_not_align.
15154 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15155 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15156 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15157 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15158 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15159 * doc/extend.texi: Document warn_if_not_aligned attribute.
15160 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15162 2017-08-17 Martin Liska <mliska@suse.cz>
15165 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15166 (get_data_dependence): Use it as pointer type.
15167 (distribute_loop): Likewise.
15169 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15171 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15172 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15173 (p8_vmrgow_<mode>_direct): New define_insn.
15174 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15175 handle endianness for vmrgew and vmrgow permute patterns.
15177 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15179 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15180 * config/rs6000/rs6000-cpus.def: Remove comment.
15181 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15182 (POWERPC_MASKS): Likewise.
15183 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15184 use of TARGET_VSX_TIMODE.
15185 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15186 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15187 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15188 (rs6000_option_override_internal): Remove dead code.
15189 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15190 (rs6000_legitimize_reload_address): Likewise.
15191 (rs6000_legitimate_address_p): Likewise.
15192 (rs6000_opt_masks): Delete "vsx-timode".
15193 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15194 from function comment.
15195 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15196 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15197 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15199 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15200 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15201 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15202 (VSX_TI): Likewise.
15204 (define_peephole2): Likewise.
15206 2017-08-17 Martin Sebor <msebor@redhat.com>
15209 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15210 past the end of an array.
15211 (test_pp_format): Add test cases.
15213 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15215 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15216 missing ECF_NOTHROW flags.
15218 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15221 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15222 operands residing in integer registers.
15223 (*vsx_le_perm_load_<mode>): Likewise.
15224 (*vsx_le_perm_store_<mode>): Likewise.
15225 (define_peephole2): Add peepholes to optimize the above.
15227 2017-08-17 Marek Polacek <polacek@redhat.com>
15229 PR middle-end/81814
15230 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15231 to mimic what shorten_compare did. Change the return type to bool.
15232 (fold_cond_expr_with_comparison): Update call to
15233 operand_equal_for_comparison_p.
15234 (fold_ternary_loc): Likewise.
15236 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15238 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15240 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15242 2017-08-17 Richard Biener <rguenther@suse.de>
15244 * tree-ssa-structalias.c (solve_graph): When propagating
15245 to successors update the graphs succ edges and avoid duplicate work.
15247 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15250 * config/i386/i386.c (ix86_option_override_internal): Save target
15251 specific options after ix86_stack_protector_guard_reg was changed.
15253 2017-08-17 Richard Biener <rguenther@suse.de>
15255 PR tree-optimization/81827
15256 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15257 (new_var_info): Initialize it conservatively.
15258 (get_call_vi): Mark register vars.
15259 (new_scalar_tmp_constraint_exp): Likewise.
15260 (handle_rhs_call): Likewise.
15261 (handle_const_call): Likewise.
15262 (create_function_info_for): Likewise.
15263 (solve_constraints): Sort varinfos to separate register from
15264 non-register vars to pack points-to solution bitmaps during
15267 2017-08-17 Marek Polacek <polacek@redhat.com>
15269 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15271 2017-08-17 Richard Biener <rguenther@suse.de>
15273 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15274 to true when overflow is undefined and we saturated the result.
15276 2017-08-17 Alan Modra <amodra@gmail.com>
15279 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15280 Don't use store multiple if only one reg needs saving.
15281 (interesting_frame_related_regno): New function.
15282 (rs6000_frame_related): Don't emit frame info for regs that
15284 (rs6000_emit_epilogue): Likewise.
15286 2017-08-16 Nathan Sidwell <nathan@acm.org>
15288 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15289 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15290 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15291 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15292 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15293 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15294 (verify_type): Adjust for TYPE_BINFO move.
15295 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15296 process TYPE_BINFO directly.
15297 (hash_tree): Likewise.
15298 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15300 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15303 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15305 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15307 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15310 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15311 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15312 Add ix86_binary_operator_ok to insn constraint.
15313 (*iordi_1_bts): Ditto.
15314 (*xordi_1_btc): Ditto.
15315 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15316 Update corresponding peephole2 pattern.
15320 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15322 PR tree-optimization/81832
15323 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15324 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15326 2017-08-16 Marek Polacek <polacek@redhat.com>
15329 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15330 regarding size_zero_node.
15332 2017-08-16 Martin Liska <mliska@suse.cz>
15335 * config.gcc: Respect previously set extra_objs in case
15338 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15340 PR tree-optimization/81835
15341 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15342 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15343 not depend on the phi.
15345 2017-08-16 Alan Modra <amodra@gmail.com>
15347 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15350 2017-08-16 Alan Modra <amodra@gmail.com>
15352 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15353 (save_reg_p): ..into this. Update all callers.
15354 (first_reg_to_save): Simplify.
15356 2017-08-16 Alan Modra <amodra@gmail.com>
15358 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15361 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15365 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15366 constant count if that count is less than 32.
15368 2017-08-15 Nathan Sidwell <nathan@acm.org>
15370 * gcc.c (execute): Emit friendlier message if inferior is killed
15371 by an external cause.
15373 2017-08-15 Richard Biener <rguenther@suse.de>
15375 PR tree-optimization/81790
15376 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15377 CONSTRUCTORs from simplifying and VN.
15379 2017-08-14 Martin Sebor <msebor@redhat.com>
15381 * builtin-attrs.def: Add comments.
15383 2017-08-14 Martin Sebor <msebor@redhat.com>
15386 * doc/extend.texi (attribute nonstring): Document new attribute.
15388 2017-08-14 Martin Sebor <msebor@redhat.com>
15391 * tree-diagnostic.c (default_tree_printer): Handle %G.
15392 * gimple-pretty-print.h (percent_G_format): Declare new function.
15393 * gimple-pretty-print.c (percent_G_format): Define.
15394 * tree-pretty-print.c (percent_K_format): Add argument.
15396 2017-08-14 Martin Sebor <msebor@redhat.com>
15398 PR translation/79998
15399 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15400 Remove a stray space.
15402 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15405 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15406 (*iordi_1_bts): Ditto.
15407 (*xordi_1_btc): Ditto.
15409 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15412 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15413 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15415 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15416 quoted strings, and make more translator-friendly.
15417 (darwin_rs6000_override_options): Likewise.
15418 (rs6000_option_override_internal): Likewise.
15419 (rs6000_return_in_memory): Fix overlong line.
15420 (init_cmulative_args): Use quoted strings, and make more
15421 translator-friendly.
15422 (rs6000_pass_by_reference): Fix overlong line.
15423 (def_builtin): Use quoted strings.
15424 (altivec_expand_predicate_builtin): Use quoted strings, and make
15425 more translator-friendly.
15426 (htm_expand_builtin): Use quoted strings.
15427 (cpu_expand_builtin): Use quoted strings, and make more
15428 translator-friendly.
15429 (altivec_expand_builtin): Likewise.
15430 (paired_expand_predicate_builtin): Likewise.
15431 (rs6000_invalid_builtin): Likewise.
15432 (builtin_function_type): Use quoted strings.
15433 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15434 more translator-friendly.
15435 (rs6000_trampoline_init): Likewise.
15436 (rs6000_handle_altivec_attribute): Likewise.
15437 (rs6000_inner_target_options): Use quoted strings.
15438 (rs6000_disable_incompatible_switches): Likewise.
15439 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15440 strings, and make more translator-friendly.
15441 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15443 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15445 PR tree-optimization/81799
15446 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15447 cond_expr to simple gimple operand.
15449 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15451 PR middle-end/46932
15452 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15454 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
15458 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
15459 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
15460 TARGET_GASISR_PROLOGUES.
15461 * config/avr/avr.c (avr_option_override): Same.
15462 (avr_pass_pre_proep::execute): Same.
15464 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
15467 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
15468 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
15470 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
15472 * config/i386/i386.md (*load_tp_<mode>): Redefine as
15473 define_insn_and_split. Split to a memory load from 0 in
15474 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
15475 using PTR mode iterator.
15476 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
15477 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
15478 (*add_tp_<mode>): Redefine as define_insn_and_split.
15479 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
15480 address space. Merge with *add_tp_x32 using PTR mode iterator.
15481 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
15482 Split to an add with a memory load from 0 in
15483 DEFAULT_TLS_SEG_REG address space.
15485 2017-08-12 Andrew Pinski <apinski@cavium.com>
15487 * config/aarch64/aarch64-option-extensions.def (rdma):
15488 Fix feature string to what Linux prints out in /proc/cpuinfo.
15490 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15493 * dwarf2out.c (modified_type_die): For C typedef types that have
15494 an ultimate origin, process the ultimate origin instead of the
15496 (gen_typedef_die): Assert that input DECLs have no ultimate
15498 (gen_type_die_with_usage): For typedef variants that have an
15499 ultimate origin, just call gen_decl_die on the original DECL.
15500 (process_scope_var): Avoid creating DIEs for local typedefs and
15501 concrete static variables.
15503 2017-08-12 Alan Modra <amodra@gmail.com>
15507 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15508 match gnu-user.h startfile.
15509 (ENDFILE_LINUX_SPEC): Similarly.
15511 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15514 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15516 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15518 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15519 * config/aarch64/aarch64.md (mov<mode>): Change.
15520 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15521 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15522 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15524 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15526 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15527 for storage order barriers.
15529 2017-08-11 Martin Liska <mliska@suse.cz>
15532 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15533 variables of void type.
15535 2017-08-11 Martin Liska <mliska@suse.cz>
15537 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15538 TARGET_SUPPORTS_ALIASES.
15539 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15540 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15541 (optimize_weakref): Likewise.
15542 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15543 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15544 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15546 2017-08-11 Martin Liska <mliska@suse.cz>
15549 * config/i386/i386.c (make_resolver_func): Do complete
15550 refactoring of the function.
15552 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15555 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15556 * config/i386/i386.c (ix86_stack_protect_guard): Use
15557 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15558 * doc/invoke.texi (x86 Options): Document
15559 -mstack-protector-guard-symbol= option.
15561 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15563 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15564 * config/i386/i386.c (ix86_split_stack_guard): New function.
15565 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15566 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15567 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15568 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15569 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15570 (split_stack_space_check): Call ix86_split_stack_guard.
15572 2017-08-10 Martin Sebor <msebor@redhat.com>
15574 * print-tree.c (print_node): Print location using the established
15576 Replace spaces with colons.
15577 (debug_raw, debug): Ditto.
15579 2017-08-10 Martin Sebor <msebor@redhat.com>
15582 * pretty-print.c (pp_format): Correct the handling of %s precision.
15584 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15587 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15589 (ix86_finalize_stack_frame_flags): This. Also clear
15590 frame_pointer_needed if -fno-omit-frame-pointer is used without
15592 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15593 with ix86_finalize_stack_frame_flags.
15594 (ix86_expand_epilogue): Likewise.
15595 (ix86_expand_split_stack_prologue): Likewise.
15596 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15598 2017-08-10 Martin Liska <mliska@suse.cz>
15601 * c-attribs.c (handle_target_attribute):
15602 Report warning for an empty string argument of target attribute.
15604 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15607 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15609 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15610 or DECL_NONLOCAL labels.
15611 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15612 or DECL_NONLOCAL labels here.
15614 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15617 to indicate when early gimple folding has been disabled.
15618 (rs6000_gimple_fold_builtin): Add debug content.
15619 (rs6000_invalid_builtin): Fix whitespace.
15620 (rs6000_expand_builtin): Fix whitespace.
15621 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15623 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15626 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15627 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15629 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15631 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15632 (qdf24xx): Likewise.
15633 * config/aarch64/aarch64.md: Include falkor.md.
15634 * config/aarch64/falkor.md: New.
15636 2017-08-09 Marek Polacek <polacek@redhat.com>
15639 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15640 * diagnostic.c (emit_diagnostic): Add a comment.
15641 (emit_diagnostic_valist): New function.
15643 2017-08-09 Marek Polacek <polacek@redhat.com>
15646 * input.c (make_location): New overload.
15647 * input.h (make_location): Declare.
15649 2017-08-08 Alan Modra <amodra@gmail.com>
15650 H.J. Lu <hongjiu.lu@intel.com>
15653 * gcc.c (NO_PIE_SPEC): Delete.
15654 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15656 (LINK_PIE_SPEC): ..to here.
15657 (LINK_COMMAND_SPEC): Support -no-pie.
15658 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15659 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15661 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15662 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15663 (ENDFILE_CRTEND_SPEC): Similarly.
15665 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15668 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15669 (mstack-protector-guard-offset=): Ditto.
15670 * config/i386/i386.c (ix86_option_override): Handle
15671 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15673 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15674 ix86_stack_protect_guard_offset variables.
15675 (TARGET_STACK_PROTECT_GUARD): Always define.
15676 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15677 and -mstack-protector-guard-offset= options.
15679 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15681 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15682 boundary case for the last candidate.
15684 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15686 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15687 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15689 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15691 PR middle-end/19706
15692 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15693 * config/aarch64/aarch64-builtins.c
15694 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15695 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15696 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15698 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15699 Andrew Pinski <pinskia@gmail.com>
15701 PR middle-end/19706
15702 * internal-fn.def (XORSIGN): New.
15703 * optabs.def (xorsign_optab): New.
15704 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15705 (convert_expand_mult_copysign): New.
15706 (pass_optimize_widening_mul::execute): Call
15707 convert_expand_mult_copysign.
15709 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15711 PR tree-optimization/81354
15712 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15713 Insert on edges rather than explicitly creating landing pads.
15714 (analyze_candidates_and_replace): Commit edge inserts.
15716 2017-08-08 Richard Biener <rguenther@suse.de>
15718 PR middle-end/81719
15719 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15720 (expand_simple_operations): Also look through ADDR_EXPRs with
15721 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15723 2017-08-08 Richard Biener <rguenther@suse.de>
15725 PR tree-optimization/81723
15726 * tree-vect-slp.c (struct bst_traits): New hash traits.
15727 (bst_fail): New global.
15728 (vect_build_slp_tree_2): New worker, split out from ...
15729 (vect_build_slp_tree): ... this now wrapping it with using
15730 bst_fail set to cache SLP tree build fails. Properly handle
15732 (vect_analyze_slp_instance): Allocate and free bst_fail.
15734 2017-08-08 Martin Liska <mliska@suse.cz>
15737 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15738 LABEL_DECLs that can be from a different function.
15740 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15742 PR tree-optimization/81744
15743 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15744 loop's number of iterations.
15746 2017-08-08 Martin Liska <mliska@suse.cz>
15748 * asan.c: Include header files.
15749 * attribs.c (build_decl_attribute_variant): New function moved
15751 (build_type_attribute_qual_variant): Likewise.
15752 (cmp_attrib_identifiers): Likewise.
15753 (simple_cst_list_equal): Likewise.
15754 (omp_declare_simd_clauses_equal): Likewise.
15755 (attribute_value_equal): Likewise.
15756 (comp_type_attributes): Likewise.
15757 (build_type_attribute_variant): Likewise.
15758 (lookup_ident_attribute): Likewise.
15759 (remove_attribute): Likewise.
15760 (merge_attributes): Likewise.
15761 (merge_type_attributes): Likewise.
15762 (merge_decl_attributes): Likewise.
15763 (merge_dllimport_decl_attributes): Likewise.
15764 (handle_dll_attribute): Likewise.
15765 (attribute_list_equal): Likewise.
15766 (attribute_list_contained): Likewise.
15767 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15768 (lookup_attribute_by_prefix): Likewise.
15769 * bb-reorder.c: Include header files.
15770 * builtins.c: Likewise.
15771 * calls.c: Likewise.
15772 * cfgexpand.c: Likewise.
15773 * cgraph.c: Likewise.
15774 * cgraphunit.c: Likewise.
15775 * convert.c: Likewise.
15776 * dwarf2out.c: Likewise.
15777 * final.c: Likewise.
15778 * fold-const.c: Likewise.
15779 * function.c: Likewise.
15780 * gimple-expr.c: Likewise.
15781 * gimple-fold.c: Likewise.
15782 * gimple-pretty-print.c: Likewise.
15783 * gimple.c: Likewise.
15784 * gimplify.c: Likewise.
15785 * hsa-common.c: Likewise.
15786 * hsa-gen.c: Likewise.
15787 * internal-fn.c: Likewise.
15788 * ipa-chkp.c: Likewise.
15789 * ipa-cp.c: Likewise.
15790 * ipa-devirt.c: Likewise.
15791 * ipa-fnsummary.c: Likewise.
15792 * ipa-inline.c: Likewise.
15793 * ipa-visibility.c: Likewise.
15795 * lto-cgraph.c: Likewise.
15796 * omp-expand.c: Likewise.
15797 * omp-general.c: Likewise.
15798 * omp-low.c: Likewise.
15799 * omp-offload.c: Likewise.
15800 * omp-simd-clone.c: Likewise.
15801 * opts-global.c: Likewise.
15802 * passes.c: Likewise.
15803 * predict.c: Likewise.
15804 * sancov.c: Likewise.
15805 * sanopt.c: Likewise.
15806 * symtab.c: Likewise.
15807 * toplev.c: Likewise.
15808 * trans-mem.c: Likewise.
15809 * tree-chkp.c: Likewise.
15810 * tree-eh.c: Likewise.
15811 * tree-into-ssa.c: Likewise.
15812 * tree-object-size.c: Likewise.
15813 * tree-parloops.c: Likewise.
15814 * tree-profile.c: Likewise.
15815 * tree-ssa-ccp.c: Likewise.
15816 * tree-ssa-live.c: Likewise.
15817 * tree-ssa-loop.c: Likewise.
15818 * tree-ssa-sccvn.c: Likewise.
15819 * tree-ssa-structalias.c: Likewise.
15820 * tree-ssa.c: Likewise.
15821 * tree-streamer-in.c: Likewise.
15822 * tree-vectorizer.c: Likewise.
15823 * tree-vrp.c: Likewise.
15824 * tsan.c: Likewise.
15825 * ubsan.c: Likewise.
15826 * varasm.c: Likewise.
15827 * varpool.c: Likewise.
15828 * tree.c: Remove functions moved to attribs.[ch].
15829 * tree.h: Likewise.
15830 * config/aarch64/aarch64.c: Add attrs.h header file.
15831 * config/alpha/alpha.c: Likewise.
15832 * config/arc/arc.c: Likewise.
15833 * config/arm/arm.c: Likewise.
15834 * config/avr/avr.c: Likewise.
15835 * config/bfin/bfin.c: Likewise.
15836 * config/c6x/c6x.c: Likewise.
15837 * config/cr16/cr16.c: Likewise.
15838 * config/cris/cris.c: Likewise.
15839 * config/darwin.c: Likewise.
15840 * config/epiphany/epiphany.c: Likewise.
15841 * config/fr30/fr30.c: Likewise.
15842 * config/frv/frv.c: Likewise.
15843 * config/ft32/ft32.c: Likewise.
15844 * config/h8300/h8300.c: Likewise.
15845 * config/i386/winnt.c: Likewise.
15846 * config/ia64/ia64.c: Likewise.
15847 * config/iq2000/iq2000.c: Likewise.
15848 * config/lm32/lm32.c: Likewise.
15849 * config/m32c/m32c.c: Likewise.
15850 * config/m32r/m32r.c: Likewise.
15851 * config/m68k/m68k.c: Likewise.
15852 * config/mcore/mcore.c: Likewise.
15853 * config/microblaze/microblaze.c: Likewise.
15854 * config/mips/mips.c: Likewise.
15855 * config/mmix/mmix.c: Likewise.
15856 * config/mn10300/mn10300.c: Likewise.
15857 * config/moxie/moxie.c: Likewise.
15858 * config/msp430/msp430.c: Likewise.
15859 * config/nds32/nds32-isr.c: Likewise.
15860 * config/nds32/nds32.c: Likewise.
15861 * config/nios2/nios2.c: Likewise.
15862 * config/nvptx/nvptx.c: Likewise.
15863 * config/pa/pa.c: Likewise.
15864 * config/pdp11/pdp11.c: Likewise.
15865 * config/powerpcspe/powerpcspe.c: Likewise.
15866 * config/riscv/riscv.c: Likewise.
15867 * config/rl78/rl78.c: Likewise.
15868 * config/rx/rx.c: Likewise.
15869 * config/s390/s390.c: Likewise.
15870 * config/sh/sh.c: Likewise.
15871 * config/sol2.c: Likewise.
15872 * config/sparc/sparc.c: Likewise.
15873 * config/spu/spu.c: Likewise.
15874 * config/stormy16/stormy16.c: Likewise.
15875 * config/tilegx/tilegx.c: Likewise.
15876 * config/tilepro/tilepro.c: Likewise.
15877 * config/v850/v850.c: Likewise.
15878 * config/vax/vax.c: Likewise.
15879 * config/visium/visium.c: Likewise.
15880 * config/xtensa/xtensa.c: Likewise.
15882 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15885 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15886 constraints since the -mupper-regs-* switches have been
15888 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15889 into a vector from a double word element that was extracted from
15890 another vector, and eliminate extra XXPERMDI instructions.
15891 (vsx_concat_<mode>_2): Likewise.
15892 (vsx_concat_<mode>_3): Likewise.
15893 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15894 concat to allow optimizing inserts from previous extracts.
15896 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15898 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15899 memory reference to a SSP offset in TLS address space.
15900 (ix86_print_operand) <case '@'>: Remove.
15901 (ix86_print_operand_punct_valid_p): Remove '@' code.
15902 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15903 UNSPEC_SP_TLS_TEST.
15904 (stack_tls_protect_set_<mode>): Remove.
15905 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15906 (stack_tls_protect_test_<mode>): Remove.
15907 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15909 2017-08-07 Olivier Hainque <hainque@adacore.com>
15912 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15914 2017-08-07 Douglas Rupp <rupp@adacore.com>
15916 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15917 variable was referenced as multidir in command.
15919 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15922 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15925 2017-08-07 Martin Liska <mliska@suse.cz>
15927 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15928 * config/rl78/rl78.c: Add include of attribs.h.
15929 * config/sh/sh.c: Likewise.
15930 * config/v850/v850.c: Likewise.
15932 2017-08-07 Tom de Vries <tom@codesourcery.com>
15934 PR middle-end/78266
15935 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15937 2017-08-07 Martin Liska <mliska@suse.cz>
15939 * config/mips/mips.c: Include attribs.h.
15941 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15944 * doc/invoke.texi: Document change in behvaior for -Ofast for
15947 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15949 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15951 (aarch64_pop_regs): Likewise.
15952 (aarch64_gen_load_pair): Likewise.
15953 (aarch64_save_callee_saves): Likewise.
15954 (aarch64_restore_callee_saves): Likewise.
15956 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15958 * config/i386/i386.c: Revert the last change.
15960 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15963 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15965 (ix86_finalize_stack_frame_flags): This. Also clear
15966 frame_pointer_needed if -fno-omit-frame-pointer is used without
15968 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15969 with ix86_finalize_stack_frame_flags.
15970 (ix86_expand_epilogue): Likewise.
15971 (ix86_expand_split_stack_prologue): Likewise.
15973 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15976 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15977 to P_AES for Westmere.
15979 2017-08-07 Jonathan Yong <10walls@gmail.com>
15981 * config/i386/mingw.opt (fset-stack-executable): Removed.
15982 * config/i386/cygming.opt (fset-stack-executable): Moved
15984 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15986 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15988 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15990 2017-08-07 Marek Polacek <polacek@redhat.com>
15992 PR middle-end/81737
15993 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15995 2017-08-07 Martin Liska <mliska@suse.cz>
15997 * attribs.h (canonicalize_attr_name): New function.
15998 (cmp_attribs): Move from c-format.c and adjusted.
15999 (is_attribute_p): Moved from tree.h.
16000 * tree-inline.c: Add new includes.
16001 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16002 (private_is_attribute_p): Remove.
16003 (private_lookup_attribute): Likewise.
16004 (private_lookup_attribute_by_prefix): Simplify.
16005 (remove_attribute): Use is_attribute_p.
16006 * tree.h: Remove removed declarations.
16008 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16010 PR middle-end/81698
16011 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16012 instead of computing it in the function. Formatting fix.
16013 (expand_case): Don't rely on default_edge being the first edge,
16014 clear it if removing it, pass default_edge to
16015 emit_case_dispatch_table.
16016 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16019 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16021 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16022 insn in the function, emit NOP after the insn.
16024 2017-08-06 Tom de Vries <tom@codesourcery.com>
16026 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16029 2017-08-06 Tom de Vries <tom@codesourcery.com>
16031 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16034 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16036 PR tree-optimization/57371
16037 * match.pd: New pattern.
16039 2017-08-04 Marek Polacek <polacek@redhat.com>
16041 PR middle-end/81695
16042 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16043 perform the computation in offset_int.
16045 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16047 PR tree-optimization/81136
16048 * tree-vectorizer.h: Include tree-hash-traits.h.
16049 (vec_base_alignments): New typedef.
16050 (vec_info): Add a base_alignments field.
16051 (vect_record_base_alignments): Declare.
16052 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16054 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16055 (create_data_ref): Add an is_conditional_in_stmt argument.
16056 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16057 the is_conditional_in_stmt field.
16058 (data_ref_loc): Add an is_conditional_in_stmt field.
16059 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16060 (find_data_references_in_stmt): Update call to create_data_ref.
16061 (graphite_find_data_references_in_stmt): Likewise.
16062 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16063 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16064 (vect_record_base_alignment): New function.
16065 (vect_record_base_alignments): Likewise.
16066 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16067 for nested statements even if we fail to compute a misalignment.
16068 Use pooled base alignments for unconditional references.
16069 (vect_find_same_alignment_drs): Compare base addresses instead
16071 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16072 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16074 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16076 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16077 Add an explicit name for the enum. Use auto_vec for slp_instances
16078 and grouped_stores.
16079 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16081 (_bb_vec_info): Add a constructor and destructor.
16082 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16083 (destroy_loop_vec_info): Delete.
16084 (vect_destroy_datarefs): Likewise.
16085 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16086 (vec_info::vec_info): New function.
16087 (vec_info::~vec_info): Likewise.
16088 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16089 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16090 destroy_loop_vec_info.
16091 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16092 (_loop_vec_info::_loop_vec_info): ...this.
16093 (destroy_loop_vec_info): Replace with...
16094 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16095 the stmt_vec_infos. Leave handling of vec_info information to its
16096 destructor. Remove explicit vector releases.
16097 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16098 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16099 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16100 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16101 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16102 (destroy_bb_vec_info): Replace with...
16103 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16104 information to its destructor.
16105 (vect_slp_analyze_bb_1): Use new and delete instead of
16106 new_bb_vec_info and destroy_bb_vec_info.
16107 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16110 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16112 * tree-data-ref.h (subscript): Add access_fn field.
16113 (data_dependence_relation): Add could_be_independent_p.
16114 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16115 (same_access_functions): Move to tree-data-ref.c.
16116 * tree-data-ref.c (ref_contains_union_access_p): New function.
16117 (access_fn_component_p): Likewise.
16118 (access_fn_components_comparable_p): Likewise.
16119 (dr_analyze_indices): Add a reference to access_fn_component_p.
16120 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16122 (constant_access_functions): Likewise.
16123 (add_other_self_distances): Likewise.
16124 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16125 (initialize_data_dependence_relation): Use XCNEW and remove
16126 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16127 of access functions that have the same type. Allow the
16128 subsequence to end with different bases in some circumstances.
16129 Record the chosen access functions in SUB_ACCESS_FN.
16130 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16131 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16132 (subscript_dependence_tester_1): Likewise dra and drb.
16133 (build_classic_dist_vector): Update calls accordingly.
16134 (subscript_dependence_tester): Likewise.
16135 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16136 DDR_COULD_BE_INDEPENDENT_P.
16137 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16138 comp_alias_ddrs instead of may_alias_ddrs.
16139 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16141 (vect_analyze_data_ref_dependence): Use it if
16142 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16143 distance vectors if that fails.
16144 (dependence_distance_ge_vf): New function.
16145 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16146 LOOP_VINFO_MAY_ALIAS_DDRS.
16148 2017-08-04 Richard Biener <rguenther@suse.de>
16150 PR middle-end/81705
16151 * fold-const.c (fold_binary_loc): Properly restrict
16152 minus_var0 && minus_var1 case when associating undefined overflow
16155 2017-08-04 Tom de Vries <tom@codesourcery.com>
16157 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16159 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16162 Don't start diagnostic messages with a capital letter.
16163 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16165 (rs6000_invalid_builtin): Likewise.
16166 (rs6000_trampoline_init): Likewise.
16168 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16171 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16172 after setting changeable df flags.
16174 2017-08-03 Richard Biener <rguenther@suse.de>
16176 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16177 up if the use is in USE - X.
16179 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16181 * toplev.c (dumpfile.h): New include.
16182 (internal_error_reentered): New static function. Use it...
16183 (internal_error_function): ...here to handle reentered internal_error.
16185 2017-08-03 Richard Biener <rguenther@suse.de>
16187 PR middle-end/81148
16188 * fold-const.c (split_tree): Add minus_var and minus_con
16189 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16190 here but always use minus_*.
16191 (associate_trees): Assert we never associate with MINUS_EXPR
16192 and NULL first operand. Do not recurse for PLUS_EXPR operands
16193 when associating as MINUS_EXPR either.
16194 (fold_binary_loc): Track minus_var and minus_con.
16196 2017-08-03 Tom de Vries <tom@codesourcery.com>
16199 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16200 ACCEL_COMPILER, apply finish_options on
16201 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16203 2017-08-03 Tom de Vries <tom@codesourcery.com>
16206 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16207 function_entry_patch_area_size > 0.
16209 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16212 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16213 instead of 10??LU, perform unit multiplication in wide_int,
16214 don't change alloc_object_size_limit if the limit is larger
16217 PR tree-optimization/81655
16218 PR tree-optimization/81588
16219 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16220 the case when ranges[i].low and high are 1 for unsigned type with
16223 PR middle-end/81052
16224 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16225 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16227 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16229 * tree-vrp.h: Add include guard.
16231 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16234 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16235 (ud2): New insn pattern.
16236 * config/i386/i386.c (ix86_expand_epilogue):
16237 For naked functions, generate ud2 instead of trap insn.
16239 2017-08-02 Marek Polacek <polacek@redhat.com>
16242 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16244 2017-08-02 Tom de Vries <tom@codesourcery.com>
16245 Cesar Philippidis <cesar@codesourcery.com>
16247 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16248 Add missing edge probabilities.
16250 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16252 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16253 Correct endianness.
16255 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16257 PR middle-end/79499
16258 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16259 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16260 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16261 prologue_seq sequences - if any.
16263 2017-08-02 Richard Biener <rguenther@suse.de>
16265 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16266 via vectors if supported, integer extracts via punning if supported
16267 or otherwise vector extracts.
16269 2017-08-02 Richard Biener <rguenther@suse.de>
16271 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16273 (bitmap_insert_into_set): ... this.
16275 2017-08-02 Richard Biener <rguenther@suse.de>
16277 PR tree-optimization/81633
16279 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16281 PR tree-optimization/71752
16282 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16284 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16286 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16287 (machine_function::call_ms2sysv_pad_out): Remove field.
16288 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16289 (ix86_compute_frame_layout): Likewise.
16291 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16294 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16295 attribute with interrupt attribute.
16297 2017-08-01 Andrew Pinski <apinski@cavium.com>
16299 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16300 BIT_INSERT_EXPR's operand 1
16301 to see if the types precision matches.
16303 2017-08-01 Martin Liska <mliska@suse.cz>
16305 PR middle-end/70140
16306 * builtins.c (expand_builtin_memcpy_args): Remove.
16307 (expand_builtin_memcpy): Call newly added function
16308 expand_builtin_memory_copy_args.
16309 (expand_builtin_memcpy_with_bounds): Likewise.
16310 (expand_builtin_mempcpy): Remove last argument.
16311 (expand_builtin_mempcpy_with_bounds): Likewise.
16312 (expand_builtin_memory_copy_args): New function created from
16313 expand_builtin_mempcpy_args with small modifications.
16314 (expand_builtin_mempcpy_args): Remove.
16315 (expand_builtin_stpcpy): Remove unused argument.
16316 (expand_builtin): Likewise.
16317 (expand_builtin_with_bounds): Likewise.
16319 2017-08-01 Martin Liska <mliska@suse.cz>
16322 Make mempcpy more optimal (PR middle-end/70140).
16324 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16328 __builtin_vec_cmpne verify both arguments are compatible vectors
16329 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16330 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16331 move computation of aligned to after checking the argument types.
16335 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16336 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16339 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16340 Doug Rupp <rupp@adacore.com>
16341 Olivier Hainque <hainque@adacore.com>
16343 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16344 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16346 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16347 for TARGET_OS_CPP_BUILTIN.
16348 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16349 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16351 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16352 passing required abi options to the assembler for EABI configurations.
16353 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16354 of .text.hot and .text.unlikely sections for kernel modules when
16355 using ARM style exceptions.
16356 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16357 options. Add EXTRA_CC1_SPEC.
16358 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16360 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16362 (ARM_TARGET2_DWARF_FORMAT): Define.
16363 * config/arm/t-vxworks: Adjust multilib control to removal of the
16364 Diab command line options.
16366 2017-08-01 Martin Liska <mliska@suse.cz>
16368 PR gcov-profile/81561
16369 * gcov.c (unblock): Make unblocking safe as we need to preserve
16370 index correspondence of blocks and block_lists.
16372 2017-08-01 Richard Biener <rguenther@suse.de>
16374 PR tree-optimization/81181
16375 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16376 (compute_antic): ... end of iteration here.
16378 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16380 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16381 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16382 (ftree-slp-vectorize): Likewise.
16383 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16384 can no longer be set independent of flag_tree_loop_vectorize.
16385 * omp-general.c (emp_max_vf): Likewise.
16386 * opts.c (enable_fdo_optimizations): Remove references to
16387 flag_tree_vectorize, these are now implicit.
16388 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16389 and leave it for the options machinery.
16391 2017-08-01 Martin Liska <mliska@suse.cz>
16393 PR middle-end/70140
16394 * builtins.c (expand_builtin_memcpy_args): Remove.
16395 (expand_builtin_memcpy): Call newly added function
16396 expand_builtin_memory_copy_args.
16397 (expand_builtin_memcpy_with_bounds): Likewise.
16398 (expand_builtin_mempcpy): Remove last argument.
16399 (expand_builtin_mempcpy_with_bounds): Likewise.
16400 (expand_builtin_memory_copy_args): New function created from
16401 expand_builtin_mempcpy_args with small modifications.
16402 (expand_builtin_mempcpy_args): Remove.
16403 (expand_builtin_stpcpy): Remove unused argument.
16404 (expand_builtin): Likewise.
16405 (expand_builtin_with_bounds): Likewise.
16407 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16410 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16411 print "ds:" only for immediates in generic address space.
16413 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16416 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16417 (ix86_function_ok_for_sibcall): Return false for naked functions.
16419 2017-08-01 Richard Biener <rguenther@suse.de>
16421 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16422 (compute_antic): Seed worklist with exit block predecessors.
16423 * cfganal.c (dfs_find_deadend): For a cycle return the source
16424 of the edge closing it.
16426 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16428 * config/aarch64/aarch64.c
16429 (aarch64_can_const_movi_rtx_p): Move 0 check.
16431 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16433 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16434 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16436 * match.pd: Ditto in address comparison pattern.
16438 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16440 PR tree-optimization/81627
16441 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16442 closed ssa form for store-store chain.
16444 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16446 PR tree-optimization/81620
16447 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16448 for store-store chain.
16450 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16452 PR tree-optimization/81588
16453 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16454 ranges[i].in_p, invert comparison code ccode. For >/>=,
16455 swap rhs1 and rhs2 and comparison code unconditionally,
16456 for </<= don't do that. Don't swap rhs1/rhs2 again if
16457 ranges[i].in_p, instead invert comparison code ccode if
16458 opcode or oe->rank is BIT_IOR_EXPR.
16461 * optabs.def (vec_extract_optab, vec_init_optab): Change from
16462 a direct optab to conversion optab.
16463 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
16464 with GET_MODE_INNER as last argument instead of optab_handler.
16465 * expmed.c (extract_bit_field_1): Likewise. Use vector from
16466 vector extraction if possible and optab is available.
16467 * expr.c (store_constructor): Use convert_optab_handler instead
16468 of optab_handler. Use vector initialization from smaller
16469 vectors if possible and optab is available.
16470 * tree-vect-stmts.c (vectorizable_load): Likewise.
16471 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16472 now have two modes.
16473 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
16474 of vec_init from half-sized vectors with the same element mode.
16475 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
16476 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
16477 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
16478 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
16479 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
16480 after mode in gen_vec_extract* calls.
16481 (vec_extract<mode>): Renamed to ...
16482 (vec_extract<mode><ssescalarmodelower>): ... this.
16483 (vec_extract<mode><ssehalfvecmodelower>): New expander.
16484 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
16485 element mode after mode in gen_vec_init* calls.
16486 (VEC_INIT_HALF_MODE): New mode iterator.
16487 (vec_init<mode>): Renamed to ...
16488 (vec_init<mode><ssescalarmodelower>): ... this.
16489 (vec_init<mode><ssehalfvecmodelower>): New expander.
16490 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
16491 (vec_extractv2sfsf): ... this.
16492 (vec_initv2sf): Renamed to ...
16493 (vec_initv2sfsf): ... this.
16494 (vec_extractv2si): Renamed to ...
16495 (vec_extractv2sisi): ... this.
16496 (vec_initv2si): Renamed to ...
16497 (vec_initv2sisi): ... this.
16498 (vec_extractv4hi): Renamed to ...
16499 (vec_extractv4hihi): ... this.
16500 (vec_initv4hi): Renamed to ...
16501 (vec_initv4hihi): ... this.
16502 (vec_extractv8qi): Renamed to ...
16503 (vec_extractv8qiqi): ... this.
16504 (vec_initv8qi): Renamed to ...
16505 (vec_initv8qiqi): ... this.
16506 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16507 (vec_init<mode>): Renamed to ...
16508 (vec_init<mode><VEC_base_l>): ... this.
16509 (vec_extract<mode>): Renamed to ...
16510 (vec_extract<mode><VEC_base_l>): ... this.
16511 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16512 (vec_initv2sfsf): ... this.
16513 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16514 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16515 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16516 element mode after mode in gen_vec_init* calls.
16517 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16518 (vec_init<mode><Vel>): ... this.
16519 (vec_extract<mode>): Renamed to ...
16520 (vec_extract<mode><Vel>): ... this.
16521 * config/aarch64/iterators.md (Vel): New mode attribute.
16522 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16523 Add element mode after mode in gen_vec_extract* calls.
16524 * config/s390/vector.md (non_vec_l): New mode attribute.
16525 (vec_extract<mode>): Renamed to ...
16526 (vec_extract<mode><non_vec_l>): ... this.
16527 (vec_init<mode>): Renamed to ...
16528 (vec_init<mode><non_vec_l>): ... this.
16529 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16530 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16532 * config/arm/iterators.md (V_elem_l): New mode attribute.
16533 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16534 (vec_extract<mode><V_elem_l>): ... this.
16535 (vec_extractv2di): Renamed to ...
16536 (vec_extractv2didi): ... this.
16537 (vec_init<mode>): Renamed to ...
16538 (vec_init<mode><V_elem_l>): ... this.
16539 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16540 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16541 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16542 Add element mode after gen_vec_extract* calls.
16543 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16544 (vec_init<mode><unitmode>): ... this.
16545 (vec_extract<mode>): Renamed to ...
16546 (vec_extract<mode><unitmode>): ... this.
16547 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16548 (vec_init<mode><unitmode>): ... this.
16549 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16550 (vec_initv2sfsf): ... this.
16551 (vec_extractv2sf): Renamed to ...
16552 (vec_extractv2sfsf): ... this.
16553 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16554 Add element mode after gen_vec_extract* calls.
16555 * config/mips/mips.md (unitmode): New mode iterator.
16556 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16557 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16558 * config/spu/spu.md (inner_l): New mode attribute.
16559 (vec_init<mode>): Renamed to ...
16560 (vec_init<mode><inner_l>): ... this.
16561 (vec_extract<mode>): Renamed to ...
16562 (vec_extract<mode><inner_l>): ... this.
16563 * config/sparc/sparc.md (veltmode): New mode iterator.
16564 (vec_init<VMALL:mode>): Renamed to ...
16565 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16566 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16567 (vec_initv2sisi): ... this.
16568 (vec_initv2sf): Renamed to ...
16569 (vec_initv2sfsf): ... this.
16570 (vec_extractv2sf): Renamed to ...
16571 (vec_extractv2sfsf): ... this.
16572 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16573 (vec_init<mode>): Renamed to ...
16574 (vec_init<mode><VEC_base_l>): ... this.
16575 (vec_extract<mode>): Renamed to ...
16576 (vec_extract<mode><VEC_base_l>): ... this.
16577 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16578 (vec_initv2sfsf): ... this.
16579 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16580 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16581 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16582 gen_vec_init* calls.
16584 2017-08-01 Richard Biener <rguenther@suse.de>
16586 PR tree-optimization/81297
16587 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16588 TREE_OVERFLOW from INTEGER_CSTs.
16590 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16592 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16594 2017-07-31 Carl Love <cel@us.ibm.com>
16596 * config/rs6000/rs6000-c: Add support for built-in functions
16597 vector signed char vec_xl_be (signed long long, signed char *);
16598 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16599 vector signed int vec_xl_be (signed long long, signed int *);
16600 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16601 vector signed long long vec_xl_be (signed long long, signed long long *);
16602 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16603 vector signed short vec_xl_be (signed long long, signed short *);
16604 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16605 vector double vec_xl_be (signed long long, double *);
16606 vector float vec_xl_be (signed long long, float *);
16607 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16608 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16609 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16611 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16612 (altivec_expand_builtin): Add switch statement to call
16613 altivec_expand_xl_be for each builtin.
16614 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16615 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16616 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16617 __builtin_vsx_le_be_v16qi.
16618 * doc/extend.texi: Update the built-in documentation file for the
16619 new built-in functions.
16621 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16624 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16626 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16628 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16630 * config.gcc: Add z14.
16631 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16632 CPU model numbers for z13s and z14.
16633 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16635 * config/s390/s390-opts.h (enum processor_type): Rename
16636 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16637 * config/s390/s390.c (processor_table): Add field for CPU name to
16638 be passed to Binutils.
16639 (s390_asm_output_machine_for_arch): Use the new field in
16640 processor_table for Binutils.
16641 (s390_expand_builtin): Replace arch12 with z14.
16642 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16643 (s390_get_sched_attrmask): Likewise.
16644 (s390_get_unit_mask): Likewise.
16645 * config/s390/s390.opt: Add z14 to processor_type enum.
16647 2017-07-31 Martin Jambor <mjambor@suse.cz>
16650 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16651 regardless of optimization level.
16653 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16654 Martin Liska <mliska@suse.cz>
16656 * predict.def: Remove old comment and adjust probability.
16657 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16658 PREDICT statements.
16660 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16663 * config/i386/i386.c (ix86_function_naked): New function.
16664 (ix86_can_use_return_insn_p): Return false for naked functions.
16665 (ix86_expand_prologue): Skip prologue for naked functions.
16666 (ix86_expand_epilogue): Skip epilogue for naked functions
16667 and emit trap instruction.
16668 (ix86_warn_func_return): New function.
16669 (ix86_attribute_table): Add "naked" attribute specification.
16670 (TARGET_WARN_FUNC_RETURN): Define.
16671 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16673 2017-07-31 Martin Liska <mliska@suse.cz>
16675 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16676 (dump_gimple_bb_header): Always dump BB info.
16677 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16679 2017-07-31 Martin Liska <mliska@suse.cz>
16682 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16683 also with current_function_decl non-null equality.
16685 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16688 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16689 change type to the element type, instead add eltype variable and
16690 use it where we are interested in the element type.
16692 PR tree-optimization/81603
16693 * ipa-polymorphic-call.c
16694 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16695 offset arithmetic in offset_int, bail out if the resulting bit offset
16696 doesn't fit into shwi.
16698 2017-07-31 Martin Liska <mliska@suse.cz>
16700 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16701 (gimplify_save_expr): Fix comment.
16703 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16706 * config/i386/i386.c (ix86_function_arg): Update arguments for
16708 (ix86_compute_frame_layout): Set the initial stack offset to
16709 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16710 INCOMING_FRAME_SP_OFFSET.
16711 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16712 stack before exception handler returns.
16713 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16714 the 'ERROR_CODE' for exception handler.
16716 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16718 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16719 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16720 (ASM_OUTPUT_REG_POP): Ditto.
16721 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16722 instead of asm_fprintf to output pure string.
16724 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16726 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16727 to imported_module_or_decl hook.
16728 (debug_nothing_tree_tree_tree_bool): Remove.
16729 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16730 * debug.c (do_nothing_debug_hooks): Use
16731 debug_nothing_tree_tree_tree_bool_bool instead of
16732 debug_nothing_tree_tree_tree_bool.
16733 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16734 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16735 * sdbout.c (sdb_debug_hooks): Likewise.
16736 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16737 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16739 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16740 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16741 attribute, don't add anything.
16743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16745 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16746 (fold_build2_stat_loc): Likewise.
16747 (fold_build3_stat_loc): Likewise.
16748 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16749 (fold_build1_loc): Remove macro.
16750 (fold_build2_loc): Likewise.
16751 (fold_build3_loc): Likewise.
16753 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16755 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16756 (gimple_build_debug_bind_source_stat): Likewise.
16757 * gimple.h (gimple_build_debug_bind): Remove macro.
16758 (gimple_build_debug_bind_source): Likewise.
16760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16762 * bitmap.c (bitmap_alloc): Adjust.
16763 (bitmap_gc_alloc): Likewise.
16764 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16766 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16768 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16769 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16770 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16771 (bitmap_gc_alloc_stat): Likewise.
16772 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16776 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16777 * rtl.h (shallow_copy_rtx): Remove macro.
16779 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16781 * emit-rtl.c (gen_raw_REG): Adjust.
16782 * gengenrtl.c (gendef): Likewise.
16783 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16784 * rtl.h (rtx_alloc): Remove macro.
16786 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16788 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16789 (build_tree_list_stat): Likewise.
16790 * tree.h (build_tree_list): Remove macro.
16791 (build_tree_list_vec): Likewise.
16793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16795 * tree.c (make_vector_stat): Remove _stat from name.
16796 (build_vector_stat): Likewise.
16797 * tree.h (make_vector_stat): Remove macro.
16798 (build_vector_stat): Likewise.
16800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16802 * tree.h (build_var_debug_value): Remove prototype.
16804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16806 * tree.c (tree_cons_stat): Remove _stat from name.
16807 * tree.h (tree_cons): Remove macro.
16809 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16811 * tree.c (build_vl_exp_stat): Remove _stat from name.
16812 * tree.h (build_vl_exp): Remove macro.
16814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16816 * tree.c (build_decl_stat): Remove _stat from name.
16817 * tree.h (build_decl): Remove macro.
16819 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16821 * gimple.c (gimple_build_with_ops_stat): Adjust.
16822 (gimple_alloc_stat): Remove _stat from name.
16823 * gimple.h (gimple_alloc): Remove macro.
16825 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16827 * tree.c (make_tree_vec_stat): Remove _stat from name.
16828 (grow_tree_vec_stat): Likewise.
16829 * tree.h (make_tree_vec_stat): Adjust prototype.
16830 (grow_tree_vec_stat): Likewise.
16831 (make_tree_vec): Remove macro.
16832 (grow_tree_vec): Likewise.
16834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16836 * fold-const.c (fold_build1_stat_loc): Adjust.
16837 (fold_build2_stat_loc): Likewise.
16838 (fold_build3_stat_loc): Likewise.
16839 * tree.c (build0_stat): Remove _stat from name.
16840 (build1_stat): Likewise.
16841 (build2_stat): Likewise.
16842 (build3_stat): Likewise.
16843 (build4_stat): Likewise.
16844 (build5_stat): Likewise.
16845 * tree.h (build1_loc): Remove macro, and rename _stat function
16847 (build2_loc): Likewise.
16848 (build3_loc): Likewise.
16849 (build4_loc): Likewise.
16850 (build5_loc): Likewise.
16852 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16854 * tree.c (make_int_cst_stat): Remove _stat from name.
16855 * tree.h (make_int_cst_stat): Adjust prototype.
16856 (make_int_cst): Remove macro.
16858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16860 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16861 * tree.h (make_tree_binfo_stat): Adjust prototype.
16862 (make_tree_binfo): Remove.
16864 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16866 * tree.c (copy_node_stat): Rename to copy_node.
16867 (build_distinct_type_copy): Adjust.
16868 * tree.h (copy_node_stat): Adjust prototype.
16869 (copy_node): Remove macro.
16871 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16873 * tree.c (make_node_stat): rename to make_node.
16874 (build_tree_list_stat): Adjust.
16875 (build0_stat): Likewise.
16876 (build2_stat): Likewise.
16877 (build3_stat): Likewise.
16878 (build4_stat): Likewise.
16879 (build5_stat): Likewise.
16880 (build_decl_stat): Likewise.
16881 * tree.h (make_node_stat): Adjust prototype.
16882 (make_node): remove macro.
16884 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16886 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16887 (PPC_FEATURE2_SCV): Likewise.
16888 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16890 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16892 * config/aarch64/aarch64.c
16893 (aarch64_internal_mov_immediate): Add new special pattern.
16894 * config/aarch64/aarch64.md (*movdi_aarch64):
16895 Add reg/32bit const mov case.
16897 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16898 Richard Sandiford <richard.sandiford@linaro.org>
16900 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16901 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16902 Add integer and movi cases.
16903 (movi-split-hf-df-sf split, fp16): New.
16904 (enabled): Added TARGET_FP_F16INST.
16905 * config/aarch64/iterators.md (GPF_HF): New.
16906 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16908 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16910 * config/aarch64/aarch64.c
16911 (aarch64_simd_container_mode): Add prototype.
16912 (aarch64_expand_mov_immediate): Add HI support.
16913 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16914 (aarch64_can_const_movi_rtx_p): New.
16915 (aarch64_preferred_reload_class):
16916 Remove restrictions of using FP registers for certain SIMD operations.
16917 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16918 (aarch64_valid_floating_const): Add integer move validation.
16919 (aarch64_simd_imm_scalar_p): Remove.
16920 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16921 (aarch64_legitimate_constant_p): Expand list of supported cases.
16922 * config/aarch64/aarch64-protos.h
16923 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16924 (aarch64_reinterpret_float_as_int): New.
16925 (aarch64_simd_imm_scalar_p): Remove.
16926 * config/aarch64/constraints.md (Uvi): New.
16927 (Dd): Split into Ds and new Dd.
16928 * config/aarch64/aarch64.md (*movsi_aarch64):
16930 (*movdi_aarch64): Add SIMD mov case.
16932 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16934 * tree-predcom.c: (struct chain): Handle store-store chain in which
16935 stores for elimination only store loop invariant values.
16936 (execute_pred_commoning_chain): Ditto.
16937 (prepare_initializers_chain_store_elim): Ditto.
16938 (prepare_finalizers): Ditto.
16939 (is_inv_store_elimination_chain): New function.
16940 (initialize_root_vars_store_elim_1): New function.
16942 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16944 * tree-predcom.c: Revise general description of the pass.
16945 (enum chain_type): New enum type for store elimination.
16946 (struct chain): New field supporting store elimination.
16947 (struct component): Ditto.
16948 (dump_chain): Dump store-stores chain.
16949 (release_chain): Release resources.
16950 (split_data_refs_to_components): Compute and create component
16951 contains only stores for elimination.
16952 (get_chain_last_ref_at): New function.
16953 (make_invariant_chain): Initialization.
16954 (make_rooted_chain): Specify chain type in parameter and record it.
16955 (add_looparound_copies): Skip for store-stores chain.
16956 (determine_roots_comp): Compute type of chain and pass it to
16958 (initialize_root_vars_store_elim_2): New function.
16959 (finalize_eliminated_stores): New function.
16960 (remove_stmt): Handle store for elimination.
16961 (execute_pred_commoning_chain): Execute predictive commoning on
16962 store-store chains.
16963 (determine_unroll_factor): Skip unroll for store-stores chain.
16964 (prepare_initializers_chain_store_elim): New function.
16965 (prepare_initializers_chain): Hanlde store-store chain.
16966 (prepare_finalizers_chain, prepare_finalizers): New function.
16967 (tree_predictive_commoning_loop): Return integer value indicating
16968 if loop is unrolled or lcssa form is corrupted.
16969 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16971 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16973 * tree-predcom.c (initialize_root): Delete.
16974 (execute_pred_commoning_chain): Initialize root vars and replace
16975 reference of non-combined chain directly, rather than call above
16978 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16980 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16981 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16983 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16985 * tree-predcom.c (struct chain): New field init_seq.
16986 (release_chain): Release init_seq.
16987 (prepare_initializers_chain): Record intialization stmts in above
16989 (insert_init_seqs): New function.
16990 (tree_predictive_commoning_loop): Call insert_init_seqs.
16992 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16994 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16996 2017-07-28 Richard Biener <rguenther@suse.de>
16998 * match.pd: Remove superfluous :c.
16999 * genmatch.c (simplify::id): Add member.
17000 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17002 (current_id): New global.
17003 (dt_node::parent): Move from ...
17004 (dt_operand::parent): ... here. Add for_id member.
17005 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17006 (decision_tree::find_node): Relax order requirement when
17007 merging DT_TRUE nodes to ones inbetween the current simplify
17008 and the one we try to merge with. Add diagnostic whenever
17009 we need to enforce pattern order by not merging.
17010 (decision_tree::insert): Set current_id.
17011 (decision_tree::print_node): Dump parent node and for_id.
17012 (parser::last_id): Add member.
17013 (parser::push_simplify): Assign unique id.
17014 (parser::parser): Initialize last_id.
17016 2017-07-28 Martin Liska <mliska@suse.cz>
17019 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17020 gimple_build_debug_bind.
17022 2017-07-28 Richard Biener <rguenther@suse.de>
17024 PR tree-optimization/81502
17025 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17027 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17029 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17030 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17031 for BIT_FIELD_REF args.
17032 * fold-const.c (make_bit_field_ref): Likewise.
17033 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17035 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17038 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17039 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17040 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17041 Or it into SANITIZER_UNDEFINED.
17042 * ubsan.c: Include gimple-fold.h and varasm.h.
17043 (ubsan_expand_ptr_ifn): New function.
17044 (instrument_pointer_overflow): New function.
17045 (maybe_instrument_pointer_overflow): New function.
17046 (instrument_object_size): Formatting fix.
17047 (pass_ubsan::execute): Call instrument_pointer_overflow
17048 and maybe_instrument_pointer_overflow.
17049 * internal-fn.c (expand_UBSAN_PTR): New function.
17050 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17051 * sanitizer.def (__ubsan_handle_pointer_overflow,
17052 __ubsan_handle_pointer_overflow_abort): New builtins.
17053 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17054 * internal-fn.def (UBSAN_PTR): New internal function.
17055 * opts.c (sanitizer_opts): Add pointer-overflow.
17056 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17057 * fold-const.c (build_range_check): Compute pointer range check in
17058 integral type if pointer arithmetics would be needed. Formatting
17061 2017-07-28 Martin Liska <mliska@suse.cz>
17064 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17065 parameters that are of a variable-length.
17067 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17069 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17071 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17072 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17073 (CRT_CALL_STATIC_FUNCTION): Likewise.
17074 (ASM_DEFAULT_SPEC): New define.
17075 (ASM_SPEC32): Likewise.
17076 (ASM_SPEC64): Likewise.
17077 (ASM_SPEC_COMMON): Likewise.
17078 (ASM_SPEC): Likewise.
17079 (INVALID_64BIT): Likewise.
17080 (LINK_OS_DEFAULT_SPEC): Likewise.
17081 (LINK_OS_SPEC32): Likewise.
17082 (LINK_OS_SPEC64): Likewise.
17083 (POWERPC_LINUX): Likewise.
17084 (PTRDIFF_TYPE): Likewise.
17085 (RESTORE_FP_PREFIX): Likewise.
17086 (RESTORE_FP_SUFFIX): Likewise.
17087 (SAVE_FP_PREFIX): Likewise.
17088 (SAVE_FP_SUFFIX): Likewise.
17089 (SIZE_TYPE): Likewise.
17090 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17091 (TARGET_64BIT): Likewise.
17092 (TARGET_64BIT): Likewise.
17093 (TARGET_AIX): Likewise.
17094 (WCHAR_TYPE_SIZE): Likewise.
17095 (WCHAR_TYPE): Undefine.
17096 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17097 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17098 (CPP_OS_RTEMS_SPEC): Delete.
17099 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17100 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17102 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17104 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17106 PR tree-optimization/81578
17107 * tree-parloops.c (build_new_reduction): Bail out if
17108 reduction_code isn't one of the standard OpenMP reductions.
17109 Move the details printing after that decision.
17111 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17113 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17114 related to reload_in_progress.
17115 (splat_input_operand): Likewise.
17116 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17118 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17120 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17121 (TARGET_INSTANTIATE_DECLS): Likewise.
17122 (legitimate_indexed_address_p): Delete reload_in_progress code.
17123 (rs6000_debug_legitimate_address_p): Likewise.
17124 (rs6000_eliminate_indexed_memrefs): Likewise.
17125 (rs6000_emit_le_vsx_store): Likewise.
17126 (rs6000_emit_move_si_sf_subreg): Likewise.
17127 (rs6000_emit_move): Likewise.
17128 (register_to_reg_type): Likewise.
17129 (rs6000_pre_atomic_barrier): Likewise.
17130 (rs6000_machopic_legitimize_pic_address): Likewise.
17131 (rs6000_allocate_stack_temp): Likewise.
17132 (rs6000_address_for_fpconvert): Likewise.
17133 (rs6000_address_for_altivec): Likewise.
17134 (rs6000_secondary_memory_needed_rtx): Delete function.
17135 (rs6000_check_sdmode): Likewise.
17136 (rs6000_alloc_sdmode_stack_slot): Likewise.
17137 (rs6000_instantiate_decls): Likewise.
17138 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17139 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17140 Delete reload_in_progress.
17141 (*vec_reload_and_plus_<mptrsize>): Likewise.
17142 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17143 (vsx_div_v2di): Likewise.
17144 (vsx_udiv_v2di): Likewise.
17146 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17148 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17149 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17150 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17151 (rs6000_debug_reg_global): Delete print of LRA status.
17152 (rs6000_option_override_internal): Delete dead LRA related code.
17153 (rs6000_lra_p): Delete function.
17154 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17156 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17158 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17159 * config/riscv/rtems.h: New file.
17161 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17162 Sudakshina Das <sudi.das@arm.com>
17164 * config/aarch64/aarch64.md
17165 (define_split for and<mode>3nr_compare): Move
17166 non aarch64_logical_operand to a register.
17167 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17168 register immediate operand to a register.
17169 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17171 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17173 PR middle-end/81564
17174 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17176 2017-07-27 Richard Biener <rguenther@suse.de>
17178 PR tree-optimization/81573
17179 PR tree-optimization/81494
17180 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17181 multi defuse cycle case.
17183 2017-07-27 Richard Biener <rguenther@suse.de>
17185 PR tree-optimization/81571
17186 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17189 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17191 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17192 earlier and only if MASK_FPU is set. Adjust formatting.
17194 2017-07-27 Martin Liska <mliska@suse.cz>
17196 * opt-functions.awk: Add validation of value of Init.
17197 * optc-gen.awk: Pass new argument.
17199 2017-07-27 Martin Liska <mliska@suse.cz>
17201 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17202 Fix wrong condition.
17204 2017-07-27 Martin Liska <mliska@suse.cz>
17206 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17207 BBs and edges seen by autoFDO.
17209 2017-07-27 Richard Biener <rguenther@suse.de>
17211 PR tree-optimization/81502
17212 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17213 with incompatible but same sized type.
17214 (execute_update_addresses_taken): Likewise.
17216 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17218 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17219 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17221 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17224 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17225 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17226 Change s_operand to memory_operand.
17228 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17230 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17231 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17232 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17233 Emit instructions rather than returning an expression. Handle TFmode
17234 and KFmode by casting to TImode.
17235 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17236 (rs6000_emit_le_vsx_store): Likewise.
17237 * config/rs6000/vsx.md (VSX_TI): New iterator.
17238 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17239 (*vsx_le_undo_permute_<mode>): Likewise.
17240 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17241 emit the split sequence.
17242 (*vsx_le_perm_store_<mode>): Likewise.
17244 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17246 PR tree-optimization/81555
17247 PR tree-optimization/81556
17248 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17249 if true, force CHANGED for the recursive invocation.
17250 (reassociate_bb): Remember original length of ops array, pass
17251 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17253 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17254 attributes for noipa attribute. For naked attribute use
17255 lookup_attribute first before lookup_attribute_spec.
17256 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17258 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17259 for functions with noipa attribute.
17260 (cgraph_externally_visible_p): Return true for functions with noipa
17262 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17263 for functions with noipa attribute.
17264 * doc/extend.texi: Document noipa function attribute.
17265 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17266 also for functions with noipa attribute.
17267 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17269 2017-07-26 Andrew Pinski <apinski@cavium.com>
17271 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17272 vec_unalign_load_cost and vec_unalign_store_cost.
17274 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17276 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17277 -mvsx-small-integer option.
17278 (ISA_3_0_MASKS_IEEE): Likewise.
17279 (OTHER_VSX_VECTOR_MASKS): Likewise.
17280 (POWERPC_MASKS): Likewise.
17281 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17282 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17283 code, only testing for DImode being allowed in non-VSX floating
17285 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17286 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17288 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17289 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17290 TARGET_P8_VECTOR test.
17291 (rs6000_secondary_reload_simple_move): Likewise.
17292 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17293 since TARGET_P9_VECTOR was already tested.
17294 (rs6000_opt_masks): Remove -mvsx-small-integer.
17295 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17296 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17298 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17299 test for TARGET_VEXTRACTUB was used, and that uses
17301 (p9 extract splitter): Likewise.
17302 (vsx_extract_<mode>_di_p9): Likewise.
17303 (vsx_extract_<mode>_store_p9): Likewise.
17304 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17305 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17306 the elimination of TARGET_VSX_SMALL_INTEGER.
17307 (vsx_extract_<mode>_p8): Likewise.
17308 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17309 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17310 (vsx_set_<mode>_p9): Likewise.
17311 (vsx_set_v4sf_p9): Likewise.
17312 (vsx_set_v4sf_p9_zero): Likewise.
17313 (vsx_insert_extract_v4sf_p9): Likewise.
17314 (vsx_insert_extract_v4sf_p9_2): Likewise.
17315 * config/rs6000/rs6000.md (sign extend splitter): Change
17316 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17317 (floatsi<mode>2_lfiwax_mem): Likewise.
17318 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17319 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17320 since a test for TARGET_P9_VECTOR was used.
17321 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17322 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17323 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17324 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17325 TARGET_P8_VECTOR test.
17326 (fix_trunc<mode>si2_stfiwx): Likewise.
17327 (fix_trunc<mode>si2_internal): Likewise.
17328 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17329 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17331 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17332 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17333 TARGET_P8_VECTOR test.
17334 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17335 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17336 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17338 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17339 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17340 since a test for TARGET_P9_VECTOR was used.
17341 (splitter for loading small constants): Likewise.
17343 2017-07-26 Andrew Pinski <apinski@cavium.com>
17345 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17348 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17351 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17352 (fp_valid_at): Likewise.
17354 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17356 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17357 (qdf24xx_addrcost_table): Likewise.
17358 (cortexa57_tunings): Update to use generic_branch_cost.
17359 (cortexa72_tunings): Likewise.
17360 (cortexa73_tunings): Likewise.
17361 (qdf24xx_tunings): Likewise.
17363 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17365 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17366 (thunderx2t99_branch_cost): Likewise.
17367 (cortexa35_tunings): Update to use generic_branch_cost.
17368 (cortexa53_tunings): Likewise.
17369 (cortexa57_tunings): Likewise.
17370 (cortexa72_tunings): Likewise.
17371 (cortexa73_tunings): Likewise.
17372 (thunderx2t99_tunings): Likewise.
17374 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17376 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17377 (sparc_option_override): Honour MASK_FSMULD.
17378 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17379 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17380 * config/sparc/sparc.opt (mfsmuld): New option.
17381 * doc/invoke.texi (mfsmuld): Document option.
17383 2017-07-26 Marek Polacek <polacek@redhat.com>
17385 PR middle-end/70992
17386 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17388 2017-07-26 Richard Biener <rguenther@suse.de>
17390 * gimple-match-head.c (do_valueize): Return OP if valueize
17392 (get_def): New helper to get at the def stmt of a SSA name
17393 if valueize allows.
17394 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17395 do_valueize to get at the def stmt.
17396 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17398 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17400 PR middle-end/46932
17401 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17403 2017-07-26 Martin Liska <mliska@suse.cz>
17406 * function.c (expand_function_start): Make expansion of
17407 nonlocal_goto_save_area after parm_birth_insn.
17409 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17411 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17412 from all CPU target flags enable members.
17414 2017-07-26 Richard Biener <rguenther@suse.de>
17416 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17417 (decision_tree::gen): Make 'type' const.
17418 (write_predicate): Likewise.
17420 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17422 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17423 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17424 (rs6000_option_override_internal): Likewise.
17425 (rs6000_expand_vector_set): Likewise.
17426 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17427 (TARGET_UPPER_REGS_SF): Likewise.
17428 (TARGET_UPPER_REGS_DI): Likewise.
17429 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17430 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17431 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17432 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17433 (Splitters for DI constants in Altivec registers): Likewise.
17434 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17435 (vsx_set_v4sf_p9): Likewise.
17436 (vsx_set_v4sf_p9_zero): Likewise.
17437 (vsx_insert_extract_v4sf_p9): Likewise.
17438 (vsx_insert_extract_v4sf_p9_2): Likewise.
17440 2017-07-25 Carl Love <cel@us.ibm.com>
17442 * doc/extend.texi: Update the built-in documentation file for the
17443 existing built-in functions
17444 vector signed char vec_cnttz (vector signed char);
17445 vector unsigned char vec_cnttz (vector unsigned char);
17446 vector signed short vec_cnttz (vector signed short);
17447 vector unsigned short vec_cnttz (vector unsigned short);
17448 vector signed int vec_cnttz (vector signed int);
17449 vector unsigned int vec_cnttz (vector unsigned int);
17450 vector signed long long vec_cnttz (vector signed long long);
17451 vector unsigned long long vec_cnttz (vector unsigned long long);
17453 2017-07-25 Andrew Pinski <apinski@cavium.com>
17455 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17456 accesses where the use is for the first operand of a BIT_INSERT.
17458 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17461 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17462 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17464 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17466 * config/i386/gstabs.h: Delete.
17467 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17469 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
17471 * config/i386/i386.c (ix86_decompose_address): Do not check for
17472 register RTX when looking at index_reg or base_reg.
17473 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
17475 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
17477 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
17478 to update EH info here.
17480 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17482 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
17484 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17486 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
17488 2017-07-25 Torsten Duwe <duwe@suse.de>
17490 * common.opt: Introduce -fpatchable-function-entry
17491 command line option, and its variables function_entry_patch_area_size
17492 and function_entry_patch_area_start.
17493 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17494 including a two-value parser.
17495 * target.def (print_patchable_function_entry): New target hook.
17496 * targhooks.h (default_print_patchable_function_entry): New function.
17497 * targhooks.c (default_print_patchable_function_entry): Likewise.
17498 * toplev.c (process_options): Switch off IPA-RA if
17499 patchable function entries are being generated.
17500 * varasm.c (assemble_start_function): Look at the
17501 patchable-function-entry command line switch and current
17502 function attributes and maybe generate NOP instructions by
17503 calling the print_patchable_function_entry hook.
17504 * doc/extend.texi: Document patchable_function_entry attribute.
17505 * doc/invoke.texi: Document -fpatchable_function_entry
17506 command line option.
17507 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17509 * doc/tm.texi: Re-generate.
17511 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17514 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17515 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17517 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17519 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17521 2017-07-25 Richard Biener <rguenther@suse.de>
17523 PR tree-optimization/81455
17524 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17525 not walk in cycles when looking for guards.
17527 2017-07-25 Richard Biener <rguenther@suse.de>
17529 PR tree-optimization/81529
17530 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17531 when optimizing backedge uses.
17533 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17535 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17537 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17538 to dl_section_ref. On AIX, append an expression to subtract
17539 the size of the section length to dl_section_ref.
17541 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17543 * configure.ac: If any of the config.* scripts fail, exit 1.
17544 * configure: Regenerate.
17546 2017-07-25 Richard Biener <rguenther@suse.de>
17548 PR middle-end/81546
17549 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17550 of immediate uses, be more verbose on errors.
17552 2017-07-25 Richard Biener <rguenther@suse.de>
17554 PR tree-optimization/81510
17555 * tree-vect-loop.c (vect_is_simple_reduction): When the
17556 reduction stmt is not inside the loop bail out.
17558 2017-07-25 Richard Biener <rguenther@suse.de>
17560 PR tree-optimization/81303
17561 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17562 profitability check against LOOP_VINFO_NITERSM1.
17564 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17566 * domwalk.c (cmp_bb_postorder): Simplify.
17567 (sort_bbs_postorder): New function. Use it...
17568 (dom_walker::walk): ...here to optimize common cases.
17570 2017-07-25 Martin Liska <mliska@suse.cz>
17573 * ipa-visibility.c (function_and_variable_visibility): Make the
17574 redirection just on target that supports aliasing.
17575 Fix GNU coding style.
17577 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17580 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17582 * config/arm/rtems.h: Likewise.
17583 * config/bfin/rtems.h: Likewise.
17584 * config/i386/rtemself.h: Likewise.
17585 * config/lm32/rtems.h: Likewise.
17586 * config/m32c/rtems.h: Likewise.
17587 * config/m68k/rtemself.h: Likewise.
17588 * config/microblaze/rtems.h: Likewise.
17589 * config/mips/rtems.h: Likewise.
17590 * config/moxie/rtems.h: Likewise.
17591 * config/nios2/rtems.h: Likewise.
17592 * config/powerpcspe/rtems.h: Likewise.
17593 * config/rs6000/rtems.h: Likewise.
17594 * config/rtems.h: Likewise.
17595 * config/sh/rtems.h: Likewise.
17596 * config/sh/rtemself.h: Likewise.
17597 * config/sparc/rtemself.h: Likewise.
17599 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17602 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17603 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17604 * tree-ssa-structalias.c (alias_get_name): Same.
17606 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17609 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17610 instructions if no du chain is found.
17612 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17614 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17616 2017-07-25 Richard Biener <rguenther@suse.de>
17618 PR middle-end/81505
17619 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17622 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17624 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17625 upper-regs options.
17626 (ISA_2_7_MASKS_SERVER): Likewise.
17627 (ISA_3_0_MASKS_IEEE): Likewise.
17628 (OTHER_P8_VECTOR_MASKS): Likewise.
17629 (OTHER_VSX_VECTOR_MASKS): Likewise.
17630 (POWERPC_MASKS): Likewise.
17631 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17632 duplicate list of options.
17633 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17634 explicit -mupper-regs options.
17635 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17636 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17637 alias for -mupper-regs-df.
17638 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17639 (rs6000_init_hard_regno_mode_ok): Likewise.
17640 (rs6000_option_override_internal): Likewise.
17641 (rs6000_opt_masks): Likewise.
17642 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17643 options in terms of whether -mvsx or -mpower8-vector was used.
17644 (TARGET_UPPER_REGS_DI): Likewise.
17645 (TARGET_UPPER_REGS_SF): Likewise.
17646 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17647 -mupper-regs-* options.
17649 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17651 * passes.c (emergency_dump_function): Print some empty lines and a
17652 header before the RTL dump.
17654 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17656 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17658 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17661 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17662 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17664 2017-07-24 Carl Love <cel@us.ibm.com>
17666 * config/rs6000/rs6000-c.c: Add support for built-in functions
17667 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17668 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17669 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17670 vec_extract_fp_from_shortl): Add defines for the two builtins.
17671 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17672 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17674 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17675 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17676 * doc/extend.texi: Update the built-in documentation file for the
17677 new built-in function.
17679 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17682 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17684 * doc/generic.texi: Likewise.
17685 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17686 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17688 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17690 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17691 (aarch64_mls_elt_merge<mode>): Likewise.
17693 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17695 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17696 having __cxa_atexit.
17698 2017-07-23 Michael Collison <michael.collison@arm.com>
17700 * config/arm/arm.c (arm_option_override): Deprecate
17701 use of -mstructure-size-boundary.
17702 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17703 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17705 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17708 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17709 Reduce cost estimate for direct moves.
17711 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17714 * config/i386/i386.c (ix86_option_override_internal): Disable
17715 BMI, BMI2 and TBM instructions for -m16.
17717 2017-07-21 Carl Love <cel@us.ibm.com>
17719 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17720 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17721 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17722 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17723 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17724 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17725 VMULOSW): New enum "unspec" values.
17726 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17727 altivec_vmulosw): New patterns.
17728 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17729 VMULOSW): Add definitions.
17731 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17733 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17734 (qdf24xx): Likewise.
17735 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17736 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17737 (AARCH64_FL_V8_1): Renumber.
17738 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17739 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17740 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17741 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17742 rdma to feature modifiers list.
17744 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17746 PR middle-end/56727
17747 * ipa-visibility (function_and_variable_visibility): Convert
17748 recursive PLT call to direct call if appropriate.
17750 2017-07-21 Andrew Pinski <apinski@cavium.com>
17752 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17753 operand 1 to see if the types precision matches.
17754 * fold-const.c (operand_equal_p): Likewise.
17756 2017-07-21 Richard Biener <rguenther@suse.de>
17758 PR tree-optimization/81303
17759 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17760 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17761 (vect_peeling_hash_get_lowest_cost): Adjust.
17762 (vect_enhance_data_refs_alignment): Likewise. Use
17763 vect_get_peeling_costs_all_drs to compute the penalty for no
17764 peeling to match up costs.
17766 2017-07-21 Richard Biener <rguenther@suse.de>
17768 PR tree-optimization/81500
17769 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17770 we didn't identify a reduction path.
17772 2017-07-21 Tom de Vries <tom@codesourcery.com>
17773 Cesar Philippidis <cesar@codesourcery.com>
17775 PR gcov-profile/81442
17776 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17779 2017-07-21 Tom de Vries <tom@codesourcery.com>
17782 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17784 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17785 nvptx_override_options_after_change.
17787 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17789 * dwarf2out.c (output_file_names): Avoid double testing for
17790 dwarf_version >= 5.
17792 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17794 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17796 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17798 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17800 (try_crossjump_to_edge): Do not punt on partitioned functions.
17802 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17804 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17805 Put all BBs reachable only via paths crossing cold region to cold
17807 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17809 2016-07-21 Richard Biener <rguenther@suse.de>
17811 PR tree-optimization/81303
17812 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17813 into account prologue and epilogue iterations when raising
17814 min_profitable_iters to sth at least covering one vector iteration.
17816 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17818 * config/arm/arm.c (arm_test_cpu_arch_dat):
17821 2017-07-20 Nathan Sidwell <nathan@acm.org>
17823 Remove TYPE_METHODS.
17824 * tree.h (TYPE_METHODS): Delete.
17825 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17826 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17827 (dbxout_type_methods): Scan TYPE_FIELDS.
17828 (dbxout_type): Don't check TYPE_METHODS here.
17829 * function.c (use_register_for_decl): Always ignore register for
17830 class types when not optimizing.
17831 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17832 * tree.c (free_lang_data_in_type): Stitch out member functions and
17833 templates from TYPE_FIELDS.
17834 (build_distinct_type_copy, verify_type_variant,
17835 verify_type): Member fns are on TYPE_FIELDS.
17836 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17837 * tree-pretty-print.c (dump_generic_node): Likewise.
17839 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17842 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17843 V2TImode and V4TImode.
17844 (ix86_expand_vector_extract): Likewise.
17845 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17846 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17847 (ssescalarmode): Handle V4TImode and V2TImode.
17848 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17849 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17850 (VEXTRACTI128_MODE): New mode iterator.
17851 (splitter for *vec_extractv?ti first element): New.
17852 (VEC_INIT_MODE): New mode iterator.
17853 (vec_init<mode>): Consolidate 3 expanders into one using
17854 VEC_INIT_MODE mode iterator.
17856 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17858 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17859 non_spilled_static_chain_regno_p.
17861 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17863 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17865 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17867 * bb-reorder.c (connect_traces): Allow copying of blocks within
17870 2017-07-20 Richard Biener <rguenther@suse.de>
17872 * gimple.h (gimple_phi_result): Add gphi * overload.
17873 (gimple_phi_result_ptr): Likewise.
17874 (gimple_phi_arg): Likewise. Adjust index assert to only
17875 allow actual argument accesses rather than all slots available
17877 (gimple_phi_arg_def): Add gphi * overload.
17878 * tree-phinodes.c (make_phi_node): Initialize only actual
17880 (resize_phi_node): Clear memory not covered by old node,
17881 do not initialize excess argument slots.
17882 (reserve_phi_args_for_new_edge): Initialize new argument slot
17885 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17887 PR tree-optimization/81388
17889 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17891 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17892 by removing computation of may_be_zero.
17894 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17895 Tom de Vries <tom@codesourcery.com>
17897 PR middle-end/81030
17898 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17899 when gimple level profile disagrees with what RTL expander did.
17901 2017-07-20 Richard Biener <rguenther@suse.de>
17903 PR tree-optimization/61171
17904 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17905 (vect_analyze_stmt): Add slp instance parameter.
17906 (vectorizable_reduction): Likewise.
17907 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17908 (vect_is_simple_reduction): Deal with chains not detected
17909 as SLP reduction chain, specifically not properly associated
17910 chains containing a mix of plus/minus.
17911 (get_reduction_op): Remove.
17912 (get_initial_defs_for_reduction): Simplify, pass in whether
17913 this is a reduction chain, pass in the SLP node for the PHIs.
17914 (vect_create_epilog_for_reduction): Get the SLP instance as
17916 (vectorizable_reduction): Get the SLP instance as arg.
17917 During analysis remember the SLP node with the PHIs in the
17918 instance. Simplify getting at the vectorized reduction PHIs.
17919 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17920 through SLP instance.
17921 (vect_slp_analyze_operations): Likewise.
17922 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17923 (vect_transform_stmt): Likewise.
17925 2017-07-20 Tom de Vries <tom@codesourcery.com>
17927 PR tree-optimization/81489
17928 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17929 read of phi arg location to before loop that modifies phi.
17931 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17933 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17936 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17938 PR middle-end/81331
17939 * except.c (execute): Fix ordering issue.
17941 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17943 PR rtl-optimization/81423
17944 * combine.c (make_compound_operation_int): Don't try to optimize
17945 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17947 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17949 PR rtl-optimization/81423
17950 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17951 with a constant that is -1 in the truncated to mode.
17953 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17955 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17956 (determine_unlikely_bbs): ... here.
17957 * predict.h (propagate_unlikely_bbs_forward): Declare.
17958 * cfgexpand.c (pass_expand::execute): Use it.
17959 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17961 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17962 propagate_unlikely_bbs_forward.
17964 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17966 PR middle-end/81331
17967 * except.c (maybe_add_nop_after_section_switch): New function.
17970 2017-07-19 Tom de Vries <tom@codesourcery.com>
17972 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17974 2017-07-19 Tom de Vries <tom@codesourcery.com>
17976 * gimple.h (gimple_phi_arg): Make assert more strict.
17978 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17980 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17981 * config/rs6000/mmintrin.h: New file.
17982 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17984 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17986 PR tree-optimization/81346
17987 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17989 2017-07-19 Tom de Vries <tom@codesourcery.com>
17991 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17993 2017-07-19 Tom de Vries <tom@codesourcery.com>
17995 * config/nvptx/nvptx-modes.def: Add V2DImode.
17996 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17997 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17998 (nvptx_output_mov_insn): Handle lack of mov.b128.
17999 (nvptx_print_operand): Handle 'H' and 'L' codes.
18000 (nvptx_vector_mode_supported): Allow V2DImode.
18001 (nvptx_preferred_simd_mode): New function.
18002 (nvptx_data_alignment): New function.
18003 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18004 nvptx_preferred_simd_mode.
18005 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18007 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18009 2017-07-19 Tom de Vries <tom@codesourcery.com>
18011 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18012 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18013 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18014 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18015 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18016 (mov<VECIM>_insn): New define_insn.
18017 (define_expand "mov<VECIM>): New define_expand.
18019 2017-07-19 Tom de Vries <tom@codesourcery.com>
18021 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18023 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18025 PR tree-optimization/81346
18026 * fold-const.h (fold_div_compare, range_check_type): Declare.
18027 * fold-const.c (range_check_type): New function.
18028 (build_range_check): Use range_check_type.
18029 (fold_div_compare): No longer static, rewritten into
18030 a match.pd helper function.
18031 (fold_comparison): Don't call fold_div_compare here.
18032 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18033 as helper function.
18035 2017-07-19 Nathan Sidwell <nathan@acm.org>
18037 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18038 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18039 * tree.c (find_decls_types_r, verify_type): Use
18040 TYPE_{MIN,MAX}_VALUE_RAW.
18041 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18042 (hash_tree): Likewise.
18043 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18045 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18048 2017-07-18 Tom de Vries <tom@codesourcery.com>
18050 PR middle-end/81464
18051 * omp-expand.c (expand_omp_for_static_chunk): Handle
18052 equal-argument loop exit phi.
18054 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18057 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18058 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18059 operand 2 predicate.
18060 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18061 operand 2 predicate.
18062 (ror,rol -> rorx splitters): Use const_int_operand as
18063 operand 2 predicate.
18065 2017-06-18 Richard Biener <rguenther@suse.de>
18067 PR tree-optimization/81410
18068 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18069 the gap in the ! slp_perm SLP case after each group.
18071 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18073 PR middle-end/81463
18074 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18077 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18079 PR middle-end/81462
18080 * predict.c (set_even_probabilities): Cleanup; do not affect
18081 probabilities that are already known.
18082 (combine_predictions_for_bb): Call even when count is set.
18084 2017-07-18 Nathan Sidwell <nathan@acm.org>
18086 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18089 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18092 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18093 optimization for loop niter analysis.
18095 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18098 * config/avr/avr.c (avr_optimize_casesi): Don't use
18099 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18101 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18103 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18104 body_cost_vec from _vect_peel_extended_info.
18105 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18106 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18109 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18111 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18113 2017-07-18 Richard Biener <rguenther@suse.de>
18115 PR tree-optimization/80620
18116 PR tree-optimization/81403
18117 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18118 info when re-using a VN table entry.
18120 2017-07-18 Richard Biener <rguenther@suse.de>
18122 PR tree-optimization/81418
18123 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18124 vectype_in. Verify that with lane-reducing reduction operations
18125 we have a single def-use cycle.
18127 2017-07-17 Carl Love <cel@us.ibm.com>
18129 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18131 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18132 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18133 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18135 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18136 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18137 VMULOSW): New enum "unspec" values.
18138 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18139 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18140 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18141 altivec_vmulosw): New patterns.
18142 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18143 VMULOSW): Add definitions.
18144 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18146 * config/alpha/alpha.c: Include predict.h.
18148 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18150 * tree-vrp.c (compare_assert_loc): Fix comparison function
18151 to return predictable results.
18153 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18155 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18157 (subdi3): Likewise.
18158 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18159 * doc/invoke.texi (mexpand-adddi): Update text.
18161 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18163 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18164 that also clobbers the CC register. The old expand code is moved
18166 (*arc_clzsi2): ... here.
18167 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18168 the CC register. The old expand code is moved to ...
18169 (arc_ctzsi2): ... here.
18171 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18173 * config/arc/arc.opt (mindexed-loads): Use initial value
18174 TARGET_INDEXED_LOADS_DEFAULT.
18175 (mauto-modify-reg): Use initial value
18176 TARGET_AUTO_MODIFY_REG_DEFAULT.
18177 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18178 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18179 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18180 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18182 2017-07-17 Martin Liska <mliska@suse.cz>
18185 * opts.c (finish_options): Do not allow -fgnu-tm
18186 w/ -fsanitize={kernel-,}address. Say sorry.
18188 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18191 * tree-loop-distribution.c (classify_partition): Only assert on
18192 numer of iterations.
18193 (merge_dep_scc_partitions): Delete prameter. Update function call.
18194 (distribute_loop): Remove code handling loop with unknown niters.
18195 (pass_loop_distribution::execute): Skip loop with unknown niters.
18197 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18200 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18201 function sort_partitions_by_post_order.
18203 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18205 PR tree-optimization/81374
18206 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18207 the max index of basic blocks, rather than number of basic blocks.
18209 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18211 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18213 (arc_legitimate_pic_operand_p): Likewise.
18214 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18216 (arc_needs_pcl_p): Likewise.
18217 (arc_legitimate_pc_offset_p): Likewise.
18218 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18219 function is also used in constrains.md.
18220 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18221 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18222 PLUS. Only return true/false in known cases, otherwise assert.
18223 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18224 is already called in arc_legitimate_constant_p.
18225 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18227 (LEGITIMATE_PIC_OPERAND_P): Use
18228 arc_raw_symbolic_reference_mentioned_p function.
18229 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18234 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18235 Andrew Burgess <andrew.burgess@embecosm.com>
18237 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18238 (arc_return_address_register): New function.
18239 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18240 (arc_handle_fndecl_attribute): Add naked attribute.
18241 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18242 (TARGET_WARN_FUNC_RETURN): Likewise.
18243 (arc_allocate_stack_slots_for_args): New function.
18244 (arc_warn_func_return): Likewise.
18245 (machine_function): Change type fn_type.
18246 (arc_compute_function_type): Consider new naked function type,
18247 change function return type.
18248 (arc_must_save_register): Adapt to handle new
18249 arc_compute_function_type's return type.
18250 (arc_expand_prologue): Likewise.
18251 (arc_expand_epilogue): Likewise.
18252 (arc_return_address_regs): Delete.
18253 (arc_return_address_register): New function.
18254 (arc_epilogue_uses): Use above function.
18255 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18256 (arc_function_type): Change encoding, add naked type.
18257 (ARC_INTERRUPT_P): Change to handle the new encoding.
18258 (ARC_FAST_INTERRUPT_P): Likewise.
18259 (ARC_NORMAL_P): Define.
18260 (ARC_NAKED_P): Likewise.
18261 (arc_compute_function_type): Delete prototype.
18262 * config/arc/arc.md (in_ret_delay_slot): Use
18263 arc_return_address_register function.
18264 (simple_return): Likewise.
18265 (p_return_i): Likewise.
18267 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18269 PR tree-optimization/81428
18270 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18271 can't be built for those types.
18273 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18275 Remove stuff dead since r239246.
18277 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18278 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18279 (avr_inform_devices): Remove dead stuff.
18281 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18283 * config/arm/arm_neon.h: Fix softp typo.
18285 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18287 PR tree-optimization/81365
18288 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18289 aggregate moves onto bb predecessor edges, make sure there are no
18290 loads that could alias the lhs in between the start of bb and the
18293 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18296 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18297 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18298 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18300 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18302 PR tree-optimization/81396
18303 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18304 (init_symbolic_number): Initialize it to 1.
18305 (perform_symbolic_merge): Add n_ops from both operands into the new
18307 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18308 without base_addr as useless if they need more than one operation.
18309 (bswap_replace): Handle !bswap case for NULL base_addr.
18311 2017-07-17 Tom de Vries <tom@codesourcery.com>
18314 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18317 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18319 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18320 conditional builtin define __FIX_LEON3FT_B2BST.
18322 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18324 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18325 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18328 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18330 PR rtl-optimization/81424
18331 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18332 to remove potential trapping from operands if -fnon-call-exceptions.
18334 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18336 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18337 profile_proability for scalling.
18338 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18340 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18342 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18344 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18346 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18347 fixpoint arithmetics.
18349 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18351 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18352 fixpoint arithmetics.
18354 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18356 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18357 fixpoint arithmetics.
18359 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18361 * profile-count.h (profile_probability::from_reg_br_prob_note,
18362 profile_probability::to_reg_br_prob_note): New functions.
18363 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18364 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18365 * predict.c (probability_reliable_p): Update.
18366 (edge_probability_reliable_p): Update.
18367 (br_prob_note_reliable_p): Update.
18368 (invert_br_probabilities): Update.
18369 (add_reg_br_prob_note): New function.
18370 (combine_predictions_for_insn): Update.
18371 * asan.c (asan_clear_shadow): Update.
18372 * cfgbuild.c (compute_outgoing_frequencies): Update.
18373 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18374 (update_br_prob_note): Update.
18375 (rtl_verify_edges): Update.
18376 (purge_dead_edges): Update.
18377 (fixup_reorder_chain): Update.
18378 * emit-rtl.c (try_split): Update.
18379 * ifcvt.c (cond_exec_process_insns): Update.
18380 (cond_exec_process_if_block): Update.
18381 (dead_or_predicable): Update.
18382 * internal-fn.c (expand_addsub_overflow): Update.
18383 (expand_neg_overflow): Update.
18384 (expand_mul_overflow): Update.
18385 * loop-doloop.c (doloop_modify): Update.
18386 * loop-unroll.c (compare_and_jump_seq): Update.
18387 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18388 * predict.h: Update.
18389 * reorg.c (mostly_true_jump): Update.
18391 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18392 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18393 * config/arc/arc.c: (emit_unlikely_jump): Update.
18394 * config/arm/arm.c: (emit_unlikely_jump): Update.
18395 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18396 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18397 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18398 (ix86_print_operand): Update.
18399 (ix86_split_fp_branch): Update.
18400 (predict_jump): Update.
18401 * config/ia64/ia64.c (ia64_print_operand): Update.
18402 * config/mmix/mmix.c (mmix_print_operand): Update.
18403 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18404 (rs6000_expand_split_stack_prologue): Update.
18405 * config/rs6000/rs6000.c: Update.
18406 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18407 (s390_expand_vec_movstr): Update.
18408 (s390_expand_cs_tdsi): Update.
18409 (s390_expand_split_stack_prologue): Update.
18410 * config/sh/sh.c (sh_print_operand): Update.
18411 (expand_cbranchsi4): Update.
18412 (expand_cbranchdi4): Update.
18413 * config/sparc/sparc.c (output_v9branch): Update.
18414 * config/spu/spu.c (get_branch_target): Update.
18415 (ea_load_store_inline): Update.
18416 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18417 * config/tilepro/tilepro.c: Update.
18419 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18421 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18422 (gimplify_save_expr): Likewise.
18424 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18426 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18428 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18430 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18432 * timevar.def (TV_IPA_FNSUMMARY): Define.
18434 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18436 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18437 to back store errata sensitive sequence from being generated.
18438 (sqrtdf2_fix): Likewise.
18440 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18442 * tree-ssa-threadupdate.c (compute_path_counts,
18443 update_joiner_offpath_counts): Use profile_probability.
18445 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18448 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18450 * config/arm/arm-c.c (arm_cpu_builtins): Define
18451 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18453 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18455 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18456 array entries to represent __ieee128 versions of the
18457 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
18458 scalar_extract_sig, and scalar_insert_exp built-in functions.
18459 (altivec_resolve_overloaded_builtin): Add special case handling
18460 for the __builtin_scalar_insert_exp function, as represented by
18461 the P9V_BUILTIN_VEC_VSIEDP constant.
18462 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
18463 exponent support for __ieee128 argument.
18464 (VSESQP): Add scalar extract signature support for __ieee128
18466 (VSTDCNQP): Add scalar test negative support for __ieee128
18468 (VSIEQP): Add scalar insert exponent support for __int128 argument
18469 with __ieee128 result.
18470 (VSIEQPF): Add scalar insert exponent support for __ieee128
18471 argument with __ieee128 result.
18472 (VSTDCQP): Add scalar test data class support for __ieee128
18474 (VSTDCNQP): Add overload support for scalar test negative with
18475 __ieee128 argument.
18476 (VSTDCQP): Add overload support for scalar test data class
18477 __ieee128 argument.
18478 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
18479 UNSPEC_VSX_SXSIGDP.
18480 (UNSPEC_VSX_SIEXPQP): New constant.
18481 (xsxexpqp): New insn for VSX scalar extract exponent quad
18483 (xsxsigqp): New insn for VSX scalar extract significand quad
18485 (xsiexpqpf): New insn for VSX scalar insert exponent quad
18486 precision with floating point argument.
18487 (xststdcqp): New expand for VSX scalar test data class quad
18489 (xststdcnegqp): New expand for VSX scalar test negative quad
18491 (xststdcqp): New insn to match expansions for VSX scalar test data
18492 class quad precision and VSX scalar test negative quad precision.
18493 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18494 special case operand checking to enforce that second operand of
18495 VSX scalar test data class with quad precision argument is a 7-bit
18497 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18498 prototypes and descriptions of __ieee128 versions of
18499 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18500 scalar_test_data_class, and scalar_test_neg built-in functions.
18502 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18504 PR tree-optimization/81162
18505 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18506 replace a negate with an add.
18508 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18510 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18512 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18514 * config/arm/arm-c.c (arm_cpu_builtins): Define
18515 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18517 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18519 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18520 (armv8-r): Set ARM Cortex-R52 as default CPU.
18521 * config/arm/arm-tables.opt: Regenerate.
18522 * config/arm/arm-tune.md: Regenerate.
18523 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18525 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18526 extension for -mcpu=cortex-r52.
18528 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18530 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18531 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18532 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18533 (fp-armv8): Define it as FP_ARMv8 only.
18534 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18535 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18537 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18538 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18539 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18540 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18541 than TARGET_FPU_ARMV8.
18542 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18543 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18544 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18546 * config/arm/neon.md (neon_vrint): Likewise.
18547 (neon_vcvt): Likewise.
18548 (neon_<fmaxmin_op><mode>): Likewise.
18549 (<fmaxmin><mode>3): Likewise.
18550 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18551 * config/arm/predicates.md (arm_cond_move_operator): Check against
18552 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18554 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18556 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18557 to top of function.
18559 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18561 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18562 loop in comment with memset.
18564 2017-07-14 Martin Liska <mliska@suse.cz>
18566 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18567 * dwarf2out.c (is_java): Remove the function.
18568 (output_pubname): Remove usage of the function.
18569 (lower_bound_default): Remove usage of DW_LANG_Java.
18570 (gen_compile_unit_die): Likewise.
18571 * gcc.c: Remove compiler defaults for .java and .zip files.
18572 * gimple-expr.c (remove_suffix): Change as there's no longer
18573 extension than 4-letter one.
18574 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18575 (gimplify_save_expr): Likewise.
18576 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18577 as it's possible even for other languages than Java.
18578 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18579 * lto-opts.c (lto_write_options): Remove reference to Java.
18580 * opts.c (strip_off_ending): Update file extension handling.
18581 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18582 * tree-eh.c (lower_resx): Likewise.
18583 * tree.c (free_lang_data_in_type): Remove dead code.
18584 (find_decls_types_r): Likewise.
18585 (build_common_builtin_nodes): Remove Java from a comment.
18586 (verify_type): Remove dead code.
18587 * varasm.c (assemble_external): Remove Java from a comment.
18589 2017-07-14 Martin Liska <mliska@suse.cz>
18591 * opts.c (finish_options): Add quotes.
18592 (common_handle_option): Likewise.
18594 2017-07-14 Martin Liska <mliska@suse.cz>
18596 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18597 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18598 Remove N_SO_PASCAL.
18599 * dwarf2out.c (lower_bound_default): Do not handle
18601 (gen_compile_unit_die): Likewise.
18602 * gcc.c: Remove default extension binding for GNU Pascal.
18603 * stmt.c: Remove Pascal language from a comment.
18604 * xcoffout.c: Likewise.
18606 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18609 * diagnostic-show-locus.c (fixit_cmp): New function.
18610 (layout::layout): Sort m_fixit_hints.
18611 (column_range::column_range): Assert that the values are valid.
18612 (struct char_span): New struct.
18613 (correction::overwrite): New method.
18614 (struct source_line): New struct.
18615 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18616 calls in terms of classes source_line and char_span, and
18617 correction::overwrite.
18618 (selftest::test_overlapped_fixit_printing_2): New function.
18619 (selftest::diagnostic_show_locus_c_tests): Call it.
18621 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18624 early if there is no lhs.
18626 2017-07-13 Martin Liska <mliska@suse.cz>
18628 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18629 (gen_reference_type_die): Likewise.
18630 * stor-layout.c: Remove Pascal-related comment.
18632 2017-07-13 Martin Liska <mliska@suse.cz>
18634 * opts.c (finish_options): Add quotes to error messages.
18635 (parse_sanitizer_options): Likewise.
18637 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18639 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18641 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18643 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18645 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18647 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18649 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18651 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18654 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18655 provides the hardware capability bits, define the macro
18656 __BUILTIN_CPU_SUPPORTS__.
18657 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18658 if GLIBC does not provide the hardware capability bits. Add a
18659 gcc_unreachable call if the built-in cpu function is neither
18660 __builtin_cpu_is nor __builtin_cpu_supports.
18661 (rs6000_get_function_versions_dispatcher): Change the warning
18662 that an old GLIBC is used which does not export the capability
18663 bits to be an error.
18664 * doc/extend.texi (target_clones attribute): Document the
18665 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18666 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18667 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18668 the macros defined by GCC if the newer GLIBC is available.
18670 2017-07-12 Jeff Law <law@redhat.com>
18672 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18673 remaining includes slightly.
18674 * config/riscv/riscv-builtins.c: Include profile-count.h.
18676 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18679 * config/avr/avr.c (avr_set_current_function): In diagnostic
18680 messages: Quote keywords and (parts of) identifiers.
18681 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18684 2017-07-12 Carl Love <cel@us.ibm.com>
18686 * config/rs6000/rs6000-c.c: Add support for built-in functions
18687 vector bool char vec_revb (vector bool char);
18688 vector bool short vec_revb (vector short char);
18689 vector bool int vec_revb (vector bool int);
18690 vector bool long long vec_revb (vector bool long long);
18691 * doc/extend.texi: Update the built-in documentation file for the
18692 new built-in functions.
18694 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18696 * config/s390/s390.md: Remove movcc splitter.
18698 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18700 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18701 load/store on condition.
18703 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18706 * config/avr/avr.c (avr_encode_section_info)
18707 [progmem && !TREE_READONLY]: Error if progmem object needs
18710 2017-07-11 Michael Collison <michael.collison@arm.com>
18712 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18715 2017-07-11 Carl Love <cel@us.ibm.com>
18717 * config/rs6000/rs6000-c.c: Add support for builtins
18718 vector unsigned int vec_parity_lsbb (vector signed int);
18719 vector unsigned int vec_parity_lsbb (vector unsigned int);
18720 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18721 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18722 vector unsigned long long vec_parity_lsbb (vector signed long long);
18723 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18724 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18725 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18726 * doc/extend.texi: Update the built-in documentation file for the
18727 new built-in functions.
18729 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18731 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18732 (layout::m_primary_loc): New field.
18733 (layout::layout): Initialize new field. Move location filtering
18734 logic from here to...
18735 (layout::maybe_add_location_range): ...this new method. Add
18736 support for filtering to just the lines already specified by other
18738 (layout::will_show_line_p): New method.
18739 (gcc_rich_location::add_location_if_nearby): New method.
18740 (selftest::test_add_location_if_nearby): New test function.
18741 (selftest::diagnostic_show_locus_c_tests): Call it.
18742 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18745 2017-07-11 Tom de Vries <tom@codesourcery.com>
18747 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18748 (bb_first_real_insn): New function.
18749 (nvptx_single): Add extra initialization of broadcasted condition
18752 2017-07-11 Nathan Sidwell <nathan@acm.org>
18754 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18756 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18758 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18759 Binutils doc as TEXI will mess them up.
18760 * doc/invoke.texi (AVR Options): Same here.
18762 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18764 * config/sparc/sparc.opt (mfix-ut700): New option.
18765 (mfix-gr712rc): Likewise.
18766 (sparc_fix_b2bst): New variable.
18767 * doc/invoke.texi (SPARC options): Document them.
18768 (ARM options): Fix warnings.
18769 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18770 instructions to prevent sequences that can trigger the store-store
18771 errata for certain LEON3FT processors.
18772 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18773 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18774 * config/sparc/sparc.md (fix_b2bst): New attribute.
18775 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18777 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18780 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18782 (*rsqrtsf2_sse): Ditto.
18784 (div<mode>3): Macroize insn from divdf3 and divsf3
18785 using MODEF mode iterator.
18787 2017-07-10 Martin Sebor <msebor@redhat.com>
18789 PR tree-optimization/80397
18790 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18791 instead of testing for equality to INTEGER_TYPE.
18793 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18795 * config.gcc: Remove uclibc from arc target spec.
18797 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18799 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18801 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18804 * lto-wrapper.c (remove_option): New function.
18805 (merge_and_complain): Merge PIC/PIE options more realistically.
18807 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18809 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18813 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18814 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18815 * config.in: Regenerate.
18816 * configure: Regenerate.
18817 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18818 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18819 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18820 (TARGET_GASISR_PROLOGUES): ...target mask.
18821 * common/config/avr/avr-common.c
18822 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18823 Set -mgas-isr-prologues.
18824 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18825 INSERT_PASS_BEFORE for it.
18826 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18827 * config/avr/avr.c (avr_option_override)
18828 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18829 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18830 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18831 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18832 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18834 (make_avr_pass_pre_proep): New function.
18835 (emit_push_sfr) <treg>: Add argument to function and use it
18836 instead of TMP_REG.
18837 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18838 and set machine->gasisr.yes.
18839 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18840 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18841 __gcc_isr.n_pushed to .L__stack_usage.
18842 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18843 (avr_asm_final_postscan_insn): ...this new static function.
18844 * config/avr/avr.h (machine_function)
18845 <is_no_gccisr, use_L__stack_usage>: New fields.
18846 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18847 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18848 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18849 (gasisr, *gasisr): New expander and insn.
18850 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18851 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18852 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18854 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18856 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18859 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18861 Move jump-tables out of .text again.
18864 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18865 (ASM_OUTPUT_ADDR_VEC): New function.
18866 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18867 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18868 INSN_ADDRESSes as asm comment.
18869 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18870 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18871 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18872 * config/avr/avr.md (*tablejump): Adjust comment.
18873 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18874 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18876 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18877 (avr_output_addr_vec): New proto.
18878 (avr_log_t) <insn_addresses>: New field.
18880 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18883 * config/i386/i386.c (ix86_function_arg_advance): Set
18884 outgoing_args_on_stack to true if there are outgoing arguments
18886 (ix86_function_arg): Likewise.
18887 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18888 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18889 * config/i386/i386.h (machine_function): Add
18890 outgoing_args_on_stack.
18892 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18894 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18895 supporting pthreds.
18896 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18898 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18900 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18901 (REAL_H): Remove $(MACHMODE_H).
18902 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18904 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18905 $(MACHMODE_H) and double-int.h.
18906 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18908 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18911 2017-07-07 Andrew Pinski <apinski@cavium.com>
18913 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18914 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18916 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18918 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18919 Add warning if GCC was not configured to link against a GLIBC that
18920 exports the hardware capability bits.
18921 (make_resolver_func): Make resolver function private and not a
18922 COMDAT function. Create the name with clone_function_name instead
18923 of make_unique_name.
18926 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18927 correct operand in doing the split.
18929 2017-07-07 Carl Love <cel@us.ibm.com>
18931 * config/rs6000/rs6000-c: Add support for built-in function
18932 vector unsigned short vec_pack_to_short_fp32 (vector float,
18934 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18935 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18936 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18937 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18938 (convert_4f32_8i16): Add define_expand.
18939 * doc/extend.texi: Update the built-in documentation file for the
18940 new built-in function.
18942 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18944 * config/sparc/m8.md: New file.
18945 * config/sparc/sparc.md: Include m8.md.
18947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18949 * config/sparc/sparc.opt: New option -mvis4b.
18950 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18951 (sparc_option_override): Handle VIS4B.
18952 (enum sparc_builtins): Define
18953 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18954 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18955 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18956 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18957 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18958 (check_constant_argument): New function.
18959 (sparc_vis_init_builtins): Define builtins
18960 __builtin_vis_dictunpack{8,16,32},
18961 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18962 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18963 __builtin_vis_fpcmpde{8,16,32}shl and
18964 __builtin_vis_fpcmpur{8,16,32}shl.
18965 (sparc_expand_builtin): Check that the constant operands to
18966 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18967 constant and in range.
18968 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18970 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18971 (SPARC_IMM5_P): Likewise.
18972 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18973 (enabled): Handle vis4b.
18974 (UNSPEC_DICTUNPACK): New unspec.
18975 (UNSPEC_FPCMPSHL): Likewise.
18976 (UNSPEC_FPUCMPSHL): Likewise.
18977 (UNSPEC_FPCMPDESHL): Likewise.
18978 (UNSPEC_FPCMPURSHL): Likewise.
18979 (cpu_feature): New CPU feature `vis4b'.
18980 (dictunpack{8,16,32}): New insns.
18981 (FPCSMODE): New mode iterator.
18982 (fpcscond): New code iterator.
18983 (fpcsucond): Likewise.
18984 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18985 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18986 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18987 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18988 * config/sparc/constraints.md: Define constraints `q' for unsigned
18989 2-bit integer constants and `t' for unsigned 5-bit integer
18991 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18993 (imm5_operand_dictunpack16): Likewise.
18994 (imm5_operand_dictunpack32): Likewise.
18995 (imm2_operand): Likewise.
18996 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18997 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18998 ditunpack* and fpcmp*shl builtins.
19000 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19002 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19003 * config.in: Add HAVE_AS_SPARC6 define.
19004 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19006 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19008 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19009 (CPP_CPU_SPEC): Handle m8.
19010 (ASM_CPU_SPEC): Likewise.
19011 * config/sparc/sparc-opts.h (enum processor_type): Add
19013 * config/sparc/sparc.c (m8_costs): New struct.
19014 (sparc_option_override): Handle TARGET_CPU_m8.
19015 (sparc32_initialize_trampoline): Likewise.
19016 (sparc64_initialize_trampoline): Likewise.
19017 (sparc_issue_rate): Likewise.
19018 (sparc_register_move_cost): Likewise.
19019 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19020 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19021 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19022 (CPP_CPU_SPEC): Handle M8.
19023 (ASM_CPU_SPEC): Likewise.
19024 (AS_M8_FLAG): Define.
19025 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19026 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19027 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19029 * configure: Regenerate.
19030 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19033 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19035 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19037 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19038 ("*movdi_insn_sp32"): Do not set v3pipe.
19039 ("*movsi_insn"): Likewise.
19040 ("*movdi_insn_sp64"): Likewise.
19041 ("*movsf_insn"): Likewise.
19042 ("*movdf_insn_sp32"): Likewise.
19043 ("*movdf_insn_sp64"): Likewise.
19044 ("*zero_extendsidi2_insn_sp64"): Likewise.
19045 ("*sign_extendsidi2_insn"): Likewise.
19046 ("*mov<VM32:mode>_insn"): Likewise.
19047 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19048 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19049 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19050 ("<vlop:code><VL:mode>3"): Likewise.
19051 ("*not_<vlop:code><VL:mode>3"): Likewise.
19052 ("*nand<VL:mode>_vis"): Likewise.
19053 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19054 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19055 ("one_cmpl<VL:mode>2"): Likewise.
19056 ("faligndata<VM64:mode>_vis"): Likewise.
19057 ("alignaddrsi_vis"): Likewise.
19058 ("alignaddrdi_vis"): Likweise.
19059 ("alignaddrlsi_vis"): Likewise.
19060 ("alignaddrldi_vis"): Likewise.
19061 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19062 ("bmaskdi_vis"): Likewise.
19063 ("bmasksi_vis"): Likewise.
19064 ("bshuffle<VM64:mode>_vis"): Likewise.
19065 ("cmask8<P:mode>_vis"): Likewise.
19066 ("cmask16<P:mode>_vis"): Likewise.
19067 ("cmask32<P:mode>_vis"): Likewise.
19068 ("pdistn<P:mode>_vis"): Likewise.
19069 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19071 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19073 * config/sparc/sparc.md ("subtype"): New insn attribute.
19074 ("*wrgsr_sp64"): Set insn subtype.
19075 ("*rdgsr_sp64"): Likewise.
19076 ("alignaddrsi_vis"): Likewise.
19077 ("alignaddrdi_vis"): Likewise.
19078 ("alignaddrlsi_vis"): Likewise.
19079 ("alignaddrldi_vis"): Likewise.
19080 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19081 ("fexpand_vis"): Likewise.
19082 ("fpmerge_vis"): Likewise.
19083 ("faligndata<VM64:mode>_vis"): Likewise.
19084 ("bshuffle<VM64:mode>_vis"): Likewise.
19085 ("cmask8<P:mode>_vis"): Likewise.
19086 ("cmask16<P:mode>_vis"): Likewise.
19087 ("cmask32<P:mode>_vis"): Likewise.
19088 ("fchksm16_vis"): Likewise.
19089 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19090 ("fmean16_vis"): Likewise.
19091 ("fp<plusminus_insn>64_vis"): Likewise.
19092 ("<plusminus_insn>v8qi3"): Likewise.
19093 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19094 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19095 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19096 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19097 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19098 ("*movqi_insn"): Likewise.
19099 ("*movhi_insn"): Likewise.
19100 ("*movsi_insn"): Likewise.
19101 ("movsi_pic_gotdata_op"): Likewise.
19102 ("*movdi_insn_sp32"): Likewise.
19103 ("*movdi_insn_sp64"): Likewise.
19104 ("movdi_pic_gotdata_op"): Likewise.
19105 ("*movsf_insn"): Likewise.
19106 ("*movdf_insn_sp32"): Likewise.
19107 ("*movdf_insn_sp64"): Likewise.
19108 ("*zero_extendhisi2_insn"): Likewise.
19109 ("*zero_extendqihi2_insn"): Likewise.
19110 ("*zero_extendqisi2_insn"): Likewise.
19111 ("*zero_extendqidi2_insn"): Likewise.
19112 ("*zero_extendhidi2_insn"): Likewise.
19113 ("*zero_extendsidi2_insn_sp64"): Likewise.
19114 ("ldfsr"): Likewise.
19115 ("prefetch_64"): Likewise.
19116 ("prefetch_32"): Likewise.
19117 ("tie_ld32"): Likewise.
19118 ("tie_ld64"): Likewise.
19119 ("*tldo_ldub_sp32"): Likewise.
19120 ("*tldo_ldub1_sp32"): Likewise.
19121 ("*tldo_ldub2_sp32"): Likewise.
19122 ("*tldo_ldub_sp64"): Likewise.
19123 ("*tldo_ldub1_sp64"): Likewise.
19124 ("*tldo_ldub2_sp64"): Likewise.
19125 ("*tldo_ldub3_sp64"): Likewise.
19126 ("*tldo_lduh_sp32"): Likewise.
19127 ("*tldo_lduh1_sp32"): Likewise.
19128 ("*tldo_lduh_sp64"): Likewise.
19129 ("*tldo_lduh1_sp64"): Likewise.
19130 ("*tldo_lduh2_sp64"): Likewise.
19131 ("*tldo_lduw_sp32"): Likewise.
19132 ("*tldo_lduw_sp64"): Likewise.
19133 ("*tldo_lduw1_sp64"): Likewise.
19134 ("*tldo_ldx_sp64"): Likewise.
19135 ("*mov<VM32:mode>_insn"): Likewise.
19136 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19137 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19139 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19141 * config/sparc/sparc.md ("type"): New insn type viscmp.
19142 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19144 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19145 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19146 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19147 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19149 ("n7_vis_logical_11cycle"): Likewise.
19150 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19151 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19152 * config/sparc/niagara.md ("niag_vis"): Likewise.
19153 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19154 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19156 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19158 * config/sparc/sparc.md: New instruction type `bmask'.
19159 (bmaskdi_vis): Use the `bmask' type.
19160 (bmasksi_vis): Likewise.
19161 * config/sparc/ultra3.md (us3_array): Likewise.
19162 * config/sparc/niagara7.md (n7_array): Likewise.
19163 * config/sparc/niagara4.md (n4_array): Likewise.
19164 * config/sparc/niagara2.md (niag2_vis): Likewise.
19165 (niag3_vis): Likewise.
19166 * config/sparc/niagara.md (niag_vis): Likewise.
19168 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19170 * ipa-comdats.c: Remove optimize check from gate.
19171 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19172 for functions not optimized.
19173 (ipa_fn_summary_read): Skip optimize check.
19174 (ipa_fn_summary_write): Likewise.
19175 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19177 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19179 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19180 (check_callers): Give up on caller which is not optimized.
19181 (inline_small_functions): Likewise.
19182 (ipa_inline): Do not give up when not optimizing.
19183 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19184 away unoptimizes cdtors.
19185 (whole_program_function_and_variable_visibility): Do
19186 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19187 * ipa.c (process_references): Do not check optimize.
19188 (symbol_table::remove_unreachable_nodes): Update optimize check.
19189 (set_writeonly_bit): Update optimize check.
19190 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19191 (pass_ipa_single_use::gate): Remove.
19193 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19195 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19196 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19197 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19198 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19199 permute_load, permute_store, adjust_extract, adjust_splat,
19200 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19201 replace_swap_with_copy, dump_swap_insn_table,
19202 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19203 recombine_lvx_pattern, recombine_stvx_pattern,
19204 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19205 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19206 to file rs6000-p8swap.c.
19207 * config/rs6000/rs6000-p8swap.c: New file.
19208 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19209 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19210 and rs6000*-*-* targets.
19212 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19214 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19216 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19218 * lto-wrapper.c (merge_and_complain): Do not merge
19219 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19220 fsigned_zeros, ftrapping_math, fwrapv.
19221 (append_compiler_options): Do not track these options.
19222 (append_linker_options): Likewie
19224 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19226 * cgraphunit.c (cgraph_node::finalize_function): When
19227 !flag_toplevel_reorde set no_reorder flag.
19228 (varpool_node::finalize_decl): Likewise.
19229 (symbol_table::compile): Drop no toplevel reorder path.
19231 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19233 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19234 edges; zero probability is not better than uninitialized.
19236 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19238 * asan.h (asan_sanitize_allocas_p): Declare.
19239 * asan.c (asan_sanitize_allocas_p): New function.
19240 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19241 (handle_builtin_alloca): Likewise.
19242 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19243 if !asan_sanitize_allocas_p.
19244 * params.def (asan-instrument-allocas): Add new option.
19245 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19246 * opts.c (common_handle_option): Disable allocas sanitization for
19249 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19251 * asan.c: Include gimple-fold.h.
19252 (get_last_alloca_addr): New function.
19253 (handle_builtin_stackrestore): Likewise.
19254 (handle_builtin_alloca): Likewise.
19255 (asan_emit_allocas_unpoison): Likewise.
19256 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19257 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19258 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19259 (instrument_builtin_call): Pass gimple iterator to
19260 get_mem_refs_of_builtin_call.
19261 (last_alloca_addr): New global.
19262 * asan.h (asan_emit_allocas_unpoison): Declare.
19263 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19264 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19265 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19266 if function calls alloca.
19267 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19268 * gimple-fold.h (replace_call_with_value): Declare.
19269 * internal-fn.c: Include asan.h.
19270 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19271 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19273 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19275 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19276 (C_SELFTEST_FLAGS): New.
19277 (CPP_SELFTEST_FLAGS): New.
19278 (SELFTEST_DEPS): New, from deps of s-selftest.
19279 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19280 (CPP_SELFTEST_DEPS): New.
19281 (selftest): Add dependency on s-selftest-c++.
19282 (s-selftest): Rename to...
19283 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19284 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19285 than SELFTEST_FLAGS.
19286 (selftest-gdb): Rename to...
19287 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19289 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19290 (selftest-valgrind): Rename to...
19291 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19293 (selftest-valgrind): Reintroduce as an alias for
19294 selftest-c-valgrind.
19295 (s-selftest-c++): New.
19296 (selftest-c++-gdb): New.
19297 (selftest-c++-valgrind): New.
19299 2017-07-06 Olivier Hainque <hainque@adacore.com>
19301 * gcc.c (process_command): When deciding if undefined variables
19302 should be ignored when processing specs, accept "gcc -v" as well.
19304 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19306 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19307 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19309 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19311 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19312 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19313 * config/arm/arm-tables.opt: Regenerate.
19314 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19316 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19318 2017-07-06 Carl Love <cel@us.ibm.com>
19320 * ChangeLog: Clean up from mid air collision
19322 2017-07-06 Carl Love <cel@us.ibm.com>
19324 * config/rs6000/rs6000-c.c: Add support for built-in functions
19325 vector signed int vec_subc (vector signed int, vector signed int);
19326 vector signed __int128 vec_subc (vector signed __int128,
19327 vector signed __int128);
19328 vector unsigned __int128 vec_subc (vector unsigned __int128,
19329 vector unsigned __int128);
19330 vector signed int vec_sube (vector signed int, vector signed int,
19331 vector signed int);
19332 vector unsigned int vec_sube (vector unsigned int,
19333 vector unsigned int,
19334 vector unsigned int);
19335 vector signed __int128 vec_sube (vector signed __int128,
19336 vector signed __int128,
19337 vector signed__int128);
19338 vector unsigned __int128 vec_sube (vector unsigned __int128,
19339 vector unsigned __int128,
19340 vector unsigned __int128);
19341 vector signed int vec_subec (vector signed int, vector signed int,
19342 vector signed int);
19343 vector unsigned int vec_subec (vector unsigned int,
19344 vector unsigned int,
19345 vector unsigned int);
19346 vector signed __int128 vec_subec (vector signed __int128,
19347 vector signed __int128,
19348 vector signed__int128);
19349 vector unsigned __int128 vec_subec (vector unsigned __int128,
19350 vector unsigned __int128,
19351 vector unsigned __int128);
19352 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19353 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19354 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19355 BU_ALTIVEC_OVERLOAD_X definitions.
19356 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19357 * doc/extend.texi: Update the built-in documentation file for the new
19358 built-in functions.
19360 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19363 * diagnostic-show-locus.c (layout::layout): Use start and finish
19364 spelling location for the start and finish of each range.
19365 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19366 Add unused aspect param.
19367 * input.c (expand_location_1): Add "aspect" param, and use it
19368 to access the correct part of the location.
19369 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19371 (expand_location_to_spelling_point): Likewise.
19372 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19373 param, and pass it to expand_location_1.
19375 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19377 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19378 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19379 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19380 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19381 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19382 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19383 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19384 _mm_maskz_getmant_ss): New intrinsics.
19385 (__builtin_ia32_getexpss128_mask): Changed to ...
19386 __builtin_ia32_getexpss128_round ... this.
19387 (__builtin_ia32_getexpsd128_mask): Changed to ...
19388 __builtin_ia32_getexpsd128_round ... this.
19389 * config/i386/i386-builtin-types.def
19390 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19391 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19392 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19393 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19394 __builtin_ia32_getmantss_mask_round): New builtins.
19395 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19396 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19397 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19398 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19399 * config/i386/sse.md
19400 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19401 avx512f_sgetexp<mode><mask_scalar_name>
19402 <round_saeonly_scalar_name> ... this.
19403 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19404 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19405 vgetexp<ssescalarmodesuffix>
19406 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19407 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19408 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19409 avx512f_vgetmant<mode><mask_scalar_name>
19410 <round_saeonly_scalar_name> ... this.
19411 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19412 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19413 vgetmant<ssescalarmodesuffix>
19414 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19415 %0<mask_scalar_operand4>, %1, %2
19416 <round_saeonly_scalar_mask_op4>, %3} ... this.
19417 * config/i386/subst.md (mask_scalar_operand4,
19418 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19419 round_saeonly_scalar_nimm_predicate): New subst attributes.
19421 2017-07-06 Julia Koval <julia.koval@intel.com>
19423 * config/i386/i386.c (ix86_erase_embedded_rounding):
19424 Remove code for old rounding pattern.
19426 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19428 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19430 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19432 * doc/sourcebuild.texi (Test Directives, Variants of
19433 dg-require-support): Add documentation for dg-require-stack-check.
19435 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19437 * config/i386/subst.md (mask_scalar, round_scalar,
19438 round_saeonly_scalar): New meta-templates.
19439 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19440 round_scalar_mask_operand3, round_scalar_mask_op3,
19441 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19442 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19443 round_saeonly_scalar_constraint,
19444 round_saeonly_scalar_prefix): New subst attribute.
19445 * config/i386/sse.md
19446 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19447 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19448 <round_scalar_name> ... this.
19449 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19450 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19451 <round_scalar_name> ... this.
19452 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19453 <sse>_vm<code><mode>3<mask_scalar_name>
19454 <round_saeonly_scalar_name> ... this.
19455 (v<plusminus_mnemonic><ssescalarmodesuffix>
19456 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19457 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19458 v<plusminus_mnemonic><ssescalarmodesuffix>
19459 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19460 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19461 (v<multdiv_mnemonic><ssescalarmodesuffix>
19462 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19463 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19464 v<multdiv_mnemonic><ssescalarmodesuffix>
19465 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19466 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19467 (v<maxmin_float><ssescalarmodesuffix>
19468 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
19469 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
19470 v<maxmin_float><ssescalarmodesuffix>
19471 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19472 %0<mask_scalar_operand3>, %1, %<iptr>2
19473 <round_saeonly_scalar_mask_op3>} ... this.
19475 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
19477 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
19478 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
19480 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
19481 Alan Hayward <alan.hayward@arm.com>
19482 David Sherwood <david.sherwood@arm.com>
19484 * combine.c (simplify_if_then_else): Remove "enum" before
19486 * compare-elim.c (can_eliminate_compare): Likewise.
19487 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19489 (aarch64_lookup_simd_builtin_type): Likewise.
19490 (aarch64_simd_builtin_type): Likewise.
19491 (aarch64_init_simd_builtin_types): Likewise.
19492 (aarch64_simd_expand_args): Likewise.
19493 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19495 (aarch64_reverse_mask): Likewise.
19496 (aarch64_simd_emit_reg_reg_move): Likewise.
19497 (aarch64_gen_adjusted_ldpstp): Likewise.
19498 (aarch64_ccmp_mode_to_code): Likewise.
19499 (aarch64_operands_ok_for_ldpstp): Likewise.
19500 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19501 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19503 (aarch64_min_divisions_for_recip_mul): Likewise.
19504 (aarch64_reassociation_width): Likewise.
19505 (aarch64_get_condition_code_1): Likewise.
19506 (aarch64_simd_emit_reg_reg_move): Likewise.
19507 (aarch64_simd_attr_length_rglist): Likewise.
19508 (aarch64_reverse_mask): Likewise.
19509 (aarch64_operands_ok_for_ldpstp): Likewise.
19510 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19511 (aarch64_gen_adjusted_ldpstp): Likewise.
19512 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19514 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19515 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19516 (arm_lookup_simd_builtin_type): Likewise.
19517 (arm_simd_builtin_type): Likewise.
19518 (arm_init_simd_builtin_types): Likewise.
19519 (arm_expand_builtin_args): Likewise.
19520 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19521 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19522 (ft32_setup_incoming_varargs): Likewise.
19523 (ft32_function_arg): Likewise.
19524 (ft32_function_arg_advance): Likewise.
19525 (ft32_pass_by_reference): Likewise.
19526 (ft32_arg_partial_bytes): Likewise.
19527 (ft32_valid_pointer_mode): Likewise.
19528 (ft32_addr_space_pointer_mode): Likewise.
19529 (ft32_addr_space_legitimate_address_p): Likewise.
19530 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19532 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19533 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19534 (iamcu_alignment): Likewise.
19535 (canonicalize_vector_int_perm): Likewise.
19536 (ix86_noce_conversion_profitable_p): Likewise.
19537 (ix86_mpx_bound_mode): Likewise.
19538 (ix86_operands_ok_for_move_multiple): Likewise.
19539 * config/microblaze/microblaze-protos.h
19540 (microblaze_expand_conditional_branch_reg): Likewise.
19541 * config/microblaze/microblaze.c
19542 (microblaze_expand_conditional_branch_reg): Likewise.
19543 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19545 (rs6000_reassociation_width): Likewise.
19546 (rs6000_invalid_binary_op): Likewise.
19547 (fusion_p9_p): Likewise.
19548 (emit_fusion_p9_load): Likewise.
19549 (emit_fusion_p9_store): Likewise.
19550 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19552 (riscv_hard_regno_mode_ok_p): Likewise.
19553 (riscv_address_insns): Likewise.
19554 (riscv_split_symbol): Likewise.
19555 (riscv_legitimize_move): Likewise.
19556 (riscv_function_value): Likewise.
19557 (riscv_hard_regno_nregs): Likewise.
19558 (riscv_expand_builtin): Likewise.
19559 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19560 (riscv_build_integer): Likewise.
19561 (riscv_split_integer): Likewise.
19562 (riscv_legitimate_constant_p): Likewise.
19563 (riscv_cannot_force_const_mem): Likewise.
19564 (riscv_regno_mode_ok_for_base_p): Likewise.
19565 (riscv_valid_base_register_p): Likewise.
19566 (riscv_valid_offset_p): Likewise.
19567 (riscv_valid_lo_sum_p): Likewise.
19568 (riscv_classify_address): Likewise.
19569 (riscv_legitimate_address_p): Likewise.
19570 (riscv_address_insns): Likewise.
19571 (riscv_load_store_insns): Likewise.
19572 (riscv_force_binary): Likewise.
19573 (riscv_split_symbol): Likewise.
19574 (riscv_force_address): Likewise.
19575 (riscv_legitimize_address): Likewise.
19576 (riscv_move_integer): Likewise.
19577 (riscv_legitimize_const_move): Likewise.
19578 (riscv_legitimize_move): Likewise.
19579 (riscv_address_cost): Likewise.
19580 (riscv_subword): Likewise.
19581 (riscv_output_move): Likewise.
19582 (riscv_canonicalize_int_order_test): Likewise.
19583 (riscv_emit_int_order_test): Likewise.
19584 (riscv_function_arg_boundary): Likewise.
19585 (riscv_pass_mode_in_fpr_p): Likewise.
19586 (riscv_pass_fpr_single): Likewise.
19587 (riscv_pass_fpr_pair): Likewise.
19588 (riscv_get_arg_info): Likewise.
19589 (riscv_function_arg): Likewise.
19590 (riscv_function_arg_advance): Likewise.
19591 (riscv_arg_partial_bytes): Likewise.
19592 (riscv_function_value): Likewise.
19593 (riscv_pass_by_reference): Likewise.
19594 (riscv_setup_incoming_varargs): Likewise.
19595 (riscv_print_operand): Likewise.
19596 (riscv_elf_select_rtx_section): Likewise.
19597 (riscv_save_restore_reg): Likewise.
19598 (riscv_for_each_saved_reg): Likewise.
19599 (riscv_register_move_cost): Likewise.
19600 (riscv_hard_regno_mode_ok_p): Likewise.
19601 (riscv_hard_regno_nregs): Likewise.
19602 (riscv_class_max_nregs): Likewise.
19603 (riscv_memory_move_cost): Likewise.
19604 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19605 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19606 (rl78_addr_space_address_mode): Likewise.
19607 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19609 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19610 (rs6000_reassociation_width): Likewise.
19611 (rs6000_invalid_binary_op): Likewise.
19612 (fusion_p9_p): Likewise.
19613 (emit_fusion_p9_load): Likewise.
19614 (emit_fusion_p9_store): Likewise.
19615 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19616 (ok_for_simple_move_operands): Likewise.
19617 (ok_for_simple_move_strict_operands): Likewise.
19618 (ok_for_simple_arith_logic_operands): Likewise.
19619 (visium_legitimize_reload_address): Likewise.
19620 (visium_select_cc_mode): Likewise.
19621 (output_cbranch): Likewise.
19622 (visium_split_double_move): Likewise.
19623 (visium_expand_copysign): Likewise.
19624 (visium_expand_int_cstore): Likewise.
19625 (visium_expand_fp_cstore): Likewise.
19626 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19627 (visium_function_arg): Likewise.
19628 (visium_function_arg_advance): Likewise.
19629 (visium_libcall_value): Likewise.
19630 (visium_setup_incoming_varargs): Likewise.
19631 (visium_legitimate_constant_p): Likewise.
19632 (visium_legitimate_address_p): Likewise.
19633 (visium_legitimize_address): Likewise.
19634 (visium_secondary_reload): Likewise.
19635 (visium_register_move_cost): Likewise.
19636 (visium_memory_move_cost): Likewise.
19637 (prepare_move_operands): Likewise.
19638 (ok_for_simple_move_operands): Likewise.
19639 (ok_for_simple_move_strict_operands): Likewise.
19640 (ok_for_simple_arith_logic_operands): Likewise.
19641 (visium_function_value_1): Likewise.
19642 (rtx_ok_for_offset_p): Likewise.
19643 (visium_legitimize_reload_address): Likewise.
19644 (visium_split_double_move): Likewise.
19645 (visium_expand_copysign): Likewise.
19646 (visium_expand_int_cstore): Likewise.
19647 (visium_expand_fp_cstore): Likewise.
19648 (visium_split_cstore): Likewise.
19649 (visium_select_cc_mode): Likewise.
19650 (visium_split_cbranch): Likewise.
19651 (output_cbranch): Likewise.
19652 (visium_print_operand_address): Likewise.
19653 * expmed.c (flip_storage_order): Likewise.
19654 * expmed.h (emit_cstore): Likewise.
19655 (flip_storage_order): Likewise.
19656 * genrecog.c (validate_pattern): Likewise.
19657 * hsa-gen.c (gen_hsa_addr): Likewise.
19658 * internal-fn.c (expand_arith_overflow): Likewise.
19659 * ira-color.c (allocno_copy_cost_saving): Likewise.
19660 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19661 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19662 (process_invariant_for_inheritance): Likewise.
19663 * lra-eliminations.c (move_plus_up): Likewise.
19664 * omp-low.c (lower_oacc_reductions): Likewise.
19665 * simplify-rtx.c (simplify_subreg): Likewise.
19666 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19667 (TARGET_CHKP_BOUND_MODE): Likewise..
19668 * targhooks.c (default_chkp_bound_mode): Likewise.
19669 (default_setup_incoming_vararg_bounds): Likewise.
19670 * targhooks.h (default_chkp_bound_mode): Likewise.
19671 (default_setup_incoming_vararg_bounds): Likewise.
19672 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19673 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19674 (have_whole_vector_shift): Likewise.
19675 * tree-vect-stmts.c (vectorizable_load): Likewise.
19676 * doc/tm.texi: Regenerate.
19678 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19680 Graceful degrade if Binutils PR21472 is not available.
19683 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19684 .rodata in flash test fails.
19685 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19686 * confgure: Regenerate.
19687 * config.in: Regenerate.
19688 * config/avr/avr.c (avr_asm_named_section)
19689 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19690 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19691 (avr_asm_init_sections): Same.
19693 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19695 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19696 (fma<VH:mode>4_intrinsic): Likewise.
19697 (*fmsub<VCVTF:mode>4): Likewise.
19698 (*fmsub<VH:mode>4_intrinsic): Likewise.
19700 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19703 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19704 Don't depend on "optimize > 0".
19705 (out_movhi_r_mr, out_movqi_mr_r): Same.
19706 (out_movhi_mr_r, out_movqi_r_mr): Same.
19707 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19708 io_address_operand on "optimize > 0".
19710 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19712 * tree-loop-distribution.c: Add general explanantion on the pass.
19713 (generate_loops_for_partition): Mark distributed loop.
19714 (pg_add_dependence_edges): New parameter. Handle alias data
19715 dependence specially and record it in the parameter if asked.
19716 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19717 (init_partition_graph_vertices, add_partition_graph_edge): New.
19718 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19719 (free_partition_graph_vdata, build_partition_graph): New.
19720 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19721 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19722 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19723 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19724 (version_for_distribution_p, finalize_partitions): New.
19725 (distribute_loop): Handle alias data dependence specially. Factor
19726 out loop fusion code as functions and call these functions.
19728 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19730 * tree-loop-distribution.c (classify_partition): New parameter and
19731 better handle reduction statement.
19732 (rdg_build_partitions): Revise comment.
19733 (distribute_loop): Compute statements in all partitions and pass it
19734 to classify_partition.
19736 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19738 * tree-loop-distribution.c (enum partition_type): New.
19739 (struct partition): New field type.
19740 (partition_merge_into): Add parameter. Update partition type.
19741 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19742 (build_rdg_partition_for_vertex): Compute partition type.
19743 (rdg_build_partitions): Dump partition type.
19744 (distribute_loop): Update calls to partition_merge_into.
19746 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19748 * tree-loop-distribution.c (struct ddr_hasher): New.
19749 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19751 (classify_partition): Call get_data_dependence.
19752 (pg_add_dependence_edges): Ditto.
19753 (distribute_loop): Release data dependence hash table.
19755 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19757 * tree-loop-distribution.c (ref_base_address): Delete.
19758 (similar_memory_accesses): Rename ...
19759 (share_memory_accesses): ... to this. Check if partitions access
19760 the same memory reference.
19761 (distribute_loop): Call share_memory_accesses.
19763 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19765 * tree-loop-distribution.c (struct partition): New field recording
19766 its data reference.
19767 (partition_alloc, partition_free): Init and release data refs.
19768 (partition_merge_into): Merge data refs.
19769 (build_rdg_partition_for_vertex): Collect data refs for partition.
19770 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19772 (distribute_loop): Remve data refs from vertice data of partition
19775 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19777 * tree-loop-distribution.c (params.h): Include header file.
19778 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19779 (datarefs_vec): New global var.
19780 (create_rdg_vertices): Use datarefs_vec directly.
19781 (free_rdg): Don't free data references.
19782 (build_rdg): Update use. Don't free data references.
19783 (distribute_loop): Compute global variable for data references.
19784 Bail out if there are too many data references.
19786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19788 * tree-loop-distribution.c (loop_nest): New global var.
19789 (build_rdg): Use loop directly, rather than loop nest.
19790 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19792 (distribute_loop): Compute global variable loop nest. Update use.
19794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19796 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19797 (partition_merge_into): New parameter. Dump reason for fusion.
19798 (distribute_loop): Update use of partition_merge_into.
19800 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19802 * tree-loop-distribution.c (bb_top_order_index): New.
19803 (bb_top_order_index_size, bb_top_order_cmp): New.
19804 (stmts_from_loop): Use topological order.
19805 (pass_loop_distribution::execute): Compute and release topological
19806 order for basic blocks.
19808 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19810 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19813 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19815 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19816 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19817 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19818 * internal-fn.def (LOOP_DIST_ALIAS): New.
19819 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19820 (fold_loop_internal_call): ... this.
19821 (vect_loop_dist_alias_call): New function.
19822 (set_uid_loop_bbs): Call fold_loop_internal_call.
19823 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19826 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19829 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19830 Require dead FLAGS_REG at the beginning of a peephole.
19832 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19835 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19836 arguments in the call to __builtin_ia32_sbb_u32.
19837 (_subborrow_u64): Swap _X and _Y arguments in the call to
19838 __builtin_ia32_sbb_u64.
19840 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19843 * tree-vrp.c (compare_assert_loc): Turn into a function template
19844 with stable template parameter. Only test if a->e is NULL,
19845 !a->e == !b->e has been verified already. Use e == NULL or
19846 e != NULL instead of e or ! e tests. If stable is true, don't use
19847 iterative_hash_expr, on the other side allow a or b or both NULL
19848 and sort the NULLs last.
19849 (process_assert_insertions): Sort using compare_assert_loc<false>
19850 instead of compare_assert_loc, later sort using
19851 compare_assert_loc<true> before calling process_assert_insertions_for
19852 in a loop. Use break instead of continue once seen NULL pointer.
19854 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19856 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19857 Cortex-R7 and Cortex-R8 processors.
19859 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19861 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19862 uninitialized while src is not.
19864 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19866 * common/config/arm/arm-common.c: Adjust include path for
19868 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19869 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19870 (arm-cpu-data.h): Likewise.
19871 (arm-cpu-cdata.h): Likewise.
19872 * config/arm/arm-cpu.h: Delete.
19873 * config/arm/arm-cpu-cdata.h: Delete.
19874 * config/arm/arm-cpu-data.h: Delete.
19876 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19878 * config/arm/arm-cpus.in (cortex-a55): New.
19879 (cortex-a75): Likewise.
19880 (cortex-a75.cortex-a55): Likewise.
19881 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19883 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19884 * config/arm/arm-cpu-cdata.h: Regenerate.
19885 * config/arm/arm-cpu-data.h: Regenerate.
19886 * config/arm/arm-cpu.h: Regenerate.
19887 * config/arm/arm-tables.opt: Regenerate.
19888 * config/arm/arm-tune.md: Regenerate.
19890 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19892 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19894 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19896 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19899 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19901 PR tree-optimization/81292
19902 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19903 full_string_p, also call adjust_related_strinfos if the adjustment
19904 is simple, otherwise invalidate related strinfos.
19906 2017-07-04 Martin Liska <mliska@suse.cz>
19909 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19910 newly created variable as DECL_IGNORED_P.
19912 2017-07-04 Martin Liska <mliska@suse.cz>
19915 * ipa-inline.c (inline_small_functions):
19916 Use xstrdup_for_dump.
19918 2017-07-04 Tom de Vries <tom@codesourcery.com>
19920 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19922 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19925 * config/darwin.c (darwin_function_switched_text_sections):
19928 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19930 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19932 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19934 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19936 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19938 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19939 min_profitable_iters, and th as inclusive lower bounds.
19940 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19941 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19942 for min_profitable_iters and min_profitable_estimate.
19943 (vect_transform_loop): Treat th as an inclusive lower bound.
19944 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19946 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19949 * config/darwin.c (darwin_function_switched_text_sections):
19950 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19951 in two pieces, and suppress the use of buf.
19953 2017-07-03 Nathan Sidwell <nathan@acm.org>
19955 * hash-table.h (hash_table_mod1): Fix indentation.
19957 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19959 PR middle-end/81290
19960 * predict.c (force_edge_cold): Be more careful about propagation
19962 * profile-count.h (profile_probability::guessed,
19963 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19965 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19967 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19969 * doc/invoke.texi (rcpc architecture extension): Document it.
19971 2017-07-03 Richard Biener <rguenther@suse.de>
19973 PR tree-optimization/60510
19974 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19975 the scalar reduction PHI and use it.
19976 (vectorizable_reduction): Properly guard the single_defuse_cycle
19977 path for non-SLP reduction chains where we cannot use it.
19978 Rework reduc_def/index and vector type deduction. Rework
19979 vector operand gathering during reduction op code-gen.
19980 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19981 chains dissolve the chain and leave it to non-SLP reduction
19984 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19986 * tree-data-ref.h (dr_alignment): Declare.
19987 * tree-data-ref.c (dr_alignment): New function.
19988 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19989 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19991 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19993 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19995 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19996 and base_misalignment fields.
19997 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19998 * tree-data-ref.c: Include builtins.h.
19999 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20000 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20001 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20002 * tree-vect-data-refs.c: Include tree-cfg.h.
20003 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20004 fields instead of calculating an alignment here.
20005 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20006 innermost_loop_behavior fields.
20008 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20010 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20012 (DR_STEP_ALIGNMENT): New macro.
20013 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20014 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20015 (create_data_ref): Print it.
20016 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20017 to tell whether the step preserves vector (mis)alignment.
20018 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20019 Move the check for an integer step and generalise to all INTEGER_CST.
20020 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20021 Print the outer step alignment.
20023 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20025 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20026 with offset_alignment.
20027 (DR_ALIGNED_TO): Delete.
20028 (DR_OFFSET_ALIGNMENT): New macro.
20029 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20030 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20031 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20032 (create_data_ref): Likewise.
20033 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20034 (vect_analyze_data_refs): Likewise.
20035 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20036 creating dummy innermost behavior.
20038 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20040 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20041 with a "innermost_loop_behavior *" and refeence tree.
20042 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20043 (create_data_ref): Update call accordingly.
20044 * tree-predcom.c (find_looparound_phi): Likewise.
20046 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20048 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20049 fields with dr_wrt_vec_loop.
20050 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20051 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20052 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20053 (vect_dr_behavior): New function.
20054 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20055 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20056 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20057 track whether the step preserves the misalignment.
20058 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20059 Use vect_dr_behavior.
20060 (vect_setup_realignment): Update call accordingly.
20061 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20062 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20063 call to vect_create_addr_base_for_vector_ref.
20064 (vect_create_cond_for_align_checks): Likewise.
20065 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20066 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20067 (vect_recog_mask_conversion_pattern): Likewise.
20068 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20069 (new_stmt_vec_info): Remove redundant zeroing.
20071 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20073 * common/config/arm/arm-common.c (arm_be8_option): New function.
20074 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20075 (ISA_ARMv6): Add isa_bit_be8.
20076 * config/arm/arm.h (arm_be8_option): Add prototype.
20077 (BE8_SPEC_FUNCTION): New define.
20078 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20079 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20080 (mlittle-endian): Similarly.
20081 (mbe8, mbe32): New options.
20082 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20083 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20085 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20087 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20089 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20091 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20092 (cleanup_tree_cfg_bb): Use it.
20093 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20095 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20097 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20100 * loop-doloop.c (add_test): Update profile.
20102 2017-07-03 Martin Liska <mliska@suse.cz>
20105 * sanopt.c (rewrite_usage_of_param): New function.
20106 (sanitize_rewrite_addressable_params): Likewise.
20107 (pass_sanopt::execute): Call rewrite_usage_of_param.
20109 2017-07-03 Richard Biener <rguenther@suse.de>
20111 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20112 back to using VIEW_CONVERT_EXPR.
20114 2017-07-03 Martin Liska <mliska@suse.cz>
20117 * doc/extend.texi: Document when a resolver function is
20118 generated for target_clones.
20120 2017-07-03 Martin Liska <mliska@suse.cz>
20122 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20123 and shadow memory of auto variables which are subject of
20124 use-after-scope sanitization.
20125 (asan_expand_mark_ifn): Add do set only when is_poison.
20127 2016-07-03 Richard Biener <rguenther@suse.de>
20129 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20131 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20132 (vectorizable_reduction): Perform reduction PHI creation when
20133 visiting a reduction PHI and adjust and simplify code generation
20134 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20135 (vect_transform_loop): Visit reduction PHIs.
20136 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20137 defs into the SLP tree.
20138 (vect_build_slp_tree): Reduction defs terminate the recursion.
20139 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20141 (vect_get_vec_defs_for_stmt_copy): Export.
20142 (vect_get_vec_defs): Likewise.
20143 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20145 (vect_get_vec_defs_for_stmt_copy): Declare.
20146 (vect_get_vec_defs): Likewise.
20148 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20150 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20151 parameter with a "loop" parameter and use it instead of the
20152 loop containing DR_STMT. Don't check simple_iv when doing
20153 BB analysis. Describe the two analysis modes in the comment.
20155 2017-07-03 Tom de Vries <tom@codesourcery.com>
20157 PR tree-optimization/69468
20158 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20159 (find_same_succ_bb): Handle ignore_edge_flags.
20161 2017-07-03 Tom de Vries <tom@codesourcery.com>
20163 PR tree-optimization/81192
20164 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20166 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20168 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20170 2017-07-03 Tom de Vries <tom@codesourcery.com>
20172 PR tree-optimization/81192
20173 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20174 BB_SAME_SUCC (bb) == NULL.
20176 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20178 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20181 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20183 * dumpfile.c: Include profile-count.h
20184 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20186 (insert_cond_bb): Update profile.
20187 * tree-cfg.h (insert_cond_bb): Update prototype.
20188 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20189 * tree-dump.c: Do not include tree-cfg.
20191 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20193 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20195 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20197 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20200 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20202 * tree-complex.c (expand_complex_div_wide): update profile.
20204 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20205 Alan Hayward <alan.hayward@arm.com>
20206 David Sherwood <david.sherwood@arm.com>
20208 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20209 (CORETYPES_H): New define.
20210 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20211 (insn-modes-inline.h, s-modes-inline-h): New rules.
20212 (generated_files): Add insn-modes-inline.h.
20213 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20214 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20215 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20216 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20217 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20218 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20219 (build/gencodes.o, build/genconditions.o): Likewise.
20220 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20221 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20222 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20223 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20224 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20225 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20226 * coretypes.h: Include everything up to real.h for generators.
20227 Include insn-modes.h first. Include wide-int-print.h after
20228 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20229 * machmode.h: Don't include insn-modes.h here.
20230 * function-tests.c: Remove includes of signop.h, machmode.h,
20231 double-int.h and wide-int.h.
20233 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20235 * optc-save-gen.awk: Likewise.
20236 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20237 * godump.c: Remove include of wide-int-print.h.
20238 * pretty-print.h: Likewise.
20239 * wide-int-print.cc: Likewise.
20240 * wide-int.cc: Likewise.
20241 * hash-map-tests.c: Remove include of signop.h.
20242 * hash-set-tests.c: Likewise.
20243 * rtl-tests.c: Likewise.
20244 * mkconfig.sh: Remove include of machmode.h.
20245 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20247 (emit_insn_modes_inline_h): ...this new function. Emit the code
20248 into an insn-modes-inline.h header file, adding appropriate
20249 include guards and end comments.
20250 (emit_insn_modes_c_header): Remove include of machmode.h.
20251 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20253 (main): Handle -i flag and call emit_insn_modes_inline_h when
20256 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20258 * tree-ssa-strlen.c (strinfo): Rename the length field to
20259 nonzero_chars. Add a full_string_p field.
20260 (compare_nonzero_chars, zero_length_string_p): New functions.
20261 (get_addr_stridx): Add an offset_out parameter.
20262 Use compare_nonzero_chars.
20263 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20264 (new_strinfo): Update after above changes to strinfo.
20265 (set_endptr_and_length): Set full_string_p.
20266 (get_string_length): Update after above changes to strinfo.
20267 (unshare_strinfo): Update call to new_strinfo.
20268 (maybe_invalidate): Likewise.
20269 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20270 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20271 as a uhwi instead of an shwi. Update after above changes to
20272 strinfo and new_strinfo.
20273 (zero_length_string): Assert that chainsi contains full strings.
20274 Use zero_length_string_p. Update call to new_strinfo.
20275 (adjust_related_strinfos): Update after above changes to strinfo.
20276 Copy full_string_p from origsi.
20277 (adjust_last_stmt): Use zero_length_string_p.
20278 (handle_builtin_strlen): Update after above changes to strinfo and
20279 new_strinfo. Install the lhs as the string length if the previous
20280 entry didn't describe a full string.
20281 (handle_builtin_strchr): Update after above changes to strinfo
20283 (handle_builtin_strcpy): Likewise.
20284 (handle_builtin_strcat): Likewise.
20285 (handle_builtin_malloc): Likewise.
20286 (handle_pointer_plus): Likewise.
20287 (handle_builtin_memcpy): Likewise. Track nonzero characters
20288 that aren't necessarily followed by a nul terminator.
20289 (handle_char_store): Likewise.
20291 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20293 PR tree-optimization/80769
20294 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20295 for malloc and calloc. Document the new invariant that all related
20296 strinfos have delayed lengths or none do.
20297 (verify_related_strinfos): Move earlier in file.
20298 (set_endptr_and_length): New function, split out from...
20299 (get_string_length): ...here. Also set the lengths of related
20301 (zero_length_string): Assert that chainsi has known (rather than
20303 (adjust_related_strinfos): Likewise.
20305 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20307 PR tree-optimization/81136
20308 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20309 assert that two references with the same misalignment have the same
20310 compile-time misalignment if those compile-time misalignments
20313 2017-07-01 Andi Kleen <ak@linux.intel.com>
20315 * print-tree.c (print_node): Print all attributes.
20317 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20319 * cfg.c (scale_bbs_frequencies): New function.
20320 * cfg.h (scale_bbs_frequencies): Declare it.
20321 * cfgloopanal.c (single_likely_exit): Cleanup.
20322 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20324 (scale_loop_profile): Likewise.
20325 (loop_version): Likewise.
20326 (create_empty_loop_on_edge): Update.
20327 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20328 scale_loop_frequencies, scale_loop_profile, loopify,
20329 loop_version): Update prototypes.
20330 * modulo-sched.c (sms_schedule): Update.
20331 * predict.c (unlikely_executed_edge_p): Also check probability.
20332 (probably_never_executed_edge_p): Fix typo.
20333 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20334 * tree-parloops.c (gen_parallel_loop): Update.
20335 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20336 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20337 * tree-ssa-loop-split.c (split_loop): Update.
20338 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20339 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20340 (vect_loop_versioning): Update.
20341 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20343 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20345 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20347 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20349 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20350 to keep profile consistent.
20352 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20354 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20355 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20356 * profile-count.h (max_safe_multiplier): Make unsigned.
20357 (profile_count::guessed_zero): New.
20359 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20361 * bb-reorder.c (fix_up_crossing_landing_pad,
20362 fix_crossing_conditional_branches): Use make_single_succ_edge
20363 to keep profile consistent.
20365 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20367 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20370 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20373 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20374 the right scopes, make sure cond_jump isn't preserved between multiple
20375 iterations. Search for fallthru edge whenever there are 3+ edges and
20376 use find_fallthru_edge for it.
20378 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20380 Patch by Alexander Monakov <amonakov@ispras.ru>
20381 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20382 probabilities consistently.
20384 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20386 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20387 emit_cmp_and_jump_insns.
20389 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20392 * tree-inline.c (expand_call_inline): Combine profile statuses.
20394 2017-06-30 Andrew Pinski <apinski@cavium.com>
20396 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20397 fold_stmt returned true.
20399 2017-06-30 Nathan Sidwell <nathan@acm.org>
20401 * ggc.h (empty_string): Delete.
20402 * cfgexpand.c (expand_asm_stmt): Use plain "".
20403 * optabs.c (expand_asm_memory_barrier): Likewise.
20404 * stringpool.c (empty_string): Delete.
20405 (digit_vector, digit_string): Delete.
20406 (ggc_alloc_string): Use plain "", don't optimize single digit
20407 strings. Use ggc_alloc_atomic.
20409 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20411 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20412 comparison set and one other set, use the cost of the non-comparison
20415 2017-06-30 Nathan Sidwell <nathan@acm.org>
20417 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20420 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20422 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20423 loops. Remove now unneeded calls to gimple_switch_set_label() that
20424 just set removed labels to NULL_TREE.
20426 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20428 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20429 (set_range_info): ...here. Only call set_range_info_raw if domain
20431 (set_nonzero_bits): Call set_range_info_raw.
20432 * tree-ssanames.h (set_range_info_raw): New.
20434 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20437 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20438 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20439 of nonimmediate_operand and <store_mask_constraint> instead of m
20440 for the input operand. For V8FI iterator, always split if input
20441 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20442 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20443 <store_mask_predicate> instead of register_operand and
20444 <store_mask_constraint> instead of v for the input operand. Make
20445 sure both operands aren't MEMs for if not <mask_applied>.
20447 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20449 * lto-wrapper.c (copy_file) Close both file descriptors before
20452 2017-06-30 Martin Liska <mliska@suse.cz>
20455 * multiple_target.c (create_dispatcher_calls): Make ifunc
20456 also for function that don't have calls or are not referenced.
20458 2017-06-30 Richard Biener <rguenther@suse.de>
20460 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
20461 analyze the first scalar stmt. Move vector type computation
20462 for the BB case here from ...
20463 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
20464 live operation processing in the SLP case properly.
20466 2017-06-30 Richard Biener <rguenther@suse.de>
20468 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
20470 2017-06-30 Martin Liska <mliska@suse.cz>
20473 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
20474 before BUILT_IN_UNWIND_RESUME when ASAN is used.
20476 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
20478 * doc/invoke.texi (AArch64): Add missing options and remove redundant
20481 2017-06-30 Richard Biener <rguenther@suse.de>
20483 PR tree-optimization/81249
20484 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
20485 condition reduction result to original scalar type.
20487 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20489 * profile-count.h (enum profile_quality): Fix typos and whitespace
20492 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20494 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20495 type for branch probabilities.
20497 2017-06-29 Julian Brown <julian@codesourcery.com>
20498 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20500 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20501 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20502 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20503 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20505 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20507 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20508 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20509 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20510 CC usage from generic code to here.
20511 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20512 CC usage into the target macros.
20514 2017-06-29 Maya Rashish <coypu@sdf.org>
20516 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20519 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20521 * arm/arm-builtins.c: Include profile-count.h
20522 * except.c (sjlj_emit_function_enter): Use
20523 profile_probability::unlikely.
20525 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20527 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20528 and tocrel_offset be pointer args rather than implicitly using
20530 (legitimate_constant_pool_address_p, rs6000_emit_move,
20531 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20532 tocrel_offset and use in toc_relative_expr_p call.
20533 (print_operand, print_operand_address): Use static tocrel_base_oac
20534 and tocrel_offset_oac.
20535 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20538 2017-06-29 Maya Rashish <coypu@sdf.org>
20540 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20542 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20544 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20545 objects, take into account only the alignment of 'op0' and 'mode1' if
20548 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20550 * ccmp.c (ccmp_tree_comparison_p): New function.
20551 (ccmp_candidate_p): Update to use above function.
20552 (get_compare_parts): New function.
20553 (expand_ccmp_next): Update to use new functions.
20554 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20556 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20557 take mode as argument.
20558 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20559 * expr.c (expand_expr_real_1): Pass mode as argument.
20561 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20563 * combine.c (combine_instructions): Print insns to dump_file, together
20566 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20568 * asan.c (asan_emit_stack_protection): Update.
20569 (create_cond_insert_point): Update.
20570 * auto-profile.c (afdo_propagate_circuit): Update.
20571 * basic-block.h (struct edge_def): Turn probability to
20572 profile_probability.
20573 (EDGE_FREQUENCY): Update.
20574 * bb-reorder.c (find_traces_1_round): Update.
20575 (better_edge_p): Update.
20576 (sanitize_hot_paths): Update.
20577 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20578 (make_single_succ_edge): Update.
20579 (check_bb_profile): Update.
20580 (dump_edge_info): Update.
20581 (update_bb_profile_for_threading): Update.
20582 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20583 probabilitycount to 0.
20584 * cfgbuild.c (compute_outgoing_frequencies): Update.
20585 * cfgcleanup.c (try_forward_edges): Update.
20586 (outgoing_edges_match): Update.
20587 (try_crossjump_to_edge): Update.
20588 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20589 (expand_gimple_tailcall): Update.
20590 (construct_init_block): Use make_single_succ_edge.
20591 (construct_exit_block): Use make_single_succ_edge.
20592 * cfghooks.c (verify_flow_info): Update.
20593 (redirect_edge_succ_nodup): Update.
20594 (split_edge): Update.
20595 (account_profile_record): Update.
20596 * cfgloopanal.c (single_likely_exit): Update.
20597 * cfgloopmanip.c (scale_loop_profile): Update.
20598 (set_zero_probability): Remove.
20599 (duplicate_loop_to_header_edge): Update.
20600 * cfgloopmanip.h (loop_version): Update prototype.
20601 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20602 (force_nonfallthru_and_redirect): Update.
20603 (update_br_prob_note): Update.
20604 (rtl_verify_edges): Update.
20605 (purge_dead_edges): Update.
20606 (rtl_lv_add_condition_to_bb): Update.
20607 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20608 * cgraphunit.c (init_lowered_empty_function): Update.
20609 (cgraph_node::expand_thunk): Update.
20610 * cilk-common.c: Include profile-count.h
20611 * dojump.c (inv): Remove.
20612 (jumpifnot): Update.
20613 (jumpifnot_1): Update.
20614 (do_jump_1): Update.
20616 (do_jump_by_parts_greater_rtx): Update.
20617 (do_compare_rtx_and_jump): Update.
20618 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20619 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20620 * dwarf2cfi.c: Include profile-count.h
20621 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20622 (sjlj_emit_dispatch_table): Likewise.
20623 * explow.c: Include profile-count.h
20624 * expmed.c (emit_store_flag_force): Update.
20625 (do_cmp_and_jump): Update.
20626 * expr.c (compare_by_pieces_d::generate): Update.
20627 (compare_by_pieces_d::finish_mode): Update.
20628 (emit_block_move_via_loop): Update.
20629 (store_expr_with_bounds): Update.
20630 (store_constructor): Update.
20631 (expand_expr_real_2): Update.
20632 (expand_expr_real_1): Update.
20633 * expr.h (try_casesi, try_tablejump): Update prototypes.
20634 * gimple-pretty-print.c (dump_probability): Update.
20635 (dump_profile): New.
20636 (dump_gimple_label): Update.
20637 (dump_gimple_bb_header): Update.
20638 * graph.c (draw_cfg_node_succ_edges): Update.
20639 * hsa-gen.c (convert_switch_statements): Update.
20640 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20641 (find_if_case_1): Update.
20642 (find_if_case_2): Update.
20643 * internal-fn.c (expand_arith_overflow_result_store): Update.
20644 (expand_addsub_overflow): Update.
20645 (expand_neg_overflow): Update.
20646 (expand_mul_overflow): Update.
20647 (expand_vector_ubsan_overflow): Update.
20648 * ipa-cp.c (good_cloning_opportunity_p): Update.
20649 * ipa-split.c (split_function): Use make_single_succ_edge.
20650 * ipa-utils.c (ipa_merge_profiles): Update.
20651 * loop-doloop.c (add_test): Update.
20652 (doloop_modify): Update.
20653 * loop-unroll.c (compare_and_jump_seq): Update.
20654 (unroll_loop_runtime_iterations): Update.
20655 * lra-constraints.c (lra_inheritance): Update.
20656 * lto-streamer-in.c (input_cfg): Update.
20657 * lto-streamer-out.c (output_cfg): Update.
20658 * mcf.c (adjust_cfg_counts): Update.
20659 * modulo-sched.c (sms_schedule): Update.
20660 * omp-expand.c (expand_omp_for_init_counts): Update.
20661 (extract_omp_for_update_vars): Update.
20662 (expand_omp_ordered_sink): Update.
20663 (expand_omp_for_ordered_loops): Update.
20664 (expand_omp_for_generic): Update.
20665 (expand_omp_for_static_nochunk): Update.
20666 (expand_omp_for_static_chunk): Update.
20667 (expand_cilk_for): Update.
20668 (expand_omp_simd): Update.
20669 (expand_omp_taskloop_for_outer): Update.
20670 (expand_omp_taskloop_for_inner): Update.
20671 * omp-simd-clone.c (simd_clone_adjust): Update.
20672 * optabs.c (expand_doubleword_shift): Update.
20673 (expand_abs): Update.
20674 (emit_cmp_and_jump_insn_1): Update.
20675 (expand_compare_and_swap_loop): Update.
20676 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20677 * predict.c (predictable_edge_p): Update.
20678 (edge_probability_reliable_p): Update.
20679 (set_even_probabilities): Update.
20680 (combine_predictions_for_insn): Update.
20681 (combine_predictions_for_bb): Update.
20682 (propagate_freq): Update.
20683 (estimate_bb_frequencies): Update.
20684 (force_edge_cold): Update.
20685 * profile-count.c (profile_count::dump): Add missing space into dump.
20686 (profile_count::debug): Add newline.
20687 (profile_count::differs_from_p): Explicitly convert to unsigned.
20688 (profile_count::stream_in): Update.
20689 (profile_probability::dump): New member function.
20690 (profile_probability::debug): New member function.
20691 (profile_probability::differs_from_p): New member function.
20692 (profile_probability::differs_lot_from_p): New member function.
20693 (profile_probability::stream_in): New member function.
20694 (profile_probability::stream_out): New member function.
20695 * profile-count.h (profile_count_quality): Rename to ...
20696 (profile_quality): ... this one.
20697 (profile_probability): New.
20698 (profile_count): Update.
20699 * profile.c (compute_branch_probabilities): Update.
20700 * recog.c (peep2_attempt): Update.
20701 * sched-ebb.c (schedule_ebbs): Update.
20702 * sched-rgn.c (find_single_block_region): Update.
20703 (compute_dom_prob_ps): Update.
20704 (schedule_region): Update.
20705 * sel-sched-ir.c (compute_succs_info): Update.
20706 * stmt.c (struct case_node): Update.
20707 (do_jump_if_equal): Update.
20708 (get_outgoing_edge_probs): Update.
20709 (conditional_probability): Update.
20710 (emit_case_dispatch_table): Update.
20711 (expand_case): Update.
20712 (expand_sjlj_dispatch_table): Update.
20713 (emit_case_nodes): Update.
20714 * targhooks.c: Update.
20715 * tracer.c (better_p): Update.
20716 (find_best_successor): Update.
20717 * trans-mem.c (expand_transaction): Update.
20718 * tree-call-cdce.c: Update.
20719 * tree-cfg.c (gimple_split_edge): Upate.
20720 (move_sese_region_to_fn): Upate.
20721 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20722 * tree-eh.c (lower_resx): Upate.
20723 (cleanup_empty_eh_move_lp): Upate.
20724 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20725 * tree-inline.c (copy_edges_for_bb): Update.
20726 (copy_cfg_body): Update.
20727 * tree-parloops.c (gen_parallel_loop): Update.
20728 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20729 (gimple_gen_time_profiler): Update.
20730 * tree-ssa-dce.c (remove_dead_stmt): Update.
20731 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20732 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20733 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20734 (unloop_loops): Update.
20735 (try_peel_loop): Update.
20736 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20737 * tree-ssa-loop-split.c (connect_loops): Update.
20738 (split_loop): Update.
20739 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20740 (hoist_guard): Update.
20741 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20742 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20743 (value_replacement): Update.
20744 * tree-ssa-reassoc.c (branch_fixup): Update.
20745 * tree-ssa-tail-merge.c (replace_block_by): Update.
20746 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20747 (create_edge_and_update_destination_phis): Update.
20748 (compute_path_counts): Update.
20749 (recompute_probabilities): Update.
20750 (update_joiner_offpath_counts): Update.
20751 (freqs_to_counts_path): Update.
20752 (duplicate_thread_path): Update.
20753 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20754 (struct switch_conv_info): Update.
20755 (gen_inbound_check): Update.
20756 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20757 (vect_do_peeling): Update.
20758 (vect_loop_versioning): Update.
20759 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20760 (optimize_mask_stores): Update.
20761 * ubsan.c (ubsan_expand_null_ifn): Update.
20762 * value-prof.c (gimple_divmod_fixed_value): Update.
20763 (gimple_divmod_fixed_value_transform): Update.
20764 (gimple_mod_pow2): Update.
20765 (gimple_mod_pow2_value_transform): Update.
20766 (gimple_mod_subtract): Update.
20767 (gimple_mod_subtract_transform): Update.
20768 (gimple_ic): Update.
20769 (gimple_stringop_fixed_value): Update.
20770 (gimple_stringops_transform): Update.
20771 * value-prof.h: Update.
20773 2017-06-29 Carl Love <cel@us.ibm.com>
20775 * config/rs6000/rs6000-c.c: Add support for built-in functions
20776 vector signed int vec_signed (vector float);
20777 vector signed long long vec_signed (vector double);
20778 vector signed int vec_signed2 (vector double, vector double);
20779 vector signed int vec_signede (vector double);
20780 vector signed int vec_signedo (vector double);
20781 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20782 instruction generator.
20783 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20784 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20785 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20787 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20788 vunsignede_v2df): Add define_expands.
20789 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20790 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20791 VEC_UNSIGNEDO): Add definitions.
20792 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20793 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20794 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20795 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20796 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20797 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20798 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20799 vec_unsignede, vec_unsignedo): Add builtin defines.
20800 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20802 * doc/extend.texi: Update the built-in documentation file for the
20803 new built-in functions.
20805 2017-06-29 Richard Biener <rguenther@suse.de>
20807 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20808 reduction chains to LOOP_VINFO_REDUCTIONS.
20809 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20810 SLP reductions after processing reduction chains.
20812 2017-06-29 Nathan Sidwell <nathan@acm.org>
20814 * builtins.c (fold_builtin_FUNCTION): Use
20815 lang_hooks.decl_printable_name.
20817 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20819 PR middle-end/81194
20820 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20821 with only one label.
20822 * stmt.c (expand_case): Assert NCASES is greater than one.
20824 2017-06-29 Richard Biener <rguenther@suse.de>
20826 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20828 (group_case_labels): Likewise.
20829 (find_taken_edge): Push sanity checking on val to workers...
20830 (find_taken_edge_cond_expr): ... here
20831 (find_taken_edge_switch_expr): ... and here, handle cases
20832 with just a default label.
20833 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20834 (group_case_labels): Likewise.
20835 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20836 group_case_labels does anything cleanup the CFG again.
20838 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20840 PR tree-optimization/81196
20841 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20842 exit condition comparing two IVs.
20844 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20846 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20847 profile to the dummy entry at the end of the list of architectures.
20848 * config/arm/arm-cpu-cdata.h: Regenerated.
20850 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20851 Michael Collison <michael.collison@arm.com>
20854 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20856 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20857 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20858 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20859 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20860 of shift when the shift amount is masked with constant equal to
20861 the size of the mode.
20862 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20865 2017-06-29 Martin Liska <mliska@suse.cz>
20867 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20869 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20872 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20875 (edge_growth_cache_entry::edge_growth_cache_entry): New
20877 (reset_edge_growth_cache): Update to use constructor.
20879 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20881 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20882 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20883 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20885 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20887 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20888 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20890 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20892 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20893 (*-linux-uclibc*): Add t-uclibc tmake_file.
20894 * config/t-musl: New.
20895 * config/t-uclibc: New.
20897 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20899 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20901 (gen_comm_data): Emit architectural setting of arch_prof.
20902 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20904 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20905 (armv8-m.base, armv8-m.main): Likewise.
20906 * arm-protos.h (arm_build_target): Add profile field.
20907 (arch_option): Likewise.
20908 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20910 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20911 arm_active_target.profile.
20913 2017-06-28 Richard Biener <rguenther@suse.de>
20915 PR middle-end/81227
20916 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20917 TYPE_OVERFLOW_WRAPS.
20918 * match.pd (negate_expr_p): Likewise.
20919 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20920 fold_build2, not fold_binary.
20922 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20924 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20925 Convert memory address to Pmode.
20926 (aarch64_print_operand): Assert MEM operands are always Pmode.
20928 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20931 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20932 Remove redundant if.
20933 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20934 * config/arm/aarch-common-protos.h
20935 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20936 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20938 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20941 * multiple_target.c (create_dispatcher_calls): Set the default
20942 clone to be static, not public.
20944 2017-06-28 Richard Biener <rguenther@suse.de>
20946 * tree-vect-loop.c (vectorizable_reduction): Move special
20947 cond reduction IV var creation ...
20948 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20949 parameter. Use STMT_VINFO_VECTYPE.
20950 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20953 2017-06-28 Martin Liska <mliska@suse.cz>
20956 * ipa-visibility.c (non_local_p): Handle visibility.
20958 2017-06-28 Martin Liska <mliska@suse.cz>
20961 * common.opt: Add IntegerRange to various options.
20962 * opt-functions.awk (integer_range_info): New function.
20963 * optc-gen.awk: Add integer_range_info to cl_options struct.
20964 * opts-common.c (decode_cmdline_option): Handle
20965 CL_ERR_INT_RANGE_ARG.
20966 (cmdline_handle_error): Likewise.
20967 * opts.c (print_filtered_help): Show valid interval in
20968 when --help is provided.
20969 * opts.h (struct cl_option): Add range_min and range_max fields.
20970 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20972 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20974 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20975 (x * C EQ/NE y * C): New transformation.
20977 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20979 * genmultilib (combination_space): Accept '+' in option names.
20981 2017-06-28 Martin Liska <mliska@suse.cz>
20984 * asan.c (instrument_derefs): Bail out inner references
20985 that are hard register variables.
20987 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20990 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20991 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20993 2017-06-28 Richard Biener <rguenther@suse.de>
20995 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20996 (vect_get_slp_defs): Adjust.
20997 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20999 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21001 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21002 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21003 (vectorizable_reduction): Adjust.
21004 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21006 (vect_get_slp_defs): Likewise.
21007 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21008 (vectorizable_bswap): Adjust.
21009 (vectorizable_call): Likewise.
21010 (vectorizable_conversion): Likewise.
21011 (vectorizable_assignment): Likewise.
21012 (vectorizable_shift): Likewise.
21013 (vectorizable_operation): Likewise.
21014 (vectorizable_store): Likewise.
21015 (vectorizable_condition): Likewise.
21016 (vectorizable_comparison): Likewise.
21018 2017-06-28 Michael Collison <michael.collison@arm.com>
21021 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21023 (arm_gen_movmemqi): Removed unused variable 'i'.
21024 Convert 'for' loop into 'while' loop.
21025 (arm_expand_prologue): Remove last unnecessary set of insn.
21026 (thumb_pop): Remove unused variable 'pushed_words'.
21027 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21029 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21031 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21032 * config/s390/s390-protos.h: Add prototype of
21033 s390_rel_address_ok_p.
21034 * config/s390/s390.c (s390_got_symbol): New function.
21035 (s390_rel_address_ok_p): New function.
21036 (legitimize_pic_address): Use s390_rel_address_ok_p.
21037 (s390_load_got): Use s390_got_symbol.
21038 (s390_option_override): Issue error if
21039 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21040 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21042 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21044 2017-06-27 Andrew Pinski <apinski@cavium.com>
21046 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21047 (X * copysign (1.0, X)): New pattern.
21048 (X * copysign (1.0, -X)): New pattern.
21049 (copysign (-1.0, CST)): New pattern.
21051 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21053 * genmultilib (combination_space): Remove variable.
21054 Validate reuse rules against regular expression for any sequence
21055 of multilib options in any order.
21057 2017-06-27 Michael Collison <michael.collison@arm.com>
21059 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21060 call aarch64_split_simd_combine.
21061 * (aarch64_combine_internal<mode>): Delete pattern.
21062 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21063 Allow register and subreg operands.
21065 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21067 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21068 specific need, just fallback on defaults.
21069 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21071 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21072 Olivier Hainque <hainque@adacore.com>
21074 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21076 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21077 targets. Pick a default if no particular attempt applied.
21078 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21081 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21083 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21084 (x86_64-wrs-vxworks7): Likewise.
21086 2017-06-27 Marek Polacek <polacek@redhat.com>
21089 * ubsan.c (instrument_null): Check get_base_address's result for null.
21091 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21093 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21095 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21097 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21098 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21099 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21100 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21101 New function types.
21102 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21103 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21104 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21105 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21106 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21107 BUILT_IN_FEUPDATEENV): New builtins.
21108 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21109 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21110 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21111 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21113 (builtin_structptr_types): Adjust size.
21114 * tree.c (builtin_structptr_types): Add four entries.
21116 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21117 Olivier Hainque <hainque@adacore.com>
21119 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21120 (TLS_SYM): New local macro, forcing reference to __tls__ on
21121 link command lines for VxWorks 7 RTPs, triggering initialization
21123 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21124 OS features TLS support, true for RTPs on VxWorks 7.
21125 * config/vxworks.c (vxworks_override_options): Setup emutls
21128 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21130 * predict.c (test_prediction_value_range): Use -1U instead of -1
21131 to avoid narrowing conversion warning.
21132 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21133 to avoid narrowing conversion warning.
21134 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21136 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21138 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21140 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21141 64bit configurations.
21142 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21143 (SIZE_TYPE): Likewise.
21144 * config/vxworks.c (vxworks_emutls_var_fields): Use
21145 long_unsigned_type_node instead of unsigned_type_node as the offset
21146 field type, which is "pointer" mode in emutls.c.
21148 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21151 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21153 PR middle-end/81207
21154 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21155 gimple_vuse copying separately from gimple_vdef copying.
21157 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21159 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21162 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21163 Olivier Hainque <hainque@adacore.com>
21165 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21166 for all vxworks7 targets.
21167 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21168 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21169 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21170 variations for VX6/VX7 and 32/64bits later on in ...
21171 (VXWORKS_LIB_SPEC): Leverage new macros.
21172 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21173 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21175 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21177 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21178 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21180 2017-06-26 Carl Love <cel@us.ibm.com>
21182 * config/rs6000/rs6000-c.c: Add support for built-in functions
21183 vector bool char vec_reve (vector bool char);
21184 vector signed char vec_reve (vector signed char);
21185 vector unsigned char vec_reve (vector unsigned char);
21186 vector bool int vec_reve (vector bool int);
21187 vector signed int vec_reve (vector signed int);
21188 vector unsigned int vec_reve (vector unsigned int);
21189 vector bool long long vec_reve (vector bool long long);
21190 vector signed long long vec_reve (vector signed long long);
21191 vector unsigned long long vec_reve (vector unsigned long long);
21192 vector bool short vec_reve (vector bool short);
21193 vector signed short vec_reve (vector signed short);
21194 vector double vec_reve (vector double);
21195 vector float vec_reve (vector float);
21196 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21197 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21198 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21199 (altivec_vreve): New pattern.
21200 * config/rs6000/altivec.h (vec_reve): New define.
21201 * doc/extend.texi (vec_rev): Update the built-in documentation file
21202 for the new built-in functions.
21204 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21206 PR tree-optimization/71815
21207 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21209 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21211 (slsr_process_phi): Likewise.
21212 (replace_uncond_cands_and_profitable_phis): Don't replace a
21213 multiply candidate with a stride of 1 (copy or cast).
21214 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21216 (lowest_cost_path): Likewise.
21217 (total_savings): Likewise.
21219 2017-06-26 Richard Biener <rguenther@suse.de>
21222 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21223 Use def_builtin_pure for all gather builtins.
21225 2017-06-26 Richard Biener <rguenther@suse.de>
21227 PR tree-optimization/81203
21228 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21229 non-dominating BBs.
21231 2017-06-26 Marek Polacek <polacek@redhat.com>
21234 * doc/invoke.texi: Document -Wmultistatement-macros.
21236 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21238 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21239 arm_neon_ok_no_float_abi effective target.
21241 2017-06-26 Richard Biener <rguenther@suse.de>
21243 PR tree-optimization/80928
21244 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21245 (copy_bbs): Set BB_DUPLICATED flag early.
21246 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21248 (execute_on_shrinking_pred): Likewise.
21249 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21250 BB_DUPLICATED blocks.
21251 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21252 iterate over all PHIs considering removal of *gsi.
21254 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21256 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21259 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21261 * config/rs6000/rs6000-string.c: (expand_block_clear,
21262 do_load_for_compare, select_block_compare_mode,
21263 compute_current_alignment, expand_block_compare,
21264 expand_strncmp_align_check, expand_strn_compare,
21265 expand_block_move, rs6000_output_load_multiple)
21266 Move functions related to string/block move/compare
21267 to a separate file.
21268 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21269 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21270 for this function which is now used in two files.
21271 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21272 * config.gcc: Add rs6000-string.o to extra_objs for
21273 targets powerpc*-*-* and rs6000*-*-*.
21275 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21278 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21279 32-bit, since indexed is not valid for DImode.
21280 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21281 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21282 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21283 (define_peephole2 for Altivec d-form store): Likewise.
21286 * multiple_target.c (create_dispatcher_calls): Only create the
21287 dispatcher call if the function is the default clone of a
21288 versioned function.
21290 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21292 PR middle-end/80902
21293 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21294 a call, force the call to not be a tail call.
21296 2017-06-23 Jeff Law <law@redhat.com>
21298 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21301 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21303 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21304 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21305 handling for early expansion of vector shifts (sl,sr,sra,rl).
21306 (builtin_function_type): Add vector shift right instructions
21307 to the unsigned argument list.
21309 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21311 rtl-optimizatoin/79286
21312 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21313 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21314 trap. PIC register plus a const unspec without offset can never trap.
21316 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21318 * tree.h (builtin_structptr_type): New type.
21319 (builtin_structptr_types): Declare new array.
21320 * tree.c (builtin_structptr_types): New array.
21321 (free_lang_data, build_common_tree_nodes): Use it.
21323 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21326 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21329 2017-06-22 Matt Turner <mattst88@gmail.com>
21331 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21332 Lake models to skylake case. Assume skylake for unknown
21333 models with clflushopt.
21335 2017-06-22 Jeff Law <law@redhat.com>
21337 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21338 frame sizes that do not satisfy aarch64_uimm12_shift.
21340 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21342 * profile-count.h (apply_probability,
21343 apply_scale, probability_in): Fix checks for zero.
21345 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21347 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21348 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21350 2016-06-22 Richard Biener <rguenther@suse.de>
21352 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21353 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21354 REDUC_MAX_EXPR support.
21355 (vectorizable_reduction): Likewise.
21356 (vect_create_epilog_for_reduction): Likewise.
21358 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21360 * match.pd (A / (1 << B) -> A >> B): New.
21361 * generic-match-head.c: Include optabs-tree.h.
21362 * gimple-match-head.c: Likewise.
21363 * optabs-tree.h (target_supports_op_p): New.
21364 * optabs-tree.c (target_supports_op_p): New.
21366 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21368 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21369 $gcc_cv_ld --help output.
21370 (gcc_cv_ld_demangle): Likewise.
21371 (gcc_cv_ld_eh_frame_hdr): Likewise.
21372 (gcc_cv_ld_pie): Likewise.
21373 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21374 (gcc_cv_ld_buildid): Likewise.
21375 (gcc_cv_ld_sysroot): Likewise.
21376 (ld_bndplt_support): Likewise.
21377 (ld_pushpopstate_support): Likewise.
21378 * configure: Regenerate.
21379 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21381 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21384 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21385 operands indexes to avoid gap between operands and match_dups.
21387 2017-06-21 Andrew Pinski <apinski@cavium.com>
21389 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21390 Increment Arith_shift and Arith_shift_reg by 1.
21391 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21393 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21394 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21395 (aarch64_strip_extend): Add new argument and test for it.
21396 (aarch64_cheap_mult_shift_p): New function.
21397 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21398 add a cost if it is true.
21399 Update calls to aarch64_strip_extend.
21400 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21402 2017-06-21 Andrew Pinski <apinski@cavium.com>
21404 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21406 (thunderxt88): Likewise.
21407 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21408 (thunderx_prefetch_tune): New variable.
21409 (thunderx2t99_prefetch_tune): Update for the correct values.
21410 (thunderxt88_tunings): New variable.
21411 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21412 generic_prefetch_tune.
21413 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21415 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21417 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21418 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21419 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21420 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21421 (aarch64_atomic_cas<mode>, GPI): Likewise.
21423 2017-06-21 Martin Liska <mliska@suse.cz>
21425 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21426 statements on cold and hot labels.
21427 * predict.c (tree_estimate_probability_bb): Remove the
21428 prediction from this place.
21430 2017-06-21 Martin Liska <mliska@suse.cz>
21432 PR tree-optimization/79489
21433 * gimplify.c (maybe_add_early_return_predict_stmt): New
21435 (gimplify_return_expr): Call the function.
21436 * predict.c (tree_estimate_probability_bb): Remove handling
21438 * predict.def: Update comment about early return predictor.
21439 * gimple-predict.h (is_gimple_predict): New function.
21440 * predict.def: Change default value of early return to 66.
21441 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21443 * passes.def: Put pass_strip_predict_hints to the beginning of
21446 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21448 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21449 FUNCTION_DECL declarations.
21450 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21452 (dwaf2out_decl): Likewise.
21453 * godump.c (go_early_global_decl): Skip call to the real debug hook
21454 for FUNCTION_DECL declarations.
21455 * passes.c (rest_of_decl_compilation): Skip call to the
21456 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21457 -fdump-go-spec is passed.
21459 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21461 * config/i386/i386.c (struct builtin_isa): New field pure_p.
21462 Reorder for compactness.
21463 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
21464 (def_builtin_pure, def_builtin_pure2): New functions.
21465 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
21467 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21469 * match.pd (nop_convert): New predicate.
21470 ((A +- CST1) +- CST2): Allow some NOP conversions.
21472 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21475 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
21476 with ctors/dtors if GOVD_SHARED is set.
21478 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21480 * config/aarch64/aarch64.md (movti_aarch64):
21481 Emit mov rather than orr.
21482 (movtf_aarch64): Likewise.
21483 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
21484 Emit mov rather than orr.
21486 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21488 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
21489 Swap alternatives, make integer dup more expensive.
21491 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21493 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21494 Return true for non-tls symbols.
21496 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21498 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21499 (cortex-a75): Likewise.
21500 (cortex-a75.cortex-a55): Likewise.
21501 * config/aarch64/aarch64-tune.md: Regenerate.
21502 * doc/invoke.texi (-mtune): Document new values for -mtune.
21504 2017-06-21 Tom de Vries <tom@codesourcery.com>
21506 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21507 stack_size feature.
21508 (Effective-Target Keywords, Other attributes): Suggest using
21509 dg-add-options stack_size feature to get stack limit in stack_size
21510 effective target documentation.
21512 2017-06-21 Julian Brown <julian@codesourcery.com>
21513 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21515 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21516 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21517 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21519 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21520 attribute type list for neon_multiply.
21521 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21522 attribute type list for neon_multiply.
21523 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21524 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21525 attribute type list for neon_multiply.
21526 * config/arm/types.md (crypto_pmull): Add.
21527 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21528 attribute type list.
21530 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21532 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21535 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21537 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21538 to make sure not to dereference a NULL cost_classes_ptr pointer.
21540 2017-06-20 Carl Love <cel@us.ibm.com>
21542 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21543 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21544 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21545 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21546 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21547 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21548 VMULOSW): New enum "unspec" values.
21549 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21550 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21551 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21552 altivec_vmulosw): New patterns.
21553 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21554 VMULOSW): Add definitions.
21556 2017-06-20 Julia Koval <julia.koval@intel.com>
21558 * config/i386/i386.c: Fix rounding expand for new pattern.
21559 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21561 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21563 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21564 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21566 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21568 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21571 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21573 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21574 architecture, then by implementer ID.
21575 * config/aarch64/aarch64-tune.md: Regenerate.
21577 2017-06-20 Richard Biener <rguenther@suse.de>
21579 PR middle-end/81097
21580 * fold-const.c (split_tree): Fold to type before negating.
21582 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21584 * diagnostic-show-locus.c
21585 (selftest::test_fixit_deletion_affecting_newline): New function.
21586 (selftest::diagnostic_show_locus_c_tests): Call it.
21588 2017-06-20 Andreas Schwab <schwab@suse.de>
21591 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21594 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21596 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21597 __ARM_FEATURE_COPROC according to support.
21599 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21601 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21602 Rewritten to avoid overflow for > 32-bit pointers.
21605 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21606 by removing enum keyword.
21607 (ubsan_type_descriptor): Likewise. Formatting fix.
21610 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21611 splitter): Require TARGET_SSE2 in the condition.
21613 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21616 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21617 for doing vector set of SFmode on ISA 3.0.
21618 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21619 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21621 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21622 SFmode value into a V4SF variable that was extracted from another
21623 V4SF variable without converting the element to double precision
21624 and back to single precision vector format.
21625 (vsx_insert_extract_v4sf_p9_2): Likewise.
21627 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21629 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21630 in UWHI to avoid undefined overflow.
21633 * ubsan.h (enum ubsan_encode_value_phase): New.
21634 (ubsan_encode_value): Change second argument to
21635 enum ubsan_encode_value_phase with default value of
21636 UBSAN_ENCODE_VALUE_GENERIC.
21637 * ubsan.c (ubsan_encode_value): Change second argument to
21638 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21639 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21640 create_tmp_var_raw instead of create_tmp_var and use a
21642 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21643 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21644 ubsan_encode_value callers.
21647 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21648 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21649 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21651 2017-06-19 Richard Biener <rguenther@suse.de>
21653 PR middle-end/81118
21654 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21655 estimates if we changed anything.
21657 2017-06-19 Richard Biener <rguenther@suse.de>
21659 PR tree-optimization/80887
21660 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21661 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21662 simplified lookups, then reset mprts_hook.
21663 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21665 (try_to_simplify): Likewise.
21667 2017-06-19 Martin Liska <mliska@suse.cz>
21670 * gimplify.c (gimplify_switch_expr):
21671 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21673 2017-06-19 Martin Liska <mliska@suse.cz>
21675 * doc/install.texi: Document that PGO runs in 4 stages.
21677 2017-06-19 Martin Liska <mliska@suse.cz>
21680 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21681 to dispatcher function name.
21682 * multiple_target.c (replace_function_decl): New function.
21683 (create_dispatcher_calls): Redirect both edges and references.
21685 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21687 * profile-count.c (profile_count::dump): Dump quality.
21688 (profile_count::differs_from_p): Update for unsigned val.
21689 * profile-count.h (profile_count_quality): New enum.
21690 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21692 2017-06-19 Richard Biener <rguenther@suse.de>
21694 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21695 struct function as arg.
21696 (estimate_numbers_of_iterations): Export overload with loop arg.
21697 (free_numbers_of_iterations_estimates_loop): Use an overload of
21698 free_numbers_of_iterations_estimates instead.
21699 * tree-cfg.c (remove_bb): Adjust.
21700 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21701 * tree-parloops.c (gen_parallel_loop): Likewise.
21702 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21704 (tree_unroll_loops_completely): Likewise.
21705 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21706 Use an overload instead and export.
21707 (estimated_loop_iterations): Adjust.
21708 (max_loop_iterations): Likewise.
21709 (likely_max_loop_iterations): Likewise.
21710 (estimate_numbers_of_iterations): Take struct function as arg
21712 (loop_exits_before_overflow): Adjust.
21713 (free_numbers_of_iterations_estimates_loop): Use an overload.
21714 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21715 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21717 2017-06-19 Richard Biener <rguenther@suse.de>
21720 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21721 RANGE_EXPR conservatively.
21723 2017-06-16 Carl Love <cel@us.ibm.com>
21725 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21726 definitions for vec_float, vec_float2, vec_floato,
21727 vec_floate built-ins.
21728 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21729 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21731 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21732 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21733 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21734 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21735 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21736 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21737 vec_floato): Add builtin defines.
21738 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21739 Update the built-in documentation file for the new built-in
21742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21744 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21745 (mthumb): Mark as the negative of -marm.
21747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21749 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21751 (ARM Options, -mtune): Document that this accepts the same
21752 extension options as -mcpu.
21753 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21757 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21758 permitted extensions.
21760 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21762 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21763 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21764 (armv8-m.main): Add option +nodsp.
21765 * config/arm/arm-cpu-cdata.h: Regenerated.
21767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21769 * config/arm/t-fuchsia: New file.
21770 * config.gcc (arm*-*-fuchsia*): Use it.
21772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21774 * config/arm/t-symbian: Rewrite for new option infrastructure.
21776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21778 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21779 (MULTILIB_REQUIRED): Likewise.
21781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21783 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21784 (MULTILIB_RESUE): Likewise.
21785 (MULTILIB_MATCHES): Likewise.
21786 (MULTLIB_REQUIRED): Likewise.
21788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21790 * config/arm/t-rtems: Rewrite for new option framework.
21792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21794 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21795 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21796 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21797 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21798 * config/arm/t-multilib: ... here.
21799 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21800 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21801 armv7-a and armv8*-a when A-profile libraries have not been built.
21802 * config/arm/t-rmprofile: Rewrite.
21804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21806 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21807 with a backslash. Remove the backslash after substituting unescaped
21809 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21813 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21814 the driver to pass through the default setting of -mfloat-abi.
21815 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21817 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21818 (all_feat_combs): New rule.
21819 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21821 * config/arm/t-aprofile: Rewrite.
21823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21825 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21826 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21827 fpu is not specified by the user/command-line.
21828 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21829 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21830 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21831 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21832 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21833 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21837 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21838 * config/arm/t-arm-elf: Rewritten.
21840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21842 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21843 have some floating-point instructions.
21844 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21845 (TARGET_MAYBE_HARD_FLOAT): New macro.
21846 * config/arm/arm-builtins.c (arm_init_builtins): Use
21847 TARGET_MAYBE_HARD_FLOAT.
21848 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21850 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21852 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21853 (configargs.h): Include it.
21854 (arm_print_hint_for_fpu_option): New function.
21855 (arm_parse_fpu_option): New function.
21856 (candidate_extension): New class.
21857 (arm_canon_for_multilib): New function.
21858 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21859 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21860 (ARCH_CANONICAL_SPECS): New macro.
21861 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21865 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21866 are set after handling multilib fragments. Set target_cpu_default2
21869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21871 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21873 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21874 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21876 (check_arch): Likewise.
21877 * config/arm/arm.c (arm_configure_build_target): Handle
21878 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21879 options in the default.
21881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21883 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21884 when an option is an alias of another.
21885 * config/arm/parsecpu.awk (optalias): New parser token.
21886 (gen_comm_data): Mark non-alias options as such. Emit entries
21887 for extension aliases.
21888 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21889 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21890 (armv6kz, armv6zk, armv6t2): Likewise.
21891 (armv7): Make vfpv3-d16 an alias.
21892 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21894 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21895 Sort in canonical order.
21896 (armv8-a): Sort in canonical order.
21897 (armv8.1-a, armv8.2-a): Likewise.
21898 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21900 (cortex-a9): Sort in canonical order.
21901 * config/arm/arm.c (selftests.h): Include it.
21902 (arm_test_cpu_arch_data): New function.
21903 (arm_run_self_tests): New function.
21904 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21905 (targetm): Move declaration to the end of the file.
21906 * arm-cpu-cdata.h: Regenerated.
21908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21910 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21911 call to target_mode_check describing the type of option passed.
21912 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21913 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21914 arm_parse_cpu_option_name to match parameters against list of
21916 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21917 arm_arch_core_flags data structure.
21918 * config/arm/arm-cpu_cdata.h: Regenerated.
21920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21922 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21924 (arm_print_hint_for_cpu_option): Likewise.
21925 (arm_print_hint_for_arch_option): Likewise.
21926 (arm_parse_cpu_option_name): Likewise.
21927 (arm_parse_arch_option_name): Likewise.
21928 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21929 of entries in the all_fpus list.
21930 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21931 (arm_parse_cpu_option_name): Declare.
21932 (arm_parse_arch_option_name): Declare.
21933 (arm_parse_option_features): Declare.
21934 (arm_intialize_isa): Declare.
21935 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21937 (gen_comm_data): ... here. Make definitions non-static.
21938 * config/arm/arm-cpu-data.h: Regenerated.
21939 * config/arm/arm-cpu-cdata.h: Regenerated.
21941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21943 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21944 (cpu_arch_extension): New structure.
21945 (cpu_arch_option, arch_option, cpu_option): New structures.
21946 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21947 architecture types.
21948 (gen_data): Generate new format data tables.
21949 * config/arm/arm.c (cpu_tune): New structure.
21950 (cpu_option, processors): Delete.
21951 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21952 (arm_print_hint_for_cpu_option): ... this and ...
21953 (arm_print_hint_for_arch_option): ... this.
21954 (arm_parse_arch_cpu_name): Delete. Replace with ...
21955 (arm_parse_cpu_option_name): ... this and ...
21956 (arm_parse_arch_option_name): ... this.
21957 (arm_unrecognized_feature): Change type of target parameter to
21959 (arm_parse_arch_cpu_features): Delete. Replace with ...
21960 (arm_parse_option_features): ... this.
21961 (arm_configure_build_target): Rework to use new configuration data
21963 (arm_print_tune_info): Rework for new configuration data tables.
21964 * config/arm/arm-cpu-data.h: Regenerated.
21965 * config/arm/arm-cpu.h: Regenerated.
21967 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21969 * Makefile.in (OBJS): Move sbitmap.o from here ...
21970 (OBJS-libcommon): ... to here.
21972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21974 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21975 (ISA_ALL_CRYPTO): New macro.
21976 (ISA_ALL_SIMD): New macro
21977 (ISA_ALL_FP): New macro.
21978 * config/arm/arm.c (fpu_bitlist): Update initializer.
21979 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21981 (arm9e): Add fpu. Add option for nofp
21982 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21983 (arm926ej-s, arm1026ej-s): Likewise.
21984 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21985 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21986 neon-fp16, neon-vfpv4, nofp and nosimd.
21987 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21988 (cortex-a8): Add fpu. Add option for nofp.
21989 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21990 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21991 (cortex-r4f): Add fpu.
21992 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21993 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21995 (cortex-r8): Likewise.
21996 (cortex-m4): Add fpu. Add option for nofp.
21997 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21998 (cortex-a17.cortex-a7): Likewise.
21999 (cortex-a32): Add fpu. Add options for crypto and nofp.
22000 (cortex-a35, cortex-a53): Likewise.
22001 (cortex-a57): Add fpu. Add option for crypto.
22002 (cortex-a72, cortex-a73): Likewise.
22003 (exynos-m1): Likewise.
22004 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22005 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22006 (cortex-m33): Add fpu. Add option for nofp.
22007 * config/arm/arm-cpu-cdata.h: Regenerated
22008 * config/arm/arm-cpu-data.h: Regenerated.
22010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22012 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22013 (armv5te, armv5tej): Likewise.
22014 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22015 (armv7): Add options fp and vfpv3-d16.
22016 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22017 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22019 (armv7ve): Likewise.
22020 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22021 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22022 (armv8-a): Add nocrypto option.
22023 (armv8.1-a, armv8.2-a): Likewise.
22024 (armv8-m.main): add options fp, fp.dp and nofp.
22026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22028 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22030 (armv8-a+crc): Delete.
22031 (armv8.1-a): Add options simd, crypto and nofp.
22032 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22033 (armv8.2-a+fp16): Delete.
22034 (armv8-m.main): Add option dsp.
22035 (armv8-m.main+dsp): Delete.
22036 (cortex-a8): Add fpu. Add nofp option.
22037 (cortex-a9): Add fpu. Add nofp and nosimd options.
22038 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22039 link to main cpu and architecture data structures.
22040 (gen_comm_data): Only put isa attributes from the main architecture
22042 (option): New statement for architecture and CPU entries.
22043 * arm.c (struct cpu_option): New structure.
22044 (struct processors): Add entry for options.
22045 (arm_unrecognized_feature): New function.
22046 (arm_parse_arch_cpu_name): Ignore any characters after the first
22048 (arm_parse_arch_cpu_feature): New function.
22049 (arm_configure_build_target): Separate out any CPU and architecture
22050 features and parse separately. Don't error out if -mfpu=auto is
22051 used with only an architecture string.
22052 (arm_print_asm_arch_directives): New function.
22053 (arm_file_start): Call it.
22054 * config/arm/arm-cpu-cdata.h: Regenerated.
22055 * config/arm/arm-cpu-data.h: Likewise.
22056 * config/arm/arm-tables.opt: Likewise.
22058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22060 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22061 assembler when it is not -mfpu=auto.
22063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22065 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22066 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22067 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22068 (ASM_CPU_SPEC): Rewrite.
22069 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22070 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22071 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22072 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22073 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22074 copied string is NUL-terminated. Also strip any characters prefixed
22076 (arm_rewrite_selected_arch): New function.
22077 (arm_rewrite_march): New function.
22079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22081 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22082 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22083 (march, mcpu, mtune): Convert to string-based options.
22084 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22085 (arm_parse_arch_cpu_name): New function.
22086 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22087 identify selected architecture or CPU.
22088 (arm_option_save): New function.
22089 (TARGET_OPTION_SAVE): Redefine.
22090 (arm_option_restore): Restore string options.
22091 (arm_option_print): Print string options.
22093 2017-06-16 Martin Sebor <msebor@redhat.com>
22095 PR tree-optimization/80933
22096 PR tree-optimization/80934
22097 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22098 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22099 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22100 (gimple_fold_builtin): Call them.
22102 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22104 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22105 as unlikely; update profile.
22107 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22109 * predict.c (force_edge_cold): Handle declaring edges impossible
22112 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22114 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22116 (try_unroll_loop_completely): Fix reporting.
22118 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22120 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22122 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22125 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22126 if given a non-constant argument for an intrinsic which requires a
22129 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22131 * profile.c (compare_freqs): New function.
22132 (branch_prob): Sort edge list.
22133 (find_spanning_tree): Assume that the list is priority sorted.
22135 2017-06-16 Richard Biener <rguenther@suse.de>
22137 PR tree-optimization/81090
22138 * passes.def (pass_record_bounds): Remove.
22139 * tree-pass.h (make_pass_record_bounds): Likewise.
22140 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22141 make_pass_record_bounds): Likewise.
22142 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22143 not free niter estimates at the beginning but at the end.
22144 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22146 2017-06-16 Richard Biener <rguenther@suse.de>
22148 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22149 initializer to workaround ICE in host GCC 4.8.
22151 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22153 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22155 (clone_inlined_nodes): Update.
22157 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22159 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22160 prefetch settings, and enable prefetching by default at -O3.
22162 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22164 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22165 Set flag_prefetch_loop_arrays according to tuning data.
22167 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22169 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22170 New tune structure.
22171 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22172 [Unrelated to main purpose of the patch] Place the pointer field last
22173 to enable type checking errors when tune structure are wrongly merged.
22174 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22175 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22176 (thunderx2t99_prefetch_tune): New tune constants.
22177 (tune_params *_tunings): Update all tunings (no functional change).
22178 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22179 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22180 from tunings structures.
22182 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22185 * ubsan.c (instrument_null): Add T argument, use it instead
22186 of computing it based on IS_LHS.
22187 (instrument_object_size): Likewise.
22188 (pass_ubsan::execute): Adjust instrument_null and
22189 instrument_object_size callers to pass gimple_get_lhs or
22190 gimple_assign_rhs1 result to it. Use instrument_null instead of
22191 calling get_base_address and instrument_mem_ref. Handle
22192 aggregate call arguments for object-size sanitization.
22194 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22196 PR tree-optimization/81089
22197 * tree-vrp.c (is_masked_range_test): Validate operands of
22200 2017-06-15 Martin Sebor <msebor@redhat.com>
22203 * dumpfile.c (dump_register): Avoid calling memset to initialize
22204 a class with a default ctor.
22205 * gcc.c (struct compiler): Remove const qualification.
22206 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22207 * hash-table.h: Ditto.
22208 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22210 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22211 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22213 * params.h (struct param_info): Make struct members non-const.
22214 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22215 with default initialization.
22216 * vec.h (vec_copy_construct, vec_default_construct): New helper
22218 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22219 with vec_copy_construct.
22220 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22221 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22222 * doc/invoke.texi (-Wclass-memaccess): Document.
22224 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22226 * emit-rtl.h (is_leaf): Update comment about local
22227 register allocator.
22229 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22232 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22233 for a variable to have a section before checking if the section has a
22235 Set section to.persistent if persistent attribute is set.
22236 Warn if .persistent attribute is used on an automatic variable.
22238 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22240 PR rtl-optimization/80474
22241 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22243 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22245 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22248 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22250 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22251 (casesi): Emit bounds checking as RTL.
22252 (casesi_internal_mips16_<mode>): Remove bounds checking.
22254 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22256 * config/xtensa/xtensa.c (xtensa_option_override): Append
22257 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22258 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22259 xtensa_doloop_hooks): Define unconditionally.
22260 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22262 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22263 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22264 for it in xtensa_option_override.
22265 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22266 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22268 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22270 * doc/cppopts.texi: Document '-' special value to -MF.
22272 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22274 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22275 (cortex_a53_fconst): Likewise.
22276 (cortex_a53_fpmul): Likewise.
22277 (cortex_a53_f_load_64): Likewise.
22278 (cortex_a53_f_load_many): Likewise.
22279 (cortex_a53_advsimd_alu): Likewise.
22280 (cortex_a53_advsimd_alu_q): Likewise.
22281 (cortex_a53_advsimd_mul): Likewise.
22282 (cortex_a53_advsimd_mul_q): Likewise.
22283 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22284 Add missing fmul, r2f_cvt and fconst cases.
22286 2017-06-14 Richard Biener <rguenther@suse.de>
22288 PR middle-end/81088
22289 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22291 (fold_binary_loc): When associating do not treat pre-existing
22292 TREE_OVERFLOW on literal constants as a reason to allow
22293 TREE_OVERFLOW on associated literal constants.
22295 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22297 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22298 (MASK_FEATURES): New macro.
22299 * config/sparc/sparc.c (sparc_option_override): Remove the special
22300 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22302 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22304 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22305 a division of 0 if non-call exceptions are enabled.
22307 2017-06-14 Andrew Pinski <apinski@cavium.com>
22308 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22311 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22312 Improve vector initialization code gen for only variable case.
22314 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22316 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22318 2017-06-14 Richard Biener <rguenther@suse.de>
22320 PR tree-optimization/81083
22321 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22324 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22326 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22327 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22328 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22329 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22330 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22331 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22333 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22335 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22336 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22340 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22342 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22344 * config/rs6000/t-rtems: Don't handle SPE.
22346 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22348 * config/rs6000/t-linux: Don't handle SPE.
22350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22352 * config/rs6000/eabispe.h: Delete file.
22354 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22356 * config/rs6000/t-spe: Delete file.
22358 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22360 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22361 (rs6000_legitimate_offset_address_p): Return false for anything in
22362 V2SImode or V2SFmode.
22364 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22366 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22367 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22368 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22370 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22371 (rs6000_legitimate_offset_address_p): Ditto.
22372 (rs6000_emit_move): Ditto.
22373 (rs6000_init_builtins): Remove V4HI_type_node.
22375 2017-06-13 Martin Liska <mliska@suse.cz>
22378 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22379 (gate_asan): Likewise.
22380 * asan.h (asan_no_sanitize_address_p): Remove the function.
22381 (sanitize_flags_p): New function.
22382 * builtins.def: Fix coding style.
22383 * common.opt: Use renamed enum value.
22384 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22385 * doc/extend.texi: Document no_sanitize attribute.
22386 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22387 to SANITIZE_UNDEFINED_NONDEFAULT.
22388 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22389 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22390 Use sanitize_flags_p.
22391 * gimplify.c (gimplify_function_tree): Likewise.
22392 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22393 * opts.c (parse_no_sanitize_attribute): New function.
22394 (common_handle_option): Use renamed enum value.
22395 * opts.h (parse_no_sanitize_attribute): Declare.
22396 * tree.c (sanitize_flags_p): New function.
22397 * tree.h: Declared here.
22398 * tsan.c: Use sanitize_flags_p.
22399 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22400 (instrument_mem_ref): Likewise.
22401 (instrument_bool_enum_load): Likewise.
22402 (do_ubsan_in_current_function): Remove the function.
22403 (pass_ubsan::execute): Use sanitize_flags_p.
22404 * ubsan.h: Remove do_ubsan_in_current_function
22405 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22406 (dump_function_to_file): Use it here.
22408 2017-06-13 Martin Jambor <mjambor@suse.cz>
22410 PR tree-optimization/80803
22411 PR tree-optimization/81063
22412 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22413 (propagate_subaccesses_across_link): Enqueue subtree whenever
22414 necessary instead of relying on the caller.
22416 2017-06-13 Martin Jambor <mjambor@suse.cz>
22418 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22419 that have a first_link.
22420 (sort_and_splice_var_accesses): Do not check first_link before
22422 (subtree_mark_written_and_enqueue): Likewise.
22423 (propagate_all_subaccesses): Likewise and do not stop at first
22424 parent with a first_link.
22426 2017-06-13 Martin Jambor <mjambor@suse.cz>
22428 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22431 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22433 * match.pd: New pattern.
22435 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22437 * tree-vrp.c (is_masked_range_test): New function.
22438 (register_edge_assert_for): Determine ranges for
22441 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22443 PR tree-optimization/67328
22444 * fold-const.c (maskable_range_p): New function.
22445 (build_range_check): Generate bittests if possible.
22447 2017-06-13 Martin Liska <mliska@suse.cz>
22449 * gimple-pretty-print.c (dump_probability): Add new argument.
22450 (dump_edge_probability): Dump both probability and count.
22451 (dump_gimple_label): Likewise.
22452 (dump_gimple_bb_header): Likewise.
22454 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22457 * config/avr/avr-devices.c: Fix indentation.
22458 * config/avr/gen-avr-mmcu-specs.c: Dito.
22460 2017-06-13 Richard Biener <rguenther@suse.de>
22462 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
22463 instead get vector type from stmt_info.
22464 (vectorizable_reduction): Adjust. Remove dead code.
22466 2017-06-13 Richard Biener <rguenther@suse.de>
22468 PR middle-end/81065
22469 * fold-const.c (extract_muldiv_1): Remove bogus distribution
22470 case of C * (x * C2 + C3).
22471 (fold_addr_of_array_ref_difference): Properly fold index difference.
22473 2017-06-12 David S. Miller <davem@davemloft.net>
22476 * config/sparc/sparc.md (return expander): Emit frame blockage if
22477 function uses alloca.
22479 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
22481 * combine.c (make_field_assignment): Check len rather than the mode
22482 precision when calling force_to_mode.
22484 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
22486 Support multilibs and devices that see flash in RAM address range.
22489 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
22490 (avr_mcu_t) <flash_pm_offset>: New field.
22491 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22492 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22493 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22494 (AVR_TINY_PM_OFFSET): Remove macro.
22495 * config/avr/avr.opt (-mshort-calls): New option.
22496 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22497 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22498 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22499 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22500 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22501 instead of avr_arch->have_jmp_call.
22502 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22503 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22504 avr_arch->flash_pm_offset to define.
22505 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22506 new field flash_pm_offset. Add entry for avrxmega3.
22507 (avr_texinfo): Add entry for avrxmega3.
22508 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22509 attiny212, attiny214,
22510 attiny412, attiny414, attiny416, attiny417,
22511 attiny814, attiny816, attiny817,
22512 attiny1614, attiny1616, attiny1617,
22513 attiny3214, attiny3216, attiny3217.
22514 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22515 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22516 (avr_print_operand_address) [AVR_TINY]: Same.
22517 (avr_asm_init_sections) <readonly_data_section>: Only patch
22518 callback if avr_arch->flash_pm_offset = 0.
22519 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22520 for rodata if avr_arch->flash_pm_offset != 0.
22521 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22522 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22523 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22524 (m_options): Append opt_rcall.
22525 (m_dirnames): Append dir_rcall.
22526 * config/avr/t-multilib: Regenerate.
22528 * configure.ac [target=avr]: Check whether avrxmega3 default
22529 linker description file works as needed.
22530 * configure: Regenerate.
22531 * doc/avr-mmcu.texi: Regenerate.
22532 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22533 <__AVR_ARCH__>: Document avrxmega3 and 103.
22534 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22535 <__AVR_SHORT_CALLS__>: Document it.
22536 <__AVR_PM_BASE_ADDRESS__>: Document it.
22537 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22538 (AVR Variable Attributes) <progmem>: Document this is
22539 not needed for avrxmega3.
22540 (AVR Named Address Spaces) <__flash>: Dito.
22542 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22544 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22546 2017-06-12 Doug Rupp <rupp@adacore.com>
22548 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22549 Append vxworks-stdint.h to the tm_file list.
22550 * config/vxworks-stdint.h: New file.
22552 2017-06-12 Martin Liska <mliska@suse.cz>
22554 PR tree-optimization/81041
22555 * tree-profile.c (gimple_gen_ic_func_profiler):
22556 Create an extra BB in profile-generate
22557 (gimple_gen_time_profiler): Likewise.
22559 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22561 PR tree-optimization/81003
22562 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22563 (update_range_test): Use it instead of force_gimple_operand_gsi.
22565 2017-06-12 Richard Biener <rguenther@suse.de>
22567 PR tree-optimization/81053
22568 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22569 with backedge value not defined in loop. Simplify def stmt
22572 2017-06-11 Tom de Vries <tom@codesourcery.com>
22575 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22577 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22578 nvptx_cannot_force_const_mem.
22580 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22582 * opts.c (finish_options): Move test for flag_split_stack after
22583 it has been initialized.
22585 2017-06-11 Jason Merrill <jason@redhat.com>
22587 * tree.h (id_equal): New.
22588 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22589 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22590 instead of strcmp of IDENTIFIER_POINTER.
22592 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22594 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22595 (mark_all_inlined_calls_cdtor): Fix formating.
22596 (inline_transform): Rescale profile before inlining.
22598 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22600 * cgraph.h (cgraph_edge::clone): Update prototype.
22601 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22602 (cgraph_node::create_clone): Update.
22603 (cgraph_node::create_version_clone): Update.
22604 * tree-inline.c (copy_bb): Update.
22605 (expand_call_inline): Update.
22607 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22609 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22610 factored out from ...
22611 (rs6000_emit_prologue): ... here.
22613 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22615 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22616 factored out from ...
22617 (rs6000_emit_prologue): ... here.
22619 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22621 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22623 (handle_missing_profiles): Fix computation of tp_first_run.
22624 (counts_to_freqs): Do not touch freqs when count is 0.
22626 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22628 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22631 2017-06-10 Tom de Vries <tom@codesourcery.com>
22633 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22634 attributes): Document signal effective target.
22636 2017-06-10 Tom de Vries <tom@codesourcery.com>
22638 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22639 Document effective target stack_size.
22641 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22643 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22644 to the edit_context if they can be auto-applied.
22646 2017-06-9 Ian Lance Taylor <iant@golang.org>
22648 * opts.c (finish_options): If -fsplit-stack, disable implicit
22649 -forder-blocks-and-partition.
22650 * doc/invoke.texi (Optimize Options): Document that when using
22651 -fsplit-stack -forder-blocks-and-partition is not implicitly
22654 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22656 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22657 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22658 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22659 * builtins.def (abort, trap, unreachable): Declare cold.
22660 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22661 * tree-core.h (ECF_COLD): New.
22662 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22663 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22665 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22667 * predict.c (unlikely_executed_stmt_p): Cleanup.
22669 2017-06-09 Richard Biener <rguenther@suse.de>
22671 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22672 model if the ref is always written to.
22674 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22676 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22678 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22680 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22683 2017-06-09 Tom de Vries <tom@codesourcery.com>
22686 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22687 "target cannot support label values" when encountering LABEL_REF.
22689 2017-06-09 Martin Liska <mliska@suse.cz>
22691 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22692 (gimple_gen_ic_func_profiler): Emit direct comparison
22693 of __gcov_indirect_call_callee with NULL.
22694 (gimple_gen_time_profiler): Change probability from
22695 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22697 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22699 * profile.c (edge_gcov_counts): Turn to pointer.
22700 (compute_branch_probabilities, compute_branch_probabilities): Update.
22701 (branch_prob): Do not clear edge_gcov_count.
22702 * profile.h (edge_gcov_counts): Turn to pointer.
22703 (edge_gcov_count): Update.
22705 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22707 * gimple.h (gimple_check_failed): Mark cold.
22709 2017-06-09 Richard Biener <rguenther@suse.de>
22711 PR tree-optimization/66623
22712 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22713 refactor check_reduction into two parts, properly computing
22714 whether we have to check reduction validity for outer loop
22717 2017-06-09 Richard Biener <rguenther@suse.de>
22719 PR tree-optimization/79483
22720 * graphite-scop-detection.c (order): New global.
22721 (get_order): Compute bb to order mapping that satisfies code
22722 generation constraints.
22723 (cmp_pbbs): New helper.
22724 (build_scops): Start domwalk at entry block, sort generated
22727 2017-06-09 Richard Biener <rguenther@suse.de>
22729 PR middle-end/81007
22730 * ipa-polymorphic-call.c
22731 (ipa_polymorphic_call_context::restrict_to_inner_class):
22732 Skip FIELD_DECLs with error_mark_node type.
22733 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22736 2017-06-09 Martin Liska <mliska@suse.cz>
22738 * predict.c (struct branch_predictor): New struct.
22739 (test_prediction_value_range): New test.
22740 (predict_c_tests): New function.
22741 * selftest-run-tests.c (selftest::run_tests): Run the function.
22742 * selftest.h: Declare new tests.
22744 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22747 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22748 gen_add3_insn did not fail.
22749 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22750 r0, construct that number in a temporary reg and add that reg to r0.
22751 If asked to put the result in r0 as well, fail.
22753 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22756 for early expansion of vec_eqv.
22758 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22760 PR middle-end/81005
22761 * ubsan.c (instrument_null): Avoid pointless code temporary.
22762 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22764 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22768 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22771 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22774 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22776 * predict.c (unlikely_executed_edge_p): Move ahead.
22777 (probably_never_executed_edge_p): Use it.
22779 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22781 PR middle-end/79988
22782 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22783 gimple_call_builtin_p call.
22785 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22787 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22788 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22789 rtl_check_failed_type2, rtl_check_failed_code1,
22790 rtl_check_failed_code2, rtl_check_failed_code_mode,
22791 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22792 rtvec_check_failed_bounds, rtl_check_failed_flag,
22793 _fatal_insn_not_found, _fatal_insn): Likewise.
22794 * tree.h (tree_contains_struct_check_failed,
22795 tree_check_failed, tree_not_check_failed,
22796 tree_class_check_failed, tree_range_check_failed,
22797 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22798 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22799 tree_operand_check_failed, omp_clause_check_failed,
22800 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22803 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22805 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22806 flag_branch_probabilities.
22807 * ipa-inline.c (edge_badness): Likewise.
22808 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22809 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22810 * predict.c (maybe_hot_frequency_p): Likewise.
22811 (probably_never_executed): Likewise.
22812 * sched-ebb.c (schedule_ebbs): Likewise.
22813 * sched-rgn.c (find_single_block_region): Likewise.
22814 * tracer.c (tail_duplicate): Likewise.
22816 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22818 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22819 longer requires x_flag_profile_use.
22821 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22823 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22824 instead of flag_reorder_blocks_and_partition.
22825 * dbxout.c (dbxout_function_end): Likewise.
22826 * dwarf2out.c (gen_subprogram_die): Likewise.
22827 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22828 * hw-doloop.c (reorg_loops): Likewise.
22829 * varasm.c (assemble_start_function,
22830 assemble_end_function): Likewise.
22831 (decide_function_section): Do not check for
22832 flag_reorder_blocks_and_partition.
22834 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22836 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22838 (chkp_get_hard_register_fake_addr_expr): Ditto.
22839 (chkp_build_addr_expr): Add check for hard reg case.
22840 (chkp_parse_array_and_component_ref): Ditto.
22841 (chkp_find_bounds_1): Ditto.
22842 (chkp_process_stmt): Don't generate bounds store for
22845 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22847 * predict.c (maybe_hot_bb_p): Do not check profile status.
22848 (maybe_hot_edge_p): Likewise.
22849 (probably_never_executed): Check for zero counts even if profile
22851 (unlikely_executed_edge_p): New function.
22852 (unlikely_executed_stmt_p): New function.
22853 (unlikely_executed_bb_p): New function.
22854 (set_even_probabilities): Use unlikely predicates.
22855 (combine_predictions_for_bb): Likewise.
22856 (predict_paths_for_bb): Likewise.
22857 (predict_paths_leading_to_edge): Likewise.
22858 (determine_unlikely_bbs): New function.
22859 (estimate_bb_frequencies): Use it.
22860 (compute_function_frequency): Use zero counts even if profile is
22862 * profile-count.h: Fix typo.
22864 2017-08-08 Julia Koval <julia.koval@intel.com>
22866 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22867 _mm512_mask_cvtsepi16_storeu_epi8,
22868 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22869 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22870 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22871 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22872 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22873 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22874 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22875 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22876 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22877 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22878 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22879 __builtin_ia32_pmovuswb256mem_mask,
22880 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22881 __builtin_ia32_pmovwb512mem_mask): New builtins.
22883 2017-08-08 Julia Koval <julia.koval@intel.com>
22885 PR target/73350,80862
22886 * config/i386/subst.md (round): Fix round pattern.
22887 * config/i386/i386.c (ix86_erase_embedded_rounding):
22888 Fix erasing rounding for the fixed pattern.
22890 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22892 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22894 2017-06-08 Martin Liska <mliska@suse.cz>
22896 PR gcov-profile/80911
22897 * gcov.c (block_info::block_info): New constructor.
22899 2017-06-07 Carl Love <cel@us.ibm.com>
22901 * config/rs6000/rs6000-c: The return type of the following
22902 built-in functions was implemented as int not long long. Fix sign
22903 of return value for the unsigned version of vec_mulo and vec_mule.
22904 vector unsigned long long vec_bperm (vector unsigned long long,
22905 vector unsigned char)
22906 vector signed long long vec_mule (vector signed int,
22908 vector unsigned long long vec_mule (vector unsigned int,
22909 vector unsigned int)
22910 vector signed long long vec_mulo (vector signed int,
22912 vector unsigned long long vec_mulo (vector unsigned int,
22913 vector unsigned int)
22914 * doc/extend.texi: Fix the documentation for the built-in
22917 2017-06-07 Carl Love <cel@us.ibm.com>
22920 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22923 2017-06-07 Carl Love <cel@us.ibm.com>
22925 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22926 support, Generate doublehv for signed int/float for BE case only.
22928 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22930 * doc/invoke.texi (mcx16): Rewrite.
22932 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22934 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22935 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22936 *mov<mode>_softfloat, and an anonymous splitter): Use
22937 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22939 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22941 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22943 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22944 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22945 (rs6000_debug_reg_global): Adjust.
22946 (rs6000_init_hard_regno_mode_ok): Adjust.
22947 (rs6000_dbx_register_number): Adjust.
22948 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22949 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22950 Remove SPE_ACC and SPEFSCR.
22951 (REG_ALLOC_ORDER): Ditto.
22952 (FRAME_POINTER_REGNUM): Change to 111.
22953 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22954 (REG_CLASS_NAMES): Ditto.
22955 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22956 (REGISTER_NAMES): Ditto.
22957 (ADDITIONAL_REG_NAMES): Ditto.
22958 (rs6000_reg_names): Ditto.
22959 * config/rs6000/rs6000.md: Renumber some register number
22962 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22964 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22966 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22967 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22969 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22970 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22971 Delete the SPE high registers.
22972 (REG_ALLOC_ORDER): Ditto.
22973 (enum reg_class): Remove SPE_HIGH_REGS.
22974 (REG_CLASS_NAMES): Ditto.
22975 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22976 (REGISTER_NAMES): Ditto.
22977 (rs6000_reg_names): Ditto.
22978 * doc/tm.texi.in: Remove SPE as example.
22979 * doc/tm.texi: Regenerate.
22981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22983 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22984 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22985 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22986 * config/rs6000/rs6000.md (type): Remove "brinc".
22988 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22990 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22991 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22992 * config/rs6000/linuxspe.h: Delete file.
22993 * config/rs6000/rs6000.md: Don't include spe.md.
22994 * config/rs6000/spe.h: Delete file.
22995 * config/rs6000/spe.md: Delete file.
22996 * config/rs6000/t-rs6000: Remove spe.md.
22998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23000 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23001 (reg_or_none500mem_operand): Delete.
23002 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23003 instead of reg_or_none500mem_operand.
23005 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23007 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23008 handling of SPE flags.
23009 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23011 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23013 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23015 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23016 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23017 paired_divv2sf3): Similar.
23018 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23019 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23020 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23022 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23023 Rename the paired_* instruction patterns.
23024 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23026 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23027 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23028 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23029 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23030 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23031 PAIRED_VECTOR_MODE.
23032 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23033 (spe_func_has_64bit_regs_p): Delete.
23034 (spe_expand_predicate_builtin): Delete.
23035 (spe_expand_evsel_builtin): Delete.
23036 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23037 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23038 (invalid_e500_subreg): Delete.
23039 (rs6000_legitimize_address): Always force_reg op2 as well, for
23040 paired single memory accesses.
23041 (rs6000_member_type_forces_blk): Delete.
23042 (rs6000_spe_function_arg): Delete.
23043 (rs6000_expand_unop_builtin): Delete SPE handling.
23044 (rs6000_expand_binop_builtin): Ditto.
23045 (spe_expand_stv_builtin): Delete.
23046 (bdesc_2arg_spe): Delete.
23047 (spe_expand_builtin): Delete.
23048 (spe_expand_predicate_builtin): Delete.
23049 (spe_expand_evsel_builtin): Delete.
23050 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23051 (spe_init_builtins): Delete.
23052 (spe_func_has_64bit_regs_p): Delete.
23053 (savres_routine_name): Delete "info" parameter. Adjust callers.
23054 (rs6000_emit_stack_reset): Ditto.
23055 (rs6000_dwarf_register_span): Delete.
23056 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23057 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23058 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23060 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23062 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23063 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23064 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23065 mulv2sf3, divv2sf3): Delete expanders.
23067 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23069 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23071 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23073 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23074 * config/rs6000/rs6000.c: Ditto.
23076 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23078 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23079 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23080 comparison_operator.
23082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23084 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23085 * config/rs6000/rs6000.opt: Ditto.
23086 * config/rs6000/t-rtems: Ditto.
23088 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23090 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23091 TARGET_E500_SINGLE by 0, simplify.
23092 * config/rs6000/rs6000.c: Ditto.
23093 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23094 (spe_build_register_parallel): Delete.
23095 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23096 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23097 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23098 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23099 (E500_CONVERT): Delete.
23100 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23102 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23104 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23105 * config/rs6000/dfp.md: Ditto.
23106 (negdd2, *negdd2_fpr): Merge.
23107 (absdd2, *absdd2_fpr): Merge.
23108 (negtd2, *negtd2_fpr): Merge.
23109 (abstd2, *abstd2_fpr): Merge.
23110 * config/rs6000/e500.h: Delete file.
23111 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23112 TARGET_FPRS by 1 and simplify.
23113 * config/rs6000/rs6000-c.c: Ditto.
23114 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23115 TARGET_DF_SPE by 0.
23116 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23118 * config/rs6000/rs6000.md: Ditto.
23119 (floatdidf2, *floatdidf2_fpr): Merge.
23120 (move_from_CR_gt_bit): Delete.
23121 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23122 (E500_CR_IOR_COMPARE): Delete.
23123 (All patterns that require !TARGET_FPRS): Delete.
23124 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23128 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23130 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23132 * graphds.c (add_edge): Intitialize edge's attached data.
23133 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23134 pointer parameter. Call pointed function on each edge during
23135 graph traversing. Skip traversing the edge when the function
23137 (graphds_dfs, graphds_scc): Ditto.
23138 (for_each_edge): New parameter. Pass the new parameter to callback
23140 * graphds.h (skip_edge_callback): New function pointer type.
23141 (graphds_dfs, graphds_scc): New function pointer parameter.
23142 (graphds_edge_callback, for_each_edge): New parameter.
23144 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23146 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23147 out code checking if runtime alias check is possible to below ...
23148 Call the new function.
23149 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23150 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23152 2017-06-07 Marek Polacek <polacek@redhat.com>
23155 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23156 TYPE_OVERFLOW_WRAPS check.
23158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23160 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23161 if versioning is required.
23162 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23163 peeling with the check for versioning.
23165 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23167 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23168 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23169 Set true to new parameter if new ssa variable is defined.
23170 (vect_gen_vector_loop_niters): Refactor. Set range information
23171 for the new vector loop bound variable.
23172 (vect_do_peeling): Ditto.
23174 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23176 * tree-affine.c (ssa.h): Include header file.
23177 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23178 has wrapping overflow behavior.
23180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23182 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23184 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23186 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23187 (tree_to_aff_combination): ... here.
23189 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23191 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23192 reg_pressure model function.
23193 (ivopts_global_cost_for_size): Delete.
23194 (determine_set_costs, iv_ca_recount_cost): Call new model function
23195 ivopts_estimate_reg_pressure.
23197 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23199 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23200 expensive than udiv. Remove floating point cases from mod.
23202 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23204 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23205 Increase idiv cost.
23207 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23209 * config/aarch64/aarch64.md
23210 (copysignsf3): Fix mask generation.
23212 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23214 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23216 (class dump_manager): Add register_dumps method.
23217 * dumpfile.c: Include langhooks.h.
23218 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23219 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23220 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23221 (dump_manager::dump_register): Start with 512 entries instead of 32.
23222 (dump_manager::register_dumps): New method.
23223 * toplev.c (general_init): Instead of invoking register_dumps
23224 langhook, invoke register_dumps method on the dump manager.
23225 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23228 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23230 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23231 Say that # requires an associated define_split to exist, and that
23232 the define_split must be suitable for use after register allocation.
23234 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23236 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23237 (compute_outgoing_frequencies): Also initialize zero counts.
23238 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23239 around loops; preserve more of profile when nothing changes.
23241 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23243 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23245 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23246 * config/arm/arm-cpu-cdata.h: Regenerate.
23247 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23248 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23249 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23250 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23251 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23253 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23255 * config/arm/t-rmprofile: Likewise.
23256 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23258 2017-06-06 David S. Miller <davem@davemloft.net>
23261 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23262 blockage if function uses alloca.
23264 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23266 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23267 New "uid" fields to hold pretty-print IDs of group and ref.
23268 Memory references are now identified as <group_id>:<ref_id>
23269 instead of using [random] addresses.
23270 (dump_mem_details): Simplify, no functional change.
23271 (dump_mem_ref): Simplify and make output more concise.
23272 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23273 (find_or_create_group): Initialize group uid.
23274 (record_ref): Initialize ref uid. Improve debug output.
23275 (prune_group_by_reuse, should_issue_prefetch_p,)
23276 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23277 (mark_nontemporal_store, determine_loop_nest_reuse):
23278 Improve debug output.
23280 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23282 * dbgcnt.def (prefetch): New debug counter.
23283 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23284 (schedule_prefetches): Stop issueing prefetches if debug counter
23287 2017-06-06 Tom de Vries <tom@codesourcery.com>
23289 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23290 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23292 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23294 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23295 Use aarch64_reg_or_zero predicate for operand 4.
23296 (aarch64_compare_and_swap<mode> define_insn_and_split):
23297 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23298 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23300 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23302 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23303 (arm_compute_save_core_reg_mask): This.
23304 (thumb1_compute_save_reg_mask): Rename into ...
23305 (thumb1_compute_save_core_reg_mask): This.
23306 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23307 (arm_compute_frame_layout): Likewise.
23309 2017-06-06 Richard Biener <rguenther@suse.de>
23311 PR tree-optimization/80974
23312 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23313 keep or clear leaders SSA info.
23315 2017-06-06 Tom de Vries <tom@codesourcery.com>
23317 * config/nvptx/nvptx.c (split_mode_p): New function.
23318 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23320 2017-06-06 Tom de Vries <tom@codesourcery.com>
23322 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23324 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23327 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23330 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23332 * shrink-wrap.c (handle_simple_exit): Update profile.
23333 (try_shrink_wrapping): Upate profile.
23335 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23337 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23338 (tree_guess_outgoing_edge_probabilities): New.
23339 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23340 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23342 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23344 * ipa-split.c (split_function): Initialize return bb profile.
23346 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23348 * profile.c (compute_branch_probabilities): Also initialize
23349 EXIT_BLOCK profile.
23351 2017-06-06 Richard Biener <rguenther@suse.de>
23353 PR tree-optimization/80928
23354 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23355 (vect_analyze_loop_operations): Properly guard analysis for
23357 (vect_transform_loop): Likewise.
23358 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23359 (vect_model_induction_cost): Do not cost for pure SLP.
23360 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23361 of induction in inner loop vectorization.
23362 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23363 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23364 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23366 (vect_analyze_slp_cost_1): Cost induction.
23367 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23368 (vect_get_slp_vect_defs): Likewise.
23369 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23370 (vect_transform_stmt): Handle SLP reductions.
23371 * tree-vectorizer.h (vectorizable_induction): Adjust.
23373 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23375 * config/rs6000/rs6000.c (make_resolver_func): Update
23376 init_lowered_empty_function call.
23378 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23380 * doc/invoke.texi: Document the -fprofile-abs-path option.
23381 * common.opt (fprofile-abs-path): New option.
23382 * gcov-io.h (gcov_write_filename): Declare.
23383 * gcov-io.c (gcov_write_filename): New function.
23384 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23385 * profile.c (output_location): Likewise.
23387 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23389 * shring-wrap.c: Revert accidental commit.
23391 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23393 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23395 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23397 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23399 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23400 profile in callgraph edge.
23401 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23402 (apply_scale): Likewise.
23403 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23404 Also scale profile when inlining function with zero profile.
23405 (initialize_cfun): Update exit block profile even when it is zero.
23406 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23407 when profile is read.
23409 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23411 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23412 (CLONE_*): New constants to define the processors we can generate
23413 code for with the target_clone attribute.
23414 (rs6000_clone_map): New array to identify which clone processors
23415 the current program is running on.
23416 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23417 target_clone attribute.
23418 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23419 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23420 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23421 (cpu_expand_builtin): Add support for target_clone attribute.
23422 (rs6000_valid_attribute_p): Allow "default" attribute.
23423 (get_decl_name): New debug function to simplify printing the
23424 current function name in debugging statements.
23425 (rs6000_clone_priority): New functions to support the target_clone
23426 attribute, and be able to generate code to switch between ISA 2.05
23427 through ISA 3.0 (power6 through power9).
23428 (rs6000_compare_version_priority): Likewise.
23429 (rs6000_get_function_versions_dispatcher): Likewise.
23430 (make_resolver_func): Likewise.
23431 (add_condition_to_bb): Likewise.
23432 (dispatch_function_versions): Likewise.
23433 (rs6000_generate_version_dispatcher_body): Likewise.
23434 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23435 (fusion_gpr_load_p): Fix a spacing issue.
23436 * doc/extend.texi (Common Function Attributes): Document that the
23437 PowerPC supports the target_clone attribute.
23439 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23441 * config/arm/arm.h: explain F symbol found in description of ARM
23442 register allocation in its legend.
23444 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23446 * config/mips/frame-header-opt.c: Include profile-count.h.
23447 * config/riscv/riscv.c: Include profile-count.h
23449 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23451 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23453 (sm_set_flag_if_changed): Add bbs field.
23454 (execute_sm_if_changed_flag_set): Pass BBS.
23455 (execute_sm): Update.
23457 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23459 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
23462 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23464 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
23465 (peephole2): New peephole2 to emit the above.
23466 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
23468 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23470 * config/aarch64/aarch64.c (define_peephole2 above
23471 *sub_<shift>_<mode>): New peephole.
23473 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23475 * config/i386/i386.c (make_resolver_func): Update.
23476 * Makefile.in: Add profile-count.h and profile-count.o
23477 * auto-profile.c (afdo_indirect_call): Update to new API.
23478 (afdo_set_bb_count): Update.
23479 (afdo_propagate_edge): Update.
23480 (afdo_propagate_circuit): Update.
23481 (afdo_calculate_branch_prob): Update.
23482 (afdo_annotate_cfg): Update.
23483 * basic-block.h: Include profile-count.h
23484 (struct edge_def): Turn count to profile_count.
23485 (struct basic_block_def): Likewie.
23486 (REG_BR_PROB_BASE): Move to profile-count.h
23487 (RDIV): Move to profile-count.h
23488 * bb-reorder.c (max_entry_count): Turn to profile_count.
23489 (find_traces): Update.
23490 (rotate_loop):Update.
23491 (connect_traces):Update.
23492 (sanitize_hot_paths):Update.
23493 * bt-load.c (migrate_btr_defs): Update.
23494 * cfg.c (RDIV): Remove.
23495 (init_flow): Use alloc_block.
23496 (alloc_block): Uninitialize count.
23497 (unchecked_make_edge): Uninitialize count.
23498 (check_bb_profile): Update.
23499 (dump_edge_info): Update.
23500 (dump_bb_info): Update.
23501 (update_bb_profile_for_threading): Update.
23502 (scale_bbs_frequencies_int): Update.
23503 (scale_bbs_frequencies_gcov_type): Update.
23504 (scale_bbs_frequencies_profile_count): New.
23505 * cfg.h (update_bb_profile_for_threading): Update.
23506 (scale_bbs_frequencies_profile_count): Declare.
23507 * cfgbuild.c (compute_outgoing_frequencies): Update.
23508 (find_many_sub_basic_blocks): Update.
23509 * cfgcleanup.c (try_forward_edges): Update.
23510 (try_crossjump_to_edge): Update.
23511 * cfgexpand.c (expand_gimple_tailcall): Update.
23512 (construct_exit_block): Update.
23513 * cfghooks.c (verify_flow_info): Update.
23514 (dump_bb_for_graph): Update.
23515 (split_edge): Update.
23516 (make_forwarder_block): Update.
23517 (duplicate_block): Update.
23518 (account_profile_record): Update.
23519 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23520 (get_estimated_loop_iterations): Update.
23521 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23522 (single_likely_exit): Update.
23523 * cfgloopmanip.c (scale_loop_profile): Update.
23525 (set_zero_probability): Update.
23526 (lv_adjust_loop_entry_edge): Update.
23527 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23528 (purge_dead_edges): Update.
23529 (rtl_account_profile_record): Update.
23530 * cgraph.c (cgraph_node::create): Uninitialize count.
23531 (symbol_table::create_edge): Uninitialize count.
23532 (cgraph_update_edges_for_call_stmt_node): Update.
23533 (cgraph_edge::dump_edge_flags): Update.
23534 (cgraph_node::dump): Update.
23535 (cgraph_edge::maybe_hot_p): Update.
23536 * cgraph.h: Include profile-count.h
23537 (create_clone), create_edge, create_indirect_edge): Update.
23538 (cgraph_node): Turn count to profile_count.
23539 (cgraph_edge0: Likewise.
23540 (make_speculative, clone): Update.
23541 (create_edge): Update.
23542 (init_lowered_empty_function): Update.
23543 * cgraphclones.c (cgraph_edge::clone): Update.
23544 (duplicate_thunk_for_node): Update.
23545 (cgraph_node::create_clone): Update.
23546 * cgraphunit.c (cgraph_node::analyze): Update.
23547 (cgraph_node::expand_thunk): Update.
23548 * final.c (dump_basic_block_info): Update.
23549 * gimple-streamer-in.c (input_bb): Update.
23550 * gimple-streamer-out.c (output_bb): Update.
23551 * graphite.c (print_global_statistics): Update.
23552 (print_graphite_scop_statistics): Update.
23553 * hsa-brig.c: Include basic-block.h.
23554 * hsa-dump.c: Include basic-block.h.
23555 * hsa-gen.c (T sum_slice): Update.
23556 (convert_switch_statements):Update.
23557 * hsa-regalloc.c: Include basic-block.h.
23558 * ipa-chkp.c (chkp_produce_thunks): Update.
23559 * ipa-cp.c (struct caller_statistics): Update.
23560 (init_caller_stats): Update.
23561 (gather_caller_stats): Update.
23562 (ipcp_cloning_candidate_p): Update.
23563 (good_cloning_opportunity_p): Update.
23564 (get_info_about_necessary_edges): Update.
23565 (dump_profile_updates): Update.
23566 (update_profiling_info): Update.
23567 (update_specialized_profile): Update.
23568 (perhaps_add_new_callers): Update.
23569 (decide_about_value): Update.
23570 (ipa_cp_c_finalize): Update.
23571 * ipa-devirt.c (struct odr_type_warn_count): Update.
23572 (struct decl_warn_count): Update.
23573 (struct final_warning_record): Update.
23574 (possible_polymorphic_call_targets): Update.
23575 (ipa_devirt): Update.
23576 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23577 * ipa-icf.c (sem_function::merge): Update.
23578 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23579 * ipa-inline.c (compute_uninlined_call_time): Update.
23580 (compute_inlined_call_time): Update.
23581 (want_inline_small_function_p): Update.
23582 (want_inline_self_recursive_call_p): Update.
23583 (edge_badness): Update.
23584 (lookup_recursive_calls): Update.
23585 (recursive_inlining): Update.
23586 (inline_small_functions): Update.
23587 (dump_overall_stats): Update.
23588 (dump_inline_stats): Update.
23589 * ipa-profile.c (ipa_profile_generate_summary): Update.
23590 (ipa_propagate_frequency): Update.
23591 (ipa_profile): Update.
23592 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23593 * ipa-utils.c (ipa_merge_profiles): Update.
23594 * loop-doloop.c (doloop_modify): Update.
23595 * loop-unroll.c (report_unroll): Update.
23596 (unroll_loop_runtime_iterations): Update.
23597 * lto-cgraph.c (lto_output_edge): Update.
23598 (lto_output_node): Update.
23599 (input_node): Update.
23600 (input_edge): Update.
23601 (merge_profile_summaries): Update.
23602 * lto-streamer-in.c (input_cfg): Update.
23603 * lto-streamer-out.c (output_cfg): Update.
23604 * mcf.c (create_fixup_graph): Update.
23605 (adjust_cfg_counts): Update.
23606 (sum_edge_counts): Update.
23607 * modulo-sched.c (sms_schedule): Update.
23608 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23609 * predict.c (maybe_hot_count_p): Update.
23610 (probably_never_executed): Update.
23611 (dump_prediction): Update.
23612 (combine_predictions_for_bb): Update.
23613 (propagate_freq): Update.
23614 (handle_missing_profiles): Update.
23615 (counts_to_freqs): Update.
23616 (rebuild_frequencies): Update.
23617 (force_edge_cold): Update.
23618 * predict.h: Include profile-count.h
23619 (maybe_hot_count_p, counts_to_freqs): UPdate.
23620 * print-rtl-function.c: Do not include cfg.h
23621 * print-rtl.c: Include basic-block.h
23622 * profile-count.c: New file.
23623 * profile-count.h: New file.
23624 * profile.c (is_edge_inconsistent): Update.
23625 (correct_negative_edge_counts): Update.
23626 (is_inconsistent): Update.
23627 (set_bb_counts): Update.
23628 (read_profile_edge_counts): Update.
23629 (compute_frequency_overlap): Update.
23630 (compute_branch_probabilities): Update; Initialize and deinitialize
23632 (branch_prob): Update.
23633 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23634 (edge_gcov_count): New.
23635 (bb_gcov_count): New.
23636 * shrink-wrap.c (try_shrink_wrapping): Update.
23637 * tracer.c (better_p): Update.
23638 * trans-mem.c (expand_transaction): Update.
23639 (ipa_tm_insert_irr_call): Update.
23640 (ipa_tm_insert_gettmclone_call): Update.
23641 * tree-call-cdce.c: Update.
23642 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23643 (gimple_duplicate_sese_tail): Update.
23644 (gimple_account_profile_record): Update.
23645 (execute_fixup_cfg): Update.
23646 * tree-inline.c (copy_bb): Update.
23647 (copy_edges_for_bb): Update.
23648 (initialize_cfun): Update.
23649 (freqs_to_counts): Update.
23650 (copy_cfg_body): Update.
23651 (expand_call_inline): Update.
23652 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23653 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23654 (try_unroll_loop_completely): Update.
23655 (try_peel_loop): Update.
23656 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23657 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23658 * tree-ssa-loop-split.c (connect_loops): Update.
23659 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23660 * tree-ssa-reassoc.c (branch_fixup): Update.
23661 * tree-ssa-tail-merge.c (replace_block_by): Update.
23662 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23663 (compute_path_counts): Update.
23664 (update_profile): Update.
23665 (recompute_probabilities): Update.
23666 (update_joiner_offpath_counts): Update.
23667 (estimated_freqs_path): Update.
23668 (freqs_to_counts_path): Update.
23669 (clear_counts_path): Update.
23670 (ssa_fix_duplicate_block_edges): Update.
23671 (duplicate_thread_path): Update.
23672 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23673 (struct switch_conv_info): Update.
23674 * tree-tailcall.c (decrease_profile): Update.
23675 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23676 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23677 * value-prof.c (check_counter): Update.
23678 (gimple_divmod_fixed_value): Update.
23679 (gimple_mod_pow2): Update.
23680 (gimple_mod_subtract): Update.
23681 (gimple_ic_transform): Update.
23682 (gimple_stringop_fixed_value): Update.
23683 * value-prof.h (gimple_ic): Update.
23685 2017-06-02 Carl Love <cel@us.ibm.com>
23687 * config/rs6000/rs6000-c: Add support for built-in functions
23688 vector double vec_doublee (vector signed int);
23689 vector double vec_doublee (vector unsigned int);
23690 vector double vec_doublee (vector float);
23691 vector double vec_doubleh (vector signed int);
23692 vector double vec_doubleh (vector unsigned int);
23693 vector double vec_doubleh (vector float);
23694 vector double vec_doublel (vector signed int);
23695 vector double vec_doublel (vector unsigned int);
23696 vector double vec_doublel (vector float);
23697 vector double vec_doubleo (vector signed int);
23698 vector double vec_doubleo (vector unsigned int);
23699 vector double vec_doubleo (vector float);.
23700 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23701 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23703 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23704 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23705 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23707 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23708 vec_doublel, vec_doubleh.
23709 * doc/extend.texi: Update the built-in documentation file for the
23710 new built-in functions.
23712 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23715 * ipa-inline-analysis.c (free_growth_caches): Set
23716 edge_removal_hook_holder to NULL after removing it.
23718 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23720 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23721 comparision with zero.
23723 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23725 for early expansion of vec_min and vec_max builtins.
23726 (builtin_function_type): Add min/max unsigned variants to those
23727 identified as having unsigned arguments.
23729 2017-06-02 Olivier Hainque <hainque@adacore.com>
23731 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23733 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23735 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23736 Use VALL_F16 iterator rather than VALL.
23738 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23740 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23741 Emit CBNZ inside loop when doing a strong exchange and comparing
23742 against zero. Generate the CC flags after the loop.
23744 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23746 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23747 (dl_section_ref): New.
23748 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23749 On AIX, append an expression to subtract the size of the
23750 section length to dl_section_ref.
23752 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23755 for early expansion of vector absolute builtins.
23757 2017-06-02 Richard Biener <rguenther@suse.de>
23759 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23760 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23762 2017-06-02 Richard Biener <rguenther@suse.de>
23764 PR tree-optimization/80948
23765 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23766 stmt order as well.
23768 2017-06-02 Richard Biener <rguenther@suse.de>
23770 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23772 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23773 for inductions as relevant.
23775 2017-06-02 Richard Biener <rguenther@suse.de>
23777 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23778 (vectorizable_induction): ... this. Remove dead code.
23780 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23782 * builtins. (expand_builtin_alloca): Remove second parameter and
23783 infer its value from the first parameter instead.
23784 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23786 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23788 PR rtl-optimization/80903
23789 * loop-doloop.c (add_test): Unshare sequence.
23791 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23793 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23795 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23797 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23799 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23800 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23801 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23802 (xlogue_layout::get_stub_rtx): Make static.
23803 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23804 (xlogue_layout::compute_stub_managed_regs): Rename to...
23805 (xlogue_layout::count_stub_managed_regs): ...this.
23806 (xlogue_layout::is_stub_managed_reg): New function.
23807 (xlogue_layout::m_stub_names): Rename to...
23808 (xlogue_layout::s_stub_names): ...this, make static.
23809 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23810 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23811 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23812 xlogue_layout::s_stub_names): Instantiate statics.
23813 (stub_managed_regs): Remove.
23814 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23815 (disable_call_ms2sysv_xlogues): Rename to...
23816 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23817 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23819 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23820 change after reload_completed.
23821 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23823 (ix86_expand_prologue): Likewise.
23824 (ix86_expand_epilogue): Likewise.
23825 (ix86_expand_split_stack_prologue): Likewise.
23826 (ix86_compute_frame_layout): Remove frame parameter ...
23827 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23828 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23830 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23831 (ix86_frame): Move from here ...
23832 * config/i386/i386.h (ix86_frame): ... to here.
23833 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23834 complete ix86_frame data structure instead. Remove some_ld_name.
23836 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23838 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23839 symbols that hold a DECL_VALUE_EXPR.
23841 2017-06-01 Martin Jambor <mjambor@suse.cz>
23843 PR tree-optimization/80898
23844 * tree-sra.c (process_subtree_disqualification): Removed.
23845 (disqualify_candidate): Do not acll
23846 process_subtree_disqualification.
23847 (subtree_mark_written_and_enqueue): New function.
23848 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23849 RHS has been disqualified and re-queue LHS if necessary. Apart
23850 from that, ignore disqualified RHS.
23852 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23854 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23855 address fetch for z10 or later.
23857 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23859 * config/arc/arc.md (tst_movb): Add guard when splitting.
23861 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23863 * config/arc/arc.c (arc_can_eliminate): Test against
23864 arc_frame_pointer_needed.
23866 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23868 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23869 to prevent store reordering.
23870 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23871 (type): Add block type.
23872 (stack_tie): Define special instruction to be used in
23875 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23877 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23878 constraint. It is not valid for the pattern.
23879 (noncommutative_binary_comparison): Likewise.
23881 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23883 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23886 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23888 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23889 be used by the reg-alloc.
23891 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23893 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23894 reg-alloc when having mul64 or mul32x16 instructions.
23895 (mulsidi3): Likewise.
23896 (umulsidi3): Likewise.
23897 (mulsi32x16): New pattern.
23898 (mulsi64): Likewise.
23899 (mulsidi64): Likewise.
23900 (umulsidi64): Likewise.
23901 (MUL32x16_REG): Define.
23902 (mul64_600): Use MUL32x16_REG.
23903 (mac64_600): Likewise.
23904 (umul64_600): Likewise.
23905 (umac64_600): Likewise.
23907 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23909 * config/arc/arc.md (mulsi3_700): Make it commutative.
23911 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23913 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23915 (*sign_extendsidi2_insn): Likewise for movstosw.
23917 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23919 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23920 the type of the input discriminant value. Convert the
23921 discriminant value of signedness vary.
23923 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23925 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23926 Add to -Wall section.
23928 2017-06-01 Richard Biener <rguenther@suse.de>
23930 PR middle-end/66313
23931 * fold-const.c (fold_plusminus_mult_expr): If the factored
23932 factor may be zero use a wrapping type for the inner operation.
23933 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23934 and handle moved defs.
23935 (process_assignment): Properly guard the unary op case. Return a
23936 tri-state indicating that moving the stmt before the call may allow
23937 to continue. Pass through to_move.
23938 (find_tail_calls): Handle moving unrelated defs before
23941 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23944 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23945 splitter result in the canonical way.
23947 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23949 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23950 also for 32bit target. Update insn attributes.
23951 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23953 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23955 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23956 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23957 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23958 (_mm_maskz_min_ss): New intrinsics.
23960 2017-05-31 Martin Liska <mliska@suse.cz>
23962 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23963 Change comment style to one we normally use.
23964 (vectorizable_reduction): Likewise.
23965 (vectorizable_induction): Likewise.
23966 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23967 (vectorizable_call): Likewise.
23968 (vectorizable_simd_clone_call): Likewise.
23969 (vectorizable_conversion): Likewise.
23970 (vectorizable_assignment): Likewise.
23971 (vectorizable_shift): Likewise.
23972 (vectorizable_operation): Likewise.
23973 (vectorizable_store): Likewise.
23974 (vectorizable_load): Likewise.
23975 * tree-vectorizer.h: Likewise.
23977 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23979 * passes.c (emergency_dump_function): New.
23980 * tree-pass.h (emergency_dump_function): Declare.
23981 * plugin.c (plugins_internal_error_function): Remove.
23982 * plugin.h (plugins_internal_error_function): Remove declaration.
23983 * toplev.c (internal_error_function): New static function. Use it...
23984 (general_init): ...here.
23986 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23988 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23989 (arc_rtx_costs): Add costs for new patterns.
23990 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23991 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23993 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23995 * tree-ssa-strlen.c (get_next_strinfo): New function.
23996 (get_stridx_plus_constant): Use it.
23997 (zero_length_string): Likewise.
23998 (adjust_related_strinfos): Likewise.
23999 (adjust_last_stmt): Likewise.
24001 2017-05-31 Richard Biener <rguenther@suse.de>
24004 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24005 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24007 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24009 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24010 loop_vinfo argument and use of dependence distance vectors.
24011 Check instead whether the two references differ only in their
24012 initial value and assume that they have the same alignment if the
24013 difference is a multiple of the vector alignment.
24014 (vect_analyze_data_refs_alignment): Update call accordingly.
24016 2017-05-31 Martin Liska <mliska@suse.cz>
24019 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24021 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24023 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24024 (create_intersect_range_checks): Move from ...
24025 * tree-data-ref.c (create_intersect_range_checks_index)
24026 (create_intersect_range_checks): ... to here.
24027 (create_runtime_alias_checks): New function factored from ...
24028 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24029 here. Call above function.
24030 * tree-data-ref.h (create_runtime_alias_checks): New function.
24032 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24034 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24035 segment length for dr_b and compute it in wide_int.
24037 2017-05-31 Richard Biener <rguenther@suse.de>
24039 PR tree-optimization/80906
24040 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24041 and pass through iv_map.
24042 (copy_bb_and_scalar_dependences): Adjust.
24043 (translate_pending_phi_nodes): Likewise.
24044 (copy_loop_close_phi_args): Handle code-generating IVs instead
24047 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24049 * diagnostic-color.c (color_dict): Add "type-diff".
24050 (parse_gcc_colors): Update comment.
24051 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24052 -fdiagnostics-show-template-tree and -fno-elide-type.
24053 (GCC_COLORS): Add type-diff to example.
24055 (-fdiagnostics-show-template-tree): New.
24056 (-fno-elide-type): New.
24057 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24058 the pp_format_decoder callback. Call any m_format_postprocessor's
24060 (pretty_printer::pretty_printer): Initialize
24061 m_format_postprocessor.
24062 (pretty_printer::~pretty_printer): Delete any
24063 m_format_postprocessor.
24064 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24065 (class format_postprocessor): New class.
24066 (struct pretty_printer::format_decoder): Document the new parameters.
24067 (struct pretty_printer::m_format_postprocessor): New field.
24068 * tree-diagnostic.c (default_tree_printer): Update for new
24069 bool and const char ** params.
24070 * tree-diagnostic.h (default_tree_printer): Likewise.
24072 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24074 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24075 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24076 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24077 rs6000_gen_cell_microcode code.
24078 (rs6000_final_prescan_insn): Delete.
24079 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24080 "warn-cell-microcode" entries.
24081 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24082 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24083 throughout. Change cc_reg_not_micro_cr0_operand to
24084 cc_reg_not_cr0_operand throughout.
24085 (*extendhi<mode>2_noload): Delete.
24086 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24087 (mwarn-cell-microcode): Delete.
24088 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24089 -mgen-cell-microcode and -mwarn-cell-microcode.
24091 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24094 * config/i386/constraints.md (Yd): New constraint.
24096 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24097 and (?Yd, r) alternatives. Update insn attributes.
24098 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24099 and (?*Yd, r) alternatives. Update insn attributes.
24100 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24102 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24104 * gimplify.c (gimplify_modify_expr): Don't create a
24105 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24108 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24110 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24112 2017-05-30 Richard Biener <rguenther@suse.de>
24114 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24115 and reduc_def fields.
24116 (STMT_VINFO_REDUC_TYPE): New define.
24117 (STMT_VINFO_REDUC_DEF): Likewise.
24118 (vect_force_simple_reduction): Adjust prototype.
24119 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24120 (vect_is_simple_reduction): Remove check_reduction argument.
24121 (vect_force_simple_reduction): Adjust and set
24122 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24123 (vectorizable_reduction): Do not re-do reduction analysis
24124 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24125 * tree-parloops.c (gather_scalar_reductions): Adjust.
24127 2017-05-30 Richard Biener <rguenther@suse.de>
24129 PR middle-end/80901
24130 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24133 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24135 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24136 Introduce unknown_misalignment parameter and remove vf.
24137 (vect_peeling_hash_get_lowest_cost):
24138 Pass unknown_misalignment parameter.
24139 (vect_enhance_data_refs_alignment):
24140 Fix unsupportable data ref treatment.
24142 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24144 * tree-vect-data-refs.c (vect_get_data_access_cost):
24145 Workaround for SLP handling.
24146 (vect_enhance_data_refs_alignment):
24147 Compute costs for doing no peeling at all, compare to the best
24148 peeling costs so far and avoid peeling if cheaper.
24150 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24152 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24153 Return peeling info and set costs to zero for unlimited cost
24155 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24156 with unknown misalignment. Compute and costs for unknown
24157 misalignment, compare them to the costs for known misalignment
24158 and choose the cheapest for peeling.
24160 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24162 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24163 (vect_get_peeling_costs_all_drs): Create function.
24164 (vect_peeling_hash_get_lowest_cost):
24165 Use vect_get_peeling_costs_all_drs.
24166 (vect_peeling_supportable): Create function.
24167 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24169 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24171 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24172 DR_HAS_NEGATIVE_STEP.
24173 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24174 (vect_enhance_data_refs_alignment): Use.
24175 (vect_duplicate_ssa_name_ptr_info): Use.
24176 * tree-vectorizer.h (dr_misalignment): Use.
24177 (known_alignment_for_access_p): Use.
24179 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24182 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24184 (has_section_name): New function.
24186 2017-05-30 Martin Liska <mliska@suse.cz>
24189 * auto-profile.c (get_function_decl_from_block): Fix
24192 2017-05-30 Richard Biener <rguenther@suse.de>
24194 PR middle-end/80876
24195 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24197 2017-05-30 Martin Liska <mliska@suse.cz>
24199 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24200 * dumpfile.h (struct dump_file_info): Remove ctors.
24202 2017-05-30 Martin Liska <mliska@suse.cz>
24204 * predict.def: Fix GNU coding style.
24206 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24208 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24209 Mark 'to' argument with ATTRIBUTE_UNUSED.
24211 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24213 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24214 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24215 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24218 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24220 * doc/install.texi (Options specification): Restore entry of
24221 --enable-sjlj-exceptions.
24223 2017-05-27 Michael Eager <eager@eagercon.com>
24226 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24228 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24230 * config/microblaze/microblaze.h
24231 (FIXED_REGISTERS): Update in macro.
24232 (CALL_USED_REGISTERS): Update in macro.
24234 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24236 * doc/install.texi: Add links to macOS binary distributions.
24238 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24242 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24244 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24246 2017-05-26 Martin Liska <mliska@suse.cz>
24248 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24250 2017-05-26 Martin Liska <mliska@suse.cz>
24252 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24253 always leading ';; '.
24254 (dump_bb_info): Likewise.
24255 (brief_dump_cfg): Likewise.
24256 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24257 * dumpfile.c: Remove usage of TDF_VERBOSE.
24258 * dumpfile.h (enum dump_kind): Likewise.
24259 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24260 * print-tree.c (debug_verbose): Remove.
24261 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24262 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24263 * tree-diagnostic.c (default_tree_printer): Replace
24264 TDF_DIAGNOSTIC with TDF_SLIM.
24266 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24268 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24269 in parameter loop, rather than loop_vinfo.
24270 (create_intersect_range_checks): Ditto.
24271 (vect_create_cond_for_alias_checks): Update call to above functions.
24273 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24275 PR tree-optimization/80815
24276 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24277 for merging runtime alias checks. Handle negative DR_STEPs.
24279 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24281 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24283 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24284 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24285 out code pruning runtime alias checks.
24286 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24287 factored out from above.
24288 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24290 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24292 (prune_runtime_alias_test_list): New decalaration.
24294 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24296 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24297 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24298 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24299 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24300 (operator==, comp_dr_with_seg_len_pair): Ditto.
24301 (vect_prune_runtime_alias_test_list): Ditto.
24303 2017-05-26 Martin Liska <mliska@suse.cz>
24306 * params.def: Bound partial-inlining-entry-probability param.
24308 2017-05-26 Marek Polacek <polacek@redhat.com>
24311 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24314 2017-05-26 Richard Biener <rguenther@suse.de>
24316 PR tree-optimization/80842
24317 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24320 2017-05-26 Richard Biener <rguenther@suse.de>
24322 PR tree-optimization/80844
24323 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24325 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24327 * doc/md.texi (Machine Constraints): Update x86 family
24328 machine constraints section to match 'config/i386/constraints.md'.
24330 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24332 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24334 2017-05-25 Nathan Sidwell <nathan@acm.org>
24336 * doc/invoke.texi (--enable-languages): Update documentation.
24338 2017-05-25 Martin Liska <mliska@suse.cz>
24340 * dumpfile.c: Add TDF_FOLDING.
24341 * dumpfile.h (enum dump_kind): Likewise.
24342 * genmatch.c (dt_simplify::gen_1): Use it.
24344 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24346 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24348 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24350 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24351 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24353 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24355 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24356 * match.pd (X == C): Rewrite it here.
24357 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24358 with_certain_nonzero_bits2): New predicates.
24359 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24361 2017-05-24 Nathan Sidwell <nathan@acm.org>
24363 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24366 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24369 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24371 * config/powerpcspe: New port. Files are copied from the rs6000
24372 port, with "rs6000" in filenames replaced by "powerpcspe".
24374 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24376 PR rtl-optimization/80754
24377 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24379 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24381 * config/sparc/sparc.md (length): Return the correct value for -mflat
24382 sibcalls to match output_sibcall.
24384 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24388 * config/rs6000/rs6000.c (struct machine_function): Add new field
24390 (rs6000_get_separate_components): Init that field, use it.
24391 (rs6000_components_for_bb): Use the field.
24393 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24395 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24397 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24399 PR middle-end/80823
24400 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24402 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24405 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24406 address against address_operand predicate.
24407 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24409 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24411 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24412 parameters passed indirectly.
24414 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24416 * config/i386/i386.md (*movdi_internal): Remove SSE4
24417 alternative 18 (?r, *v). Update insn attributes.
24418 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24419 Update insn attributes.
24420 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24421 Update insn attributes.
24422 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24423 alternative 1 (r, v). Remove isa attribute.
24424 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24425 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24426 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24428 2017-05-23 Tom de Vries <tom@codesourcery.com>
24430 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24433 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24435 * cgraphunit.c (symbol_table::process_new_functions): Update.
24436 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24437 (inline_generate_summary): Rename to ...
24438 (ipa_fn_summary_generate): ... this one.
24439 (inline_read_summary): Rename to ...
24440 (ipa_fn_summary_read): ... this one.
24441 (inline_write_summary): Rename to ...
24442 (ipa_fn_summary_write): ... this one.
24443 (inline_free_summary): Rename to ...
24444 (ipa_free_fn_summary): ... this one.
24445 (pass_data_local_fn_summary, pass_local_fn_summary,
24446 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24447 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24448 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24449 make_pass_ipa_fn_summary): New.
24450 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24451 inline_write_summary, inline_free_summary): Remove.
24452 (ipa_free_fn_summary) : New.
24453 * ipa-inline.c (ipa_inline): Update.
24454 (pass_ipa_inline): Do not generate summaries.
24455 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24457 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
24458 and add pass_ipa_fn_summary.
24459 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
24461 (make_pass_inline_parameters): Remove.
24463 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
24465 * omp-low.c (struct omp_context): Remove "default_kind" member.
24468 * omp-offload.c (execute_oacc_device_lower): Remove the
24469 parallelism dimensions function attributes for unparallelized
24470 OpenACC kernels constructs.
24472 2017-05-23 Martin Liska <mliska@suse.cz>
24474 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
24476 (cgraph_edge::make_speculative): Likewise.
24477 (cgraph_edge::resolve_speculation): Likewise.
24478 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24479 (cgraph_node::dump): Likewise.
24480 * cgraph.h: Likewise.
24481 * cgraphunit.c (analyze_functions): Likewise.
24482 (symbol_table::compile): Likewise.
24483 * ipa-cp.c (print_all_lattices): Likewise.
24484 (determine_versionability): Likewise.
24485 (initialize_node_lattices): Likewise.
24486 (ipcp_verify_propagated_values): Likewise.
24487 (estimate_local_effects): Likewise.
24488 (update_profiling_info): Likewise.
24489 (create_specialized_node): Likewise.
24490 (perhaps_add_new_callers): Likewise.
24491 (decide_about_value): Likewise.
24492 (decide_whether_version_node): Likewise.
24493 (identify_dead_nodes): Likewise.
24494 (ipcp_store_bits_results): Likewise.
24495 * ipa-devirt.c (dump_targets): Likewise.
24496 (ipa_devirt): Likewise.
24497 * ipa-icf.c (sem_item::dump): Likewise.
24498 (sem_function::equals): Likewise.
24499 (sem_variable::equals): Likewise.
24500 (sem_item_optimizer::read_section): Likewise.
24501 (sem_item_optimizer::execute): Likewise.
24502 (congruence_class::dump): Likewise.
24503 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24504 (dump_inline_summary): Likewise.
24505 (estimate_node_size_and_time): Likewise.
24506 (inline_analyze_function): Likewise.
24507 * ipa-inline-transform.c (inline_call): Likewise.
24508 * ipa-inline.c (report_inline_failed_reason): Likewise.
24509 (want_early_inline_function_p): Likewise.
24510 (edge_badness): Likewise.
24511 (update_edge_key): Likewise.
24512 (inline_small_functions): Likewise.
24513 * ipa-profile.c (ipa_profile): Likewise.
24514 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24515 (ipa_make_edge_direct_to_target): Likewise.
24516 (remove_described_reference): Likewise.
24517 (ipa_impossible_devirt_target): Likewise.
24518 (propagate_controlled_uses): Likewise.
24519 (ipa_print_node_params): Likewise.
24520 (ipcp_transform_function): Likewise.
24521 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24522 (propagate_pure_const): Likewise.
24523 * ipa-reference.c (generate_summary): Likewise.
24524 (read_write_all_from_decl): Likewise.
24525 (propagate): Likewise.
24526 (ipa_reference_read_optimization_summary): Likewise.
24527 * ipa-utils.c (ipa_merge_profiles): Likewise.
24528 * ipa.c (walk_polymorphic_call_targets): Likewise.
24529 (symbol_table::remove_unreachable_nodes): Likewise.
24530 (ipa_single_use): Likewise.
24531 * passes.c (execute_todo): Likewise.
24532 * predict.c (drop_profile): Likewise.
24533 * symtab.c (symtab_node::get_dump_name): New function.
24534 (symtab_node::dump_name): Likewise.
24535 (symtab_node::dump_asm_name): Likewise.
24536 (symtab_node::dump_references): Likewise.
24537 (symtab_node::dump_referring): Likewise.
24538 (symtab_node::dump_base): Likewise.
24539 (symtab_node::debug_symtab): Likewise.
24540 * tree-sra.c (convert_callers_for_node): Likewise.
24541 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24542 * value-prof.c (init_node_map): Likewise.
24544 2017-05-23 Martin Liska <mliska@suse.cz>
24546 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24547 and symtab_node::debug_symtab to symbol_table::debug.
24548 * cgraphunit.c (analyze_functions): Use the renamed function.
24549 (symbol_table::compile): Likewise.
24550 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24551 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24552 * passes.c (execute_todo): Likewise.
24553 * symtab.c (symbol_table::dump): New function.
24554 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24556 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24558 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24559 that nonconst implies exec.
24561 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24563 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24564 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24565 (inline_edge_summary_vec): Turn into ...
24566 (ipa_call_summaries): ... this one.
24567 (redirect_to_unreachable, edge_set_predicate,
24568 evaluate_properties_for_edge, inline_summary_alloc,
24569 reset_ipa_call_summary, reset_inline_summary,
24570 inline_summary_t::duplicate): Update.
24571 (inline_edge_duplication_hook): Turn to ...
24572 (ipa_call_summary_t::duplicate): ... this one.
24573 (inline_edge_removal_hook): Turn to ...
24574 (ipa_call_summary_t::remove): ... this one.
24575 (dump_inline_edge_summary): Turn to ...
24576 (dump_ipa_call_summary): ... this one.
24577 (estimate_function_body_sizes): Update.
24578 (inline_update_callee_summaries): Update.
24579 (remap_edge_change_prob): Update.
24580 (remap_edge_summaries): Update.
24581 (inline_merge_summary): Update.
24582 (do_estimate_edge_time): Update.
24583 (inline_generate_summary): Update.
24584 (inline_read_section): Update.
24585 (inline_read_summary): Update.
24586 (inline_free_summary): Update.
24587 * ipa-inline.c (can_inline_edge_p): Update.
24588 (compute_inlined_call_time): Update.
24589 (want_inline_small_function_p): Update.
24590 (edge_badness): Update.
24591 (early_inliner): Update.
24592 * ipa-inline.h (inline_edge_summary): Turn to ...
24593 (ipa_call_summary): ... this one.
24594 (ipa_call_summary_t): New class.
24595 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24596 (ipa_call_summaries): New.
24597 (inline_edge_summary): Remove.
24598 (estimate_edge_growth): Update.
24599 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24600 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24601 * ipa-split.c (execute_split_functions): Update.
24602 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24604 2017-05-23 Tom de Vries <tom@codesourcery.com>
24606 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24607 attributes): Document rdrand effective target.
24609 2017-05-23 Tom de Vries <tom@codesourcery.com>
24611 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24612 attributes): Sort alphabetically.
24614 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24616 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24618 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24621 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24622 V2DF/V2DI splat into two separate patterns, one that handles
24623 registers, and the other that only handles memory. Drop support
24624 for splatting from a GPR on ISA 2.07 and then splitting the
24625 splat into direct move and splat.
24626 (vsx_splat_<mode>_reg): Likewise.
24627 (vsx_splat_<mode>_mem): Likewise.
24629 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24631 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24633 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24635 PR middle-end/80809
24636 * omp-low.c (finish_taskreg_remap): New function.
24637 (finish_taskreg_scan): If unit size of ctx->record_type
24638 is non-constant, unshare the size expression and replace
24639 decls in it with possible outer var refs.
24641 PR middle-end/80809
24642 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24643 GOVD_SHARED rather than GOVD_PRIVATE with it.
24644 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24645 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24647 PR middle-end/80853
24648 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24649 as last argument to build_outer_var_ref for pointer bases of array
24650 section reductions.
24652 2017-05-19 Martin Sebor <msebor@redhat.com>
24654 * print-tree.c (print_node): Print DECL_READ_P flag.
24656 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24658 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24659 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24660 * cgraph.c: Likewise.
24661 * cgraphunit.c: Likewise.
24662 * gengtype.c: Likewise.
24663 * ipa-cp.c: Likewise.
24664 * ipa-devirt.c: Likewise.
24665 * ipa-icf.c: Likewise.
24666 * ipa-predicate.c: Likewise.
24667 * ipa-profile.c: Likewise.
24668 * ipa-prop.c: Likewise.
24669 * ipa-split.c: Likewise.
24671 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24672 edge_predicate_pool, dump_inline_hints,
24673 inline_summary::account_size_time, redirect_to_unreachable,
24674 edge_set_predicate, set_hint_predicate,
24675 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24676 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24677 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24678 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24679 ipa_call_summary_t::remove, initialize_growth_caches,
24680 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24681 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24682 mark_modified, unmodified_parm_1, unmodified_parm,
24683 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24684 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24685 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24686 will_be_nonconstant_predicate, record_modified_bb_info,
24687 get_minimal_bb, record_modified, param_change_prob,
24688 phi_result_unknown_predicate, predicate_for_phi_result,
24689 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24690 estimate_function_body_sizes, compute_inline_parameters,
24691 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24692 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24693 inline_update_callee_summaries, remap_edge_change_prob,
24694 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24695 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24696 inline_analyze_function, inline_summary_t::insert,
24697 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24698 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24699 inline_free_summary): Move to ipa-fnsummary.h
24700 (predicate_t): Remove.
24701 * ipa-fnsummary.c: New file.
24702 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24703 (enum inline_hints_vals, inline_hints, agg_position_info,
24704 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24705 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24706 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24707 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24708 inline_read_summary, inline_write_summary, inline_free_summary,
24709 inline_analyze_function, initialize_inline_failed,
24710 inline_merge_summary, inline_update_overall_summary,
24711 compute_inline_parameters): Move to ipa-fnsummary.h
24712 * ipa-fnsummary.h: New file.
24713 * ipa-inline-transform.h: Include ipa-inline.h.
24714 * ipa-inline.c: LIkewise.
24716 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24718 * ipa-inline.c (edge_badness): Use inlined_time instead of
24719 inline_summaries->get.
24721 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24723 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24725 2017-05-22 Nathan Sidwell <nathan@acm.org>
24727 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24728 (fdump-lang): Document 'raw' option.
24729 * dumpfile.h (TDI_tu): Delete.
24730 * dumpfile.c (dump_files): Remove translation-unit.
24731 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24733 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24735 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24736 command option from $(AWK) call.
24737 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24738 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24739 [FORMAT]: Remove handling of variable.
24740 * config/avr/t-multilib: Regenerate.
24742 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24744 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24746 (dump_inline_summary): Do not print self_time.
24747 (estimate_function_body_sizes): Do not set self_time.
24748 (compute_inline_parameters): Likewise.
24749 (inline_read_section, inline_write_summary): Do not stream self_time.
24750 * ipa-inline.h (inline_summary): Drop self_time.
24752 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24754 * ipa-inline-analysis.c (account_size_time): Rename to ...
24755 (inline_summary::account_size_time): ... this one.
24756 (reset_ipa_call_summary): Turn to ...
24757 (ipa_call_summary::reset): ... this one.
24758 (reset_inline_summary): Turn to ...
24759 (inline_summary::reset): ... this one.
24760 (inline_summary_t::remove): Update.
24761 (inline_summary_t::duplicate): Update.
24762 (ipa_call_summary_t::remove): Update.
24763 (dump_inline_summary): Update.
24764 (estimate_function_body_sizes): Update.
24765 (compute_inline_parameters): Update.
24766 (estimate_node_size_and_time): Update.
24767 (inline_merge_summary): Update.
24768 (inline_update_overall_summary): Update.
24769 (inline_read_section): Update.
24770 (inline_write_summary): Update.
24771 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24772 add account_size_time and reset member functions.
24773 (ipa_call_summary): Add reset function.
24774 * ipa-predicate.h (predicate::operator &): Constify.
24776 2017-05-22 Richard Biener <rguenther@suse.de>
24778 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24780 2017-05-19 Jason Merrill <jason@redhat.com>
24782 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24784 2017-05-19 Marek Polacek <polacek@redhat.com>
24787 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24788 TYPE_OVERFLOW_WRAPS checks.
24790 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24792 * tree-core.h (enum omp_clause_default_kind): Add
24793 "OMP_CLAUSE_DEFAULT_PRESENT".
24794 * tree-pretty-print.c (dump_omp_clause): Handle it.
24795 * gimplify.c (enum gimplify_omp_var_data): Add
24796 "GOVD_MAP_FORCE_PRESENT".
24797 (gimplify_adjust_omp_clauses_1): Map it to
24798 "GOMP_MAP_FORCE_PRESENT".
24799 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24801 * gimplify.c (oacc_default_clause): Clarify.
24803 2017-05-19 Nathan Sidwell <nathan@acm.org>
24805 LANG_HOOK_REGISTER_DUMPS
24806 * toplev.c (general_init): Call register dump lang hook.
24807 * doc/invoke.texi: Document -fdump-lang option family.
24808 * dumpfile.c (dump_files): Remove class dump here.
24809 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24810 * dumpfile.h (tree_dump_index): Remove TDI_class.
24811 * langhooks-def.h (lhd_register_dumps): Declare.
24812 (LANG_HOOKS_REGISTER_DUMPS): Define.
24813 (LANG_HOOKS_INITIALIZER): Add it.
24814 * langhooks.c (lhd_register_dumps): Define.
24815 * langhooks.h (struct lang_hooks): Add register_dumps.
24817 2017-05-19 Nathan Sidwell <nathan@acm.org>
24819 * context.h (context::set_passes): New.
24820 * context.c (context::context): Do not create pass manager.
24821 * toplev.c (general_init): Create pass manager here.
24823 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24825 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24826 use this splitter if two add or or instructions would also work for
24827 the constant we want to generate.
24829 2017-05-19 Richard Biener <rguenther@suse.de>
24832 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24833 predicate evaluation.
24835 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24837 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24839 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24842 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24844 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24845 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24846 (inline_edge_summary_vec): Turn into ...
24847 (ipa_call_summaries): ... this one.
24848 (redirect_to_unreachable, edge_set_predicate,
24849 evaluate_properties_for_edge, inline_summary_alloc,
24850 reset_ipa_call_summary, reset_inline_summary,
24851 inline_summary_t::duplicate): Update.
24852 (inline_edge_duplication_hook): Turn to ...
24853 (ipa_call_summary_t::duplicate): ... this one.
24854 (inline_edge_removal_hook): Turn to ...
24855 (ipa_call_summary_t::remove): ... this one.
24856 (dump_inline_edge_summary): Turn to ...
24857 (dump_ipa_call_summary): ... this one.
24858 (estimate_function_body_sizes): Update.
24859 (inline_update_callee_summaries): Update.
24860 (remap_edge_change_prob): Update.
24861 (remap_edge_summaries): Update.
24862 (inline_merge_summary): Update.
24863 (do_estimate_edge_time): Update.
24864 (inline_generate_summary): Update.
24865 (inline_read_section): Update.
24866 (inline_read_summary): Update.
24867 (inline_free_summary): Update.
24868 * ipa-inline.c (can_inline_edge_p): Update.
24869 (compute_inlined_call_time): Update.
24870 (want_inline_small_function_p): Update.
24871 (edge_badness): Update.
24872 (early_inliner): Update.
24873 * ipa-inline.h (inline_edge_summary): Turn to ...
24874 (ipa_call_summary): ... this one.
24875 (ipa_call_summary_t): New class.
24876 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24877 (ipa_call_summaries): New.
24878 (inline_edge_summary): Remove.
24879 (estimate_edge_growth): Update.
24880 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24881 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24882 * ipa-split.c (execute_split_functions): Update.
24883 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24885 2017-05-19 Richard Biener <rguenther@suse.de>
24887 PR middle-end/80764
24888 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24890 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24892 * config/rs6000/rs6000.c (struct machine_function): Add field
24893 fpr_is_wrapped_separately.
24894 (rs6000_get_separate_components): Use 64 components. Handle the
24895 new FPR components.
24896 (rs6000_components_for_bb): Handle the FPR components.
24897 (rs6000_emit_prologue_components): Handle the FPR components.
24898 (rs6000_emit_epilogue_components): Handle the FPR components.
24899 (rs6000_set_handled_components): Handle the FPR components.
24900 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24901 that are already separately shrink-wrapped.
24902 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24903 that are already separately shrink-wrapped.
24905 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24908 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24911 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24912 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24913 cases where the register allocator uses a move and an offsettable
24914 memory operation to/from a FPR register on ISA 2.06/2.07.
24915 (define_peephole2 for Altivec d-form store): Likewise.
24917 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24920 * config/i386/mmx.md (*mov<mode>_internal): Enable
24921 alternatives 11, 12, 13 and 14 also for 32bit targets.
24922 Remove alternatives 15, 16, 17 and 18.
24923 * config/i386/sse.md (vec_concatv2di): Change
24924 alternative (!x, *y) to (x, ?!*Yn).
24926 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24928 * dumpfile.h (enum dump_kind): Remove stray comma.
24930 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24932 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24933 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24934 predicate::num_conditions
24935 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24936 (CHANGED): turn into predicate::changed.
24937 (agg_position_info): Move to ipa-predicate.h
24938 (add_condition, predicate::add_clause, predicate::operator &=,
24939 predicate::or_with, predicate::evaluate, predicate::probability,
24940 dump_condition, dump_clause, predicate::dump,
24941 predicate::remap_after_duplication, predicate::remap_after_inlining,
24942 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24943 (evaluate_conditions_for_known_args): Update.
24944 (set_cond_stmt_execution_predicate): Update.
24945 * ipa-inline.h: Include ipa-predicate.h
24946 (condition, inline_param_summary, conditions, agg_position_info,
24947 predicate): Move to ipa-predicate.h
24948 * ipa-predicate.c: New file.
24949 * ipa-predicate.h: New file.
24951 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24953 * final.c (leaf_function_p): Check we are not in a sequence.
24955 2017-05-18 Martin Liska <mliska@suse.cz>
24957 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24958 * dumpfile.c (dump_register): Use new enum dump_kind.
24959 (get_dump_file_name): Likewise.
24960 (dump_enable_all): Likewise.
24961 (dump_switch_p_1): Likewise.
24962 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24963 * dumpfile.h (enum dump_kind): New enum type.
24964 (struct dump_file_info): Create constructor and
24965 format fields and comments.
24966 * passes.c (pass_manager::register_one_dump_file):
24968 * statistics.c (statistics_early_init): Likewise.
24969 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24970 TDF_TREE with TDF_SLIM.
24971 (gather_memory_references_ref): Likewise.
24973 2017-05-18 Martin Liska <mliska@suse.cz>
24975 * vec.h (struct vnull): Use it.
24977 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24979 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24980 (true_predicate, false_predicate, true_predicate_p,
24981 false_predicate_p): Remove.
24982 (single_cond_predicate, not_inlined_predicate): Turn to member function
24984 (add_condition): Update.
24985 (add_clause): Turn to...
24986 (predicate::add_clause): ... this one; update; allow passing NULL
24988 (and_predicates): Turn to ...
24989 (predicate::operator &=): ... this one.
24990 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24991 (or_predicates): Turn to ...
24992 (predicate::or_with): ... this one.
24993 (evaluate_predicate): Turn to ...
24994 (predicate::evaluate): ... this one.
24995 (predicate_probability): Turn to ...
24996 (predicate::probability): ... this one.
24997 (dump_condition): Update.
24998 (dump_predicate): Turn to ...
24999 (predicate::dump): ... this one.
25000 (account_size_time): Update.
25001 (edge_set_predicate): Update.
25002 (set_hint_predicate): UPdate.
25003 (evaluate_conditions_for_known_args): Update.
25004 (evaluate_properties_for_edge): Update.
25005 (remap_predicate_after_duplication): Turn to...
25006 (predicate::remap_after_duplication): ... this one.
25007 (remap_hint_predicate_after_duplication): Update.
25008 (inline_summary_t::duplicate): UPdate.
25009 (dump_inline_edge_summary): Update.
25010 (dump_inline_summary): Update.
25011 (set_cond_stmt_execution_predicate): Update.
25012 (set_switch_stmt_execution_predicate): Update.
25013 (compute_bb_predicates): Update.
25014 (will_be_nonconstant_expr_predicate): Update.
25015 (will_be_nonconstant_predicate): Update.
25016 (phi_result_unknown_predicate): Update.
25017 (predicate_for_phi_result): Update.
25018 (array_index_predicate): Update.
25019 (estimate_function_body_sizes): Update.
25020 (estimate_node_size_and_time): Update.
25021 (estimate_ipcp_clone_size_and_time): Update.
25022 (remap_predicate): Rename to ...
25023 (predicate::remap_after_inlining): ... this one.
25024 (remap_hint_predicate): Update.
25025 (inline_merge_summary): Update.
25026 (inline_update_overall_summary): Update.
25027 (estimate_size_after_inlining): Update.
25028 (read_predicate): Rename to ...
25029 (predicate::stream_in): ... this one.
25030 (read_inline_edge_summary): Update.
25031 (write_predicate): Rename to ...
25032 (predicate::stream_out): ... this one.
25033 (write_inline_edge_summary): Update.
25034 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25035 (clause_t): Turn to uint32_t
25036 (predicate): Turn to class; implement constructor and operators
25038 (size_time_entry): Update.
25039 (inline_summary): Update.
25040 (inline_edge_summary): Update.
25042 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25044 * fold-const.c (fold_binary_loc): Move transformation...
25045 * match.pd (C - X CMP X): ... here.
25047 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25049 * config/sparc/sparc.c (sparc_option_override): Set function
25050 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25051 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25053 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25055 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25057 2017-05-18 Marek Polacek <polacek@redhat.com>
25060 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25061 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25062 gimple_assign_load_p.
25064 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25066 PR middle-end/80692
25067 * real.c (do_compare): Give decimal_do_compare preference over
25068 comparing just the signs.
25070 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25072 * doc/md.texi (Canonicalization of Instructions): Describe the
25073 canonical form of instructions that inherently set a condition
25076 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25078 PR middle-end/80775
25079 * tree-cfg.c: Move deletion of unreachable case statements to after
25080 the merging of consecutive case labels.
25082 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25084 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25085 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25086 restoring of callee-saved registers.
25088 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25090 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25091 * config/visium/visium.c (single_set_and_flags): Likewise.
25092 * config/visium/visium.md (Substitutions): Likewise.
25094 2017-05-17 Martin Liska <mliska@suse.cz>
25096 * cfg.c: Introduce dump_flags_t type and
25097 use it instead of int type.
25099 * cfghooks.c: Likewise.
25100 * cfghooks.h (struct cfg_hooks): Likewise.
25101 * cfgrtl.c: Likewise.
25102 * cfgrtl.h: Likewise.
25103 * cgraph.c (cgraph_node::get_body): Likewise.
25104 * coretypes.h: Likewise.
25105 * domwalk.c: Likewise.
25106 * domwalk.h: Likewise.
25107 * dumpfile.c (struct dump_option_value_info): Likewise.
25108 (dump_enable_all): Likewise.
25109 (dump_switch_p_1): Likewise.
25110 (opt_info_switch_p): Likewise.
25111 * dumpfile.h (enum tree_dump_index): Likewise.
25112 (struct dump_file_info): Likewise.
25113 * genemit.c: Likewise.
25114 * generic-match-head.c: Likewise.
25115 * gengtype.c (open_base_files): Likewise.
25116 * gimple-pretty-print.c: Likewise.
25117 * gimple-pretty-print.h: Likewise.
25118 * graph.c (print_graph_cfg): Likewise.
25119 * graphite-scop-detection.c (dot_all_sese): Likewise.
25120 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25121 * loop-unroll.c (report_unroll): Likewise.
25122 * passes.c (pass_manager::register_one_dump_file): Likewise.
25123 * print-tree.c: Likewise.
25124 * statistics.c: Likewise.
25125 * tree-cfg.c: Likewise.
25126 * tree-cfg.h: Likewise.
25127 * tree-dfa.c: Likewise.
25128 * tree-dfa.h: Likewise.
25129 * tree-dump.c (dump_function): Likewise.
25130 * tree-dump.h (struct dump_info): Likewise.
25131 * tree-pretty-print.c: Likewise.
25132 * tree-pretty-print.h: Likewise.
25133 * tree-ssa-live.c: Likewise.
25134 * tree-ssa-live.h: Likewise.
25135 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25136 * tree-vect-loop.c: Likewise.
25137 * tree-vect-slp.c: Likewise.
25139 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25140 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25142 PR tree-optimization/80457
25143 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25144 of all arguments to a statement as scalar_to_vec operations.
25145 (vectorizable_call): Adjust call to vect_model_simple_cost for
25147 (vectorizable_conversion): Likewise.
25148 (vectorizable_assignment): Likewise.
25149 (vectorizable_shift): Likewise.
25150 (vectorizable_operation): Likewise.
25151 (vectorizable_comparison): Likewise.
25152 (vect_is_simple_cond): Record the def types for operands.
25153 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25154 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25155 for statement argument count.
25157 2017-05-16 Carl Love <cel@us.ibm.com>
25159 * config/rs6000/rs6000-c: Add support for built-in functions
25160 vector unsigned long long vec_bperm (vector unsigned long long,
25161 vector unsigned char)
25162 vector signed long long vec_mule (vector signed int,
25164 vector unsigned long long vec_mule (vector unsigned int,
25165 vector unsigned int)
25166 vector signed long long vec_mulo (vector signed int,
25168 vector unsigned long long vec_mulo (vector unsigned int,
25169 vector unsigned int)
25170 vector signed char vec_sldw (vector signed char,
25171 vector signed char,
25173 vector unsigned char vec_sldw (vector unsigned char,
25174 vector unsigned char,
25176 vector signed short vec_sldw (vector signed short,
25177 vector signed short,
25179 vector unsigned short vec_sldw (vector unsigned short,
25180 vector unsigned short,
25182 vector signed int vec_sldw (vector signed int,
25185 vector unsigned int vec_sldw (vector unsigned int,
25186 vector unsigned int,
25188 vector signed long long vec_sldw (vector signed long long,
25189 vector signed long long,
25191 vector unsigned long long vec_sldw (vector unsigned long long,
25192 vector unsigned long long,
25194 * config/rs6000/rs6000-c: Add support for built-in functions
25195 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25196 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25197 * doc/extend.texi: Update the built-in documentation for the
25198 new built-in functions.
25200 2017-05-16 Marek Polacek <polacek@redhat.com>
25204 * tree.c (save_expr): Don't fold the expression.
25206 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25208 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25209 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25210 and (?*y,m). Update insn attributes.
25212 2017-05-16 Martin Liska <mliska@suse.cz>
25214 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25215 flags argument of print_gimple_stmt, print_gimple_expr,
25216 print_generic_stmt and print_generic_expr.
25217 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25218 * coretypes.h: Likewise.
25219 * except.c (dump_eh_tree): Likewise.
25220 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25221 * gimple-pretty-print.h: Likewise.
25222 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25223 (backprop::push_to_worklist): Likewise.
25224 (backprop::pop_from_worklist): Likewise.
25225 (backprop::process_use): Likewise.
25226 (backprop::intersect_uses): Likewise.
25227 (note_replacement): Likewise.
25228 * gimple-ssa-store-merging.c
25229 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25230 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25231 (pass_store_merging::execute): Likewise.
25232 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25233 (ssa_base_cand_dump_callback): Likewise.
25234 (dump_incr_vec): Likewise.
25235 (replace_refs): Likewise.
25236 (replace_mult_candidate): Likewise.
25237 (create_add_on_incoming_edge): Likewise.
25238 (create_phi_basis): Likewise.
25239 (insert_initializers): Likewise.
25240 (all_phi_incrs_profitable): Likewise.
25241 (introduce_cast_before_cand): Likewise.
25242 (replace_one_candidate): Likewise.
25243 * gimplify.c (gimplify_expr): Likewise.
25244 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25245 (set_rename): Likewise.
25246 (rename_uses): Likewise.
25247 (copy_loop_phi_nodes): Likewise.
25248 (add_close_phis_to_merge_points): Likewise.
25249 (copy_loop_close_phi_args): Likewise.
25250 (copy_cond_phi_args): Likewise.
25251 (graphite_copy_stmts_from_block): Likewise.
25252 (translate_pending_phi_nodes): Likewise.
25253 * graphite-poly.c (print_pdr): Likewise.
25254 (dump_gbb_cases): Likewise.
25255 (dump_gbb_conditions): Likewise.
25256 (print_scop_params): Likewise.
25257 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25258 (build_cross_bb_scalars_use): Likewise.
25259 (gather_bbs::before_dom_children): Likewise.
25260 * hsa-dump.c (dump_hsa_immed): Likewise.
25261 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25262 (get_replacement_map): Likewise.
25263 * ipa-inline-analysis.c (dump_condition): Likewise.
25264 (estimate_function_body_sizes): Likewise.
25265 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25266 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25267 * ipa-prop.c (ipa_dump_param): Likewise.
25268 (ipa_print_node_jump_functions_for_edge): Likewise.
25269 (ipa_modify_call_arguments): Likewise.
25270 (ipa_modify_expr): Likewise.
25271 (ipa_dump_param_adjustments): Likewise.
25272 (ipa_dump_agg_replacement_values): Likewise.
25273 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25274 * ipa-pure-const.c (check_stmt): Likewise.
25275 (pass_nothrow::execute): Likewise.
25276 * ipa-split.c (execute_split_functions): Likewise.
25277 * omp-offload.c (dump_oacc_loop_part): Likewise.
25278 (dump_oacc_loop): Likewise.
25279 * trans-mem.c (tm_log_emit): Likewise.
25280 (tm_memopt_accumulate_memops): Likewise.
25281 (dump_tm_memopt_set): Likewise.
25282 (dump_tm_memopt_transform): Likewise.
25283 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25284 (print_loop): Likewise.
25285 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25286 (chkp_gather_checks_info): Likewise.
25287 (chkp_get_check_result): Likewise.
25288 (chkp_remove_check_if_pass): Likewise.
25289 (chkp_use_outer_bounds_if_possible): Likewise.
25290 (chkp_reduce_bounds_lifetime): Likewise.
25291 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25292 (chkp_mark_completed_bounds): Likewise.
25293 (chkp_register_incomplete_bounds): Likewise.
25294 (chkp_mark_invalid_bounds): Likewise.
25295 (chkp_maybe_copy_and_register_bounds): Likewise.
25296 (chkp_build_returned_bound): Likewise.
25297 (chkp_get_bound_for_parm): Likewise.
25298 (chkp_build_bndldx): Likewise.
25299 (chkp_get_bounds_by_definition): Likewise.
25300 (chkp_generate_extern_var_bounds): Likewise.
25301 (chkp_get_bounds_for_decl_addr): Likewise.
25302 * tree-chrec.c (chrec_apply): Likewise.
25303 * tree-data-ref.c (dump_data_reference): Likewise.
25304 (dump_subscript): Likewise.
25305 (dump_data_dependence_relation): Likewise.
25306 (analyze_overlapping_iterations): Likewise.
25307 * tree-inline.c (expand_call_inline): Likewise.
25308 (tree_function_versioning): Likewise.
25309 * tree-into-ssa.c (dump_defs_stack): Likewise.
25310 (dump_currdefs): Likewise.
25311 (dump_names_replaced_by): Likewise.
25312 (dump_update_ssa): Likewise.
25313 (update_ssa): Likewise.
25314 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25315 * tree-parloops.c (build_new_reduction): Likewise.
25316 (try_create_reduction_list): Likewise.
25317 (ref_conflicts_with_region): Likewise.
25318 (oacc_entry_exit_ok_1): Likewise.
25319 (oacc_entry_exit_single_gang): Likewise.
25320 * tree-pretty-print.h: Likewise.
25321 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25322 (get_scalar_evolution): Likewise.
25323 (add_to_evolution): Likewise.
25324 (get_loop_exit_condition): Likewise.
25325 (analyze_evolution_in_loop): Likewise.
25326 (analyze_initial_condition): Likewise.
25327 (analyze_scalar_evolution): Likewise.
25328 (instantiate_scev): Likewise.
25329 (number_of_latch_executions): Likewise.
25330 (gather_chrec_stats): Likewise.
25331 (final_value_replacement_loop): Likewise.
25332 (scev_const_prop): Likewise.
25333 * tree-sra.c (dump_access): Likewise.
25334 (disqualify_candidate): Likewise.
25335 (create_access): Likewise.
25336 (reject): Likewise.
25337 (maybe_add_sra_candidate): Likewise.
25338 (create_access_replacement): Likewise.
25339 (analyze_access_subtree): Likewise.
25340 (analyze_all_variable_accesses): Likewise.
25341 (sra_modify_assign): Likewise.
25342 (initialize_constant_pool_replacements): Likewise.
25343 (find_param_candidates): Likewise.
25344 (decide_one_param_reduction): Likewise.
25345 (replace_removed_params_ssa_names): Likewise.
25346 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25347 * tree-ssa-copy.c (dump_copy_of): Likewise.
25348 (copy_prop_visit_cond_stmt): Likewise.
25349 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25350 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25351 (record_equivalences_from_stmt): Likewise.
25352 * tree-ssa-dse.c (compute_trims): Likewise.
25353 (delete_dead_call): Likewise.
25354 (delete_dead_assignment): Likewise.
25355 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25356 (forward_propagate_into_cond): Likewise.
25357 (pass_forwprop::execute): Likewise.
25358 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25359 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25361 (move_computations_worker): Likewise.
25362 (execute_sm): Likewise.
25363 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25364 (remove_exits_and_undefined_stmts): Likewise.
25365 (remove_redundant_iv_tests): Likewise.
25366 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25367 (adjust_iv_update_pos): Likewise.
25368 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25369 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25370 (value_replacement): Likewise.
25371 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25372 * tree-ssa-pre.c (print_pre_expr): Likewise.
25373 (get_representative_for): Likewise.
25374 (create_expression_by_pieces): Likewise.
25375 (insert_into_preds_of_block): Likewise.
25376 (eliminate_insert): Likewise.
25377 (eliminate_dom_walker::before_dom_children): Likewise.
25378 (eliminate): Likewise.
25379 (remove_dead_inserted_code): Likewise.
25380 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25381 * tree-ssa-reassoc.c (get_rank): Likewise.
25382 (eliminate_duplicate_pair): Likewise.
25383 (eliminate_plus_minus_pair): Likewise.
25384 (eliminate_not_pairs): Likewise.
25385 (undistribute_ops_list): Likewise.
25386 (eliminate_redundant_comparison): Likewise.
25387 (update_range_test): Likewise.
25388 (optimize_range_tests_var_bound): Likewise.
25389 (optimize_vec_cond_expr): Likewise.
25390 (rewrite_expr_tree): Likewise.
25391 (rewrite_expr_tree_parallel): Likewise.
25392 (linearize_expr): Likewise.
25393 (break_up_subtract): Likewise.
25394 (linearize_expr_tree): Likewise.
25395 (attempt_builtin_powi): Likewise.
25396 (attempt_builtin_copysign): Likewise.
25397 (transform_stmt_to_copy): Likewise.
25398 (transform_stmt_to_multiply): Likewise.
25399 (dump_ops_vector): Likewise.
25400 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25401 (print_scc): Likewise.
25402 (set_ssa_val_to): Likewise.
25403 (visit_reference_op_store): Likewise.
25404 (visit_use): Likewise.
25405 (sccvn_dom_walker::before_dom_children): Likewise.
25406 (run_scc_vn): Likewise.
25407 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25409 (expr_hash_elt::print): Likewise.
25410 (const_and_copies::pop_to_marker): Likewise.
25411 (const_and_copies::record_const_or_copy_raw): Likewise.
25412 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25413 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25414 (dump_predicates): Likewise.
25415 (find_uninit_use): Likewise.
25416 (warn_uninitialized_phi): Likewise.
25417 (pass_late_warn_uninitialized::execute): Likewise.
25418 * tree-ssa.c (verify_vssa): Likewise.
25419 (verify_ssa): Likewise.
25420 (maybe_optimize_var): Likewise.
25421 * tree-vrp.c (dump_value_range): Likewise.
25422 (dump_all_value_ranges): Likewise.
25423 (dump_asserts_for): Likewise.
25424 (register_edge_assert_for_2): Likewise.
25425 (vrp_visit_cond_stmt): Likewise.
25426 (vrp_visit_switch_stmt): Likewise.
25427 (vrp_visit_stmt): Likewise.
25428 (vrp_visit_phi_node): Likewise.
25429 (simplify_cond_using_ranges_1): Likewise.
25430 (fold_predicate_in): Likewise.
25431 (evrp_dom_walker::before_dom_children): Likewise.
25432 (evrp_dom_walker::push_value_range): Likewise.
25433 (evrp_dom_walker::pop_value_range): Likewise.
25434 (execute_early_vrp): Likewise.
25436 2017-05-16 Richard Biener <rguenther@suse.de>
25438 * dwarf2out.c (loc_list_from_tree_1): Do not create
25439 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25441 2017-05-16 Richard Biener <rguenther@suse.de>
25443 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25445 (note_variable_value_in_expr): If we resolved the decl ref
25446 do not push to the stack.
25448 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25450 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25451 operations in fast-math mode.
25452 (vaddq_f16): Likewise.
25453 (vmul_f16): Likewise.
25454 (vmulq_f16): Likewise.
25455 (vsub_f16): Likewise.
25456 (vsubq_f16): Likewise.
25457 * config/arm/neon.md (add<mode>3): New.
25459 (fma:<VH:mode>3): New. Also remove outdated comment.
25462 2017-05-16 Martin Liska <mliska@suse.cz>
25466 * ipa-devirt.c (warn_types_mismatch): Fix typo.
25467 (odr_types_equivalent_p): Likewise.
25469 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
25471 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
25473 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
25476 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
25477 non-interunit SSE move alternatives with '?'.
25478 (zero-extendsidi peephole2): New peephole to skip intermediate
25479 general register in SSE zero-extend sequence.
25481 2017-05-15 Jeff Law <law@redhat.com>
25483 * reorg.c (relax_delay_slots): Create a new variable to hold
25484 the temporary target rather than clobbering TARGET_LABEL.
25486 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
25487 missing argument to extract_bit_field call.
25488 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
25490 2017-05-15 Martin Liska <mliska@suse.cz>
25493 * gcc.c (process_command): Do not allow empty argument of -o option.
25495 2017-05-15 Renlin Li <renlin.li@arm.com>
25497 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25498 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25499 * config/aarch64/constraints.md (Usf): Add long call check.
25500 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25501 (call_value): Likewise.
25502 (sibcall): Likewise.
25503 (sibcall_value): Likewise.
25505 (call_value_insn): New.
25506 (sibcall_insn): Update rtx pattern.
25507 (sibcall_value_insn): Likewise.
25508 (call_internal): Remove.
25509 (call_value_internal): Likewise.
25510 (sibcall_internal): Likewise.
25511 (sibcall_value_internal): Likewise.
25512 (call_reg): Likewise.
25513 (call_symbol): Likewise.
25514 (call_value_reg): Likewise.
25515 (call_value_symbol): Likewise.
25517 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25520 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25522 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25524 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25525 compatible with CCGOCmode and with CCZmode.
25527 2017-05-14 Martin Sebor <msebor@redhat.com>
25529 PR middle-end/77671
25530 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25531 (gimple_fold_builtin_snprintf): Same.
25532 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25533 (gimple_fold_builtin_snprintf): Same.
25534 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25535 of character types.
25536 (is_call_safe): New function.
25537 (try_substitute_return_value): Call it.
25538 (try_simplify_call): New function.
25539 (pass_sprintf_length::handle_gimple_call): Call it.
25541 2017-05-14 Martin Sebor <msebor@redhat.com>
25543 PR middle-end/80669
25544 * builtins.c (expand_builtin_stpncpy): Simplify.
25546 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25548 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25549 * config/i386/i386.h
25550 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25551 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25552 (struct machine_function): Add new members call_ms2sysv,
25553 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25554 (struct machine_frame_state): New fields sp_realigned and
25555 sp_realigned_offset.
25556 * config/i386/i386.c
25557 (enum xlogue_stub): New enum.
25558 (enum xlogue_stub_sets): New enum.
25559 (class xlogue_layout): New class.
25560 (struct ix86_frame): New fields stack_realign_allocate_offset,
25561 stack_realign_offset and outlined_save_offset. Modify comments to
25562 detail stack layout when using out-of-line stubs.
25563 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25564 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25565 -mcall-ms2sysv-xlogues.
25566 (stub_managed_regs): New static variable.
25567 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25568 registers managed by out-of-line stub.
25569 (disable_call_ms2sysv_xlogues): New function.
25570 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25571 m->call_ms2sysv when appropriate and compute frame layout for
25573 (sp_valid_at, fp_valid_at): New inline functions.
25574 (choose_basereg): New function.
25575 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25577 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25578 Use align parameter of choose_baseaddr to generated aligned SSE movs
25580 (pro_epilogue_adjust_stack): Modify to track
25581 machine_frame_state::sp_realigned.
25582 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25583 (ix86_nsaved_sseregs): Likewise.
25584 (ix86_emit_save_regs): Likewise.
25585 (ix86_emit_save_regs_using_mov): Likewise.
25586 (ix86_emit_save_sse_regs_using_mov): Likewise.
25587 (get_scratch_register_on_entry): Likewise.
25588 (gen_frame_set): New function.
25589 (gen_frame_load): Likewise.
25590 (gen_frame_store): Likewise.
25591 (emit_outlined_ms2sysv_save): Likewise.
25592 (emit_outlined_ms2sysv_restore): Likewise.
25593 (ix86_expand_prologue): Modify stack re-alignment code and call
25594 emit_outlined_ms2sysv_save when appropriate.
25595 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25596 parameter rtx_insn *insn, which allows the function to be used to only
25597 generate the notes.
25598 (ix86_expand_epilogue): Modify validity checks of frame and stack
25599 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25600 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25601 * config/i386/predicates.md
25602 (save_multiple): New predicate.
25603 (restore_multiple): Likewise.
25604 * config/i386/sse.md
25605 (save_multiple<mode>): New pattern.
25606 (save_multiple_realign<mode>): Likewise.
25607 (restore_multiple<mode>): Likewise.
25608 (restore_multiple_and_return<mode>): Likewise.
25609 (restore_multiple_leave_return<mode>): Likewise.
25610 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25612 2017-05-14 Julia Koval <julia.koval@intel.com>
25614 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25615 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25616 (__builtin_ia32_xsetbv): New builtins.
25617 * config/i386/i386.c (ix86_expand_special_args_builtin):
25619 (ix86_expand_builtin): Special expand for new intrinsics.
25620 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25621 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25622 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25624 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25626 * cfganal.c (inverted_post_order_compute): Change argument type
25628 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25629 * df-core.c (rest_of_handle_df_initialize): Adjust.
25630 (rest_of_handle_df_finish): Likewise.
25631 (df_analyze_1): Likewise.
25632 (df_analyze): Likewise.
25633 (loop_inverted_post_order_compute): Change argument to be a vec *.
25634 (df_analyze_loop): Adjust.
25635 (df_get_n_blocks): Likewise.
25636 (df_get_postorder): Likewise.
25637 * df.h (struct df_d): Change field to be a vec.
25638 * lcm.c (compute_laterin): Adjust.
25639 (compute_available): Likewise.
25640 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25641 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25642 * tree-ssa-pre.c (compute_antic): Likewise.
25644 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25646 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25647 (depth_first_search::depth_first_search): Change structure init
25648 function to this constructor.
25649 (depth_first_search::add_bb): Rename function to this member.
25650 (depth_first_search::execute): Likewise.
25651 (flow_dfs_compute_reverse_finish): Adjust.
25653 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25655 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25656 (longest_simple_path): Likewise.
25657 * shrink-wrap.c (spread_components): Likewise.
25658 (disqualify_problematic_components): Likewise.
25659 (emit_common_heads_for_components): Likewise.
25660 (emit_common_tails_for_components): Likewise.
25661 (insert_prologue_epilogue_for_components): Likewise.
25663 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25665 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25668 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25670 * df-core.c (df_set_blocks): Start using auto_bitmap.
25671 (df_compact_blocks): Likewise.
25672 * df-problems.c (df_rd_confluence_n): Likewise.
25673 * df-scan.c (df_insn_rescan_all): Likewise.
25674 (df_process_deferred_rescans): Likewise.
25675 (df_update_entry_block_defs): Likewise.
25676 (df_update_exit_block_uses): Likewise.
25677 (df_entry_block_bitmap_verify): Likewise.
25678 (df_exit_block_bitmap_verify): Likewise.
25679 (df_scan_verify): Likewise.
25680 * lra-constraints.c (lra_constraints): Likewise.
25681 (undo_optional_reloads): Likewise.
25682 (lra_undo_inheritance): Likewise.
25683 * lra-remat.c (calculate_gen_cands): Likewise.
25684 (do_remat): Likewise.
25685 * lra-spills.c (assign_spill_hard_regs): Likewise.
25686 (spill_pseudos): Likewise.
25687 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25688 (bitmap_set_subtract_values): Likewise.
25690 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25692 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25693 management with auto_bitmap.
25694 (fix_inter_tick): Likewise.
25695 (fix_recovery_deps): Likewise.
25696 * ira.c (add_store_equivs): Likewise.
25697 (find_moveable_pseudos): Likewise.
25698 (split_live_ranges_for_shrink_wrap): Likewise.
25699 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25700 (rtx_reuse_manager::seen_def_p): Likewise.
25701 (rtx_reuse_manager::set_seen_def): Likewise.
25702 * print-rtl.h (class rtx_reuse_manager): Likewise.
25704 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25706 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25708 (migrate_btr_def): Likewise.
25709 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25710 * df-core.c (loop_post_order_compute): Likewise.
25711 (loop_inverted_post_order_compute): Likewise.
25712 * hsa-common.h: Likewise.
25713 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25714 * init-regs.c (initialize_uninitialized_regs): Likewise.
25715 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25716 (inline_small_functions): Likewise.
25717 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25718 * ira.c (combine_and_move_insns): Likewise.
25719 (build_insn_chain): Likewise.
25720 * loop-invariant.c (find_invariants): Likewise.
25721 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25722 * predict.c (tree_predict_by_opcode): Likewise.
25723 (predict_paths_leading_to): Likewise.
25724 (predict_paths_leading_to_edge): Likewise.
25725 (estimate_loops_at_level): Likewise.
25726 (estimate_loops): Likewise.
25727 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25728 (spread_components): Likewise.
25729 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25730 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25731 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25732 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25733 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25734 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25735 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25736 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25737 (mark_threaded_blocks): Likewise.
25738 (thread_through_all_blocks): Likewise.
25739 * tree-ssa.c (verify_ssa): Likewise.
25740 (execute_update_addresses_taken): Likewise.
25741 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25743 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25745 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25747 (post_order_compute): Likewise.
25748 (inverted_post_order_compute): Likewise.
25749 (pre_and_rev_post_order_compute_fn): Likewise.
25751 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25753 * genrecog.c (int_set::int_set): Explicitly construct our
25754 auto_vec base class.
25755 * vec.h (auto_vec::auto_vec): New constructor.
25757 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25759 * bitmap.h (class auto_bitmap): New constructor taking
25760 bitmap_obstack * argument.
25762 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25764 * bitmap.h (class auto_bitmap): Change type of m_bits to
25765 bitmap_head, and adjust ctor / dtor and member operators.
25767 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25769 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25770 when returned register mode doesn't match original mode.
25772 2017-05-12 Jeff Law <law@redhat.com>
25773 Jakub Jelinek <jakub@redhat.com>
25775 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25776 we look for cc setter after the compare-elim changes.
25777 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25778 within the vector to match what compare-elim now expects.
25779 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25780 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25782 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25783 after the compare-elim changes.
25784 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25785 the vector to match what compare-elim now expects.
25786 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25787 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25788 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25789 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25790 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25792 * config/visium/visium.c (single_set_and_flags): Fix where
25793 we look for cc setter after the compare-elim changes.
25794 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25795 with the vector to match what compare-elim now expects.
25796 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25797 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25798 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25799 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25800 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25801 (neg<mode>2_insn_set_overflow): Likewise.
25803 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25805 PR middle-end/79794
25806 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25807 maybe_expand_insn call, set ops[0].target. If still set after call,
25808 set alt_rtl. Add extra arg to recursive calls.
25809 (extract_bit_field): Add alt_rtl argument. Pass to
25811 * expmed.h (extract_bit_field): Fix prototype.
25812 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25813 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25814 to extract_bit_field_calls.
25815 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25816 Pass alt_rtl to extract_bit_field calls.
25817 * calls.c (store_unaligned_arguments_into_psuedos)
25818 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25819 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25821 * optabs.h (struct expand_operand): Add target bitfield.
25823 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25825 * compare-elim.c (try_eliminate_compare): Canonicalize
25826 operation with embedded compare to
25827 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25828 (set (reg) (operation)].
25830 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25832 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25835 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25836 cost of adding a carry flag for ADC instruction.
25837 [case MINUS]: Ignore the cost of subtracting a carry flag
25838 for SBB instruction.
25840 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25842 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25844 * config/rs6000/bmiintrin.h: New file.
25845 * config/rs6000/bmi2intrin.h: New file.
25846 * config/rs6000/x86intrin.h: New file.
25848 2017-05-12 Jeff Law <law@redhat.com>
25850 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25853 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25855 PR middle-end/80707
25856 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25858 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25860 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25861 early expansion of vector divide builtins.
25862 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25863 builtins identified as having unsigned arguments.
25865 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25867 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25868 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25869 expansion of vector logical operations (and, andc, or, xor,
25872 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25874 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25875 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25877 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25879 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25880 early GIMPLE expansion of vector multiplies.
25882 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25884 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25885 TARGET_HAVE_MOVT conditional.
25886 (movt splitter): Likewise.
25888 2017-05-12 Richard Biener <rguenther@suse.de>
25890 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25891 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25892 Fold all stmts not inplace.
25894 2017-05-12 Richard Biener <rguenther@suse.de>
25896 PR tree-optimization/80713
25897 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25898 inserted_exprs bit for not removed stmts.
25900 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25902 PR middle-end/69921
25903 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25904 parallelized" attribute for parallelized OpenACC kernels.
25905 * omp-offload.c (execute_oacc_device_lower): Use it.
25907 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25908 Set "oacc kernels" attribute.
25909 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25910 parameter. Adjust all users.
25911 (oacc_fn_attrib_kernels_p): Remove function.
25912 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25913 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25914 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25915 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25916 assert "oacc kernels" attribute is set.
25918 2017-05-11 Carl Love <cel@us.ibm.com>
25920 * config/rs6000/rs6000-c: Add support for built-in functions
25921 vector unsigned char vec_popcnt (vector signed char)
25922 vector unsigned char vec_popcnt (vector unsigned char)
25923 vector unsigned short vec_popcnt (vector signed short)
25924 vector unsigned short vec_popcnt (vector unsigned short)
25925 vector unsigned int vec_popcnt (vector signed int)
25926 vector unsigned int vec_popcnt (vector unsigned int)
25927 vector unsigned long long vec_popcnt (vector signed long long)
25928 vector unsigned long long vec_popcnt (vector unsigned long long)
25929 vector signed long long vec_slo (vector signed long long,
25930 vector signed char)
25931 vector signed long long vec_slo (vector signed long long,
25932 vector unsigned char)
25933 vector unsigned long long vec_slo (vector unsigned long long,
25934 vector signed char)
25935 vector unsigned long long vec_slo (vector unsigned long long,
25936 vector unsigned char)
25937 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25938 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25939 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25940 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25941 * doc/extend.texi: Update the built-in documentation file for the
25942 new built-in functions.
25944 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25946 * attribs.h (sorted_attr_string): Move machine independent
25947 functions for target clone support from the i386 port to common
25948 code. Rename ix86_function_versions to common_function_versions.
25949 Rename make_name to make_unique_name.
25950 (common_function_versions): Likewise.
25951 (make_unique_name): Likewise.
25952 (make_dispatcher_decl): Likewise.
25953 (is_function_default_version): Likewise.
25954 * attribs.c (attr_strcmp): Likewise.
25955 (sorted_attr_string): Likewise.
25956 (common_function_versions): Likewise.
25957 (make_unique_name): Likewise.
25958 (make_dispatcher_decl): Likewise.
25959 (is_function_default_version): Likewise.
25960 * config/i386/i386.c (attr_strcmp): Likewise.
25961 (sorted_attr_string): Likewise.
25962 (ix86_function_versions): Likewise.
25963 (make_name): Likewise.
25964 (make_dispatcher_decl): Likewise.
25965 (is_function_default_version): Likewise.
25966 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25968 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25971 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25972 Account for direct move costs for vec_construct of integer
25975 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25978 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25979 (UNSPEC_STX_ATOMIC): Ditto.
25980 (loaddi_via_sse): New insn.
25981 (storedi_via_sse): Ditto.
25982 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25983 Update corresponding peephole2 patterns.
25984 (atomic_storedi_fpu): Ditto.
25986 2017-05-11 Julia Koval <julia.koval@intel.com>
25988 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25989 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25991 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25992 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25993 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25995 2017-05-11 Nathan Sidwell <nathan@acm.org>
25997 * graphite-poly.c: Include dumpfile.h.
25999 * dumpfle.h (dump_function): Declare here ...
26000 * tree-dump.h (dump_function): ... not here.
26001 * dumpfile.c: #include tree-cfg.h.
26002 (dump_function): Move here from ...
26003 * tree-dump.c (dump_function): ... here.
26004 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26005 * graphite-poly.c: Don't include tree-dump.h.
26006 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26007 * print-tree.c: Likewise.
26008 * stor-layout.c: Likewise.
26009 * tree-nested.c: Likewise.
26011 * dumpfile.c (dump_start): Use TDF_FLAGS.
26012 (dump_enable_all): Fix TDF_KIND check thinko.
26014 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26016 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26017 array entries to represent two legal parameterizations of the
26018 overloaded __builtin_cmpb function, as represented by the
26019 P6_OV_BUILTIN_CMPB constant.
26020 (altivec_resolve_overloaded_builtin): Add special case handling
26021 for the __builtin_cmpb function, as represented by the
26022 P6_OV_BUILTIN_CMPB constant.
26023 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26024 (BU_P6_64BIT_2): New macro.
26025 (BU_P6_OVERLOAD_2): New macro
26026 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26027 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26028 (CMPB): Add overload support to represent both 32-bit and 64-bit
26029 compare-bytes function.
26030 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26031 support for TARGET_CMPB.
26032 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26033 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26034 documentation of the __builtin_cmpb overloaded built-in function.
26036 2017-05-11 Richard Biener <rguenther@suse.de>
26038 PR tree-optimization/80705
26039 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26040 bases are not vectorizable.
26042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26044 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26045 when counting register pressure.
26047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26049 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26050 (iv_ca_more_deps): Renamed to ...
26051 (iv_ca_compare_deps): ... this.
26052 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26056 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26058 (determine_group_iv_costs): ... here.
26059 (find_inv_vars_cb): Record inv var if it's not recorded before.
26061 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26063 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26064 (get_shiftadd_cost): Ditto.
26066 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26068 * tree-ssa-address.c: Include header file.
26069 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26071 (add_to_parts): Refactor.
26072 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26073 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26076 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26078 PR tree-optimization/53090
26079 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26081 (extract_cond_operands): Detect condition with IV on both sides
26082 and return COMP_IV_EXPR_2.
26083 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26084 (rewrite_use_compare): Simplify by removing call to function
26085 extract_cond_operands.
26087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26089 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26090 (extract_cond_operands): Detect condition comparing against non-
26091 invariant bound and return appropriate enum value.
26092 (find_interesting_uses_cond): Update use of extract_cond_operands.
26093 Handle its return value accordingly.
26094 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26096 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26098 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26099 nonlinear iv_use computation in loop invariant sensitive way.
26101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26103 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26104 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26106 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26108 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26109 (dump_cand): Support iv_cand.inv_exprs.
26110 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26112 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26117 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26119 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26120 as local function. Include necessary header files.
26121 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26125 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26127 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26129 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26130 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26131 RSHIFT_EXPR and BIT_NOT_EXPR.
26133 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26135 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26136 (adjust_setup_cost): New parameter supporting round up adjustment.
26137 (struct address_cost_data): Delete.
26138 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26139 (split_address_cost, ptr_difference_cost): Delete.
26140 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26141 (struct ainc_cost_data): New struct.
26142 (get_address_cost_ainc): New function.
26143 (get_address_cost, get_computation_cost): Reimplement.
26144 (determine_group_iv_cost_address): Record inv_expr for all uses of
26146 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26147 (iv_ca_has_deps): Reimplemented to ...
26148 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26150 (iv_ca_extend): Call iv_ca_more_deps.
26152 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26154 * tree-ssa-address.c (struct mem_address): Move to header file.
26155 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26156 * tree-ssa-address.h (struct mem_address): Move from C file.
26157 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26159 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26161 * tree-affine.h (aff_combination_type): New interface.
26162 (aff_combination_zero_p): Remove static.
26163 (aff_combination_const_p): New interface.
26164 (aff_combination_singleton_var_p): New interfaces.
26166 2017-05-11 Richard Biener <rguenther@suse.de>
26168 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26169 Skip unreachable blocks and destinations.
26170 (eliminate): Move stmt removal and fixup ...
26171 (fini_eliminate): ... here. Skip inserted exprs.
26172 (pass_pre::execute): Move fini_pre after fini_eliminate.
26173 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26174 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26175 PRE to get rid of dead code that has invalid SSA form and
26176 split critical edges again.
26178 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26180 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26182 2017-05-11 Richard Biener <rguenther@suse.de>
26184 * passes.c (execute_function_todo): Verify loops if they are
26185 said to be up-to-date.
26186 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26187 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26189 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26192 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26193 handle calling assemble_external ourself.
26196 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26197 modes with zero size. Enhance comment.
26199 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26201 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26202 built-ins for vec_xl and vec_xst with short and char pointer
26205 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26207 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26208 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26209 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26210 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26211 (_mm_maskz_min_round_ss): New intrinsics.
26212 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26213 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26214 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26215 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26216 (__builtin_ia32_minss_mask_round): New builtins.
26217 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26218 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26219 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26221 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26222 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26224 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26227 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26229 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26230 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26231 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26232 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26233 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26234 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26235 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26236 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26237 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26238 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26239 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26240 (__builtin_ia32_mulss_mask_round): New builtins.
26241 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26242 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26243 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26245 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26246 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26248 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26251 2017-05-10 Julia Koval <julia.koval@intel.com>
26253 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26254 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26255 (_mm256_setr_m128i): New intrinsics.
26257 2017-05-10 Julia Koval <julia.koval@intel.com>
26259 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26260 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26261 (_mm_maskz_rcp14_ss): New intrinsics.
26262 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26263 (__builtin_ia32_rcp14ss_mask): New builtins.
26264 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26266 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26268 PR tree-optimization/51513
26269 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26270 (assert_unreachable_fallthru_edge_p): Use it.
26271 (group_case_labels_stmt): Likewise.
26272 * tree-cfg.h: Prototype it.
26273 * stmt.c: Include cfghooks.h and tree-cfg.h.
26274 (emit_case_dispatch_table) <gap_label>: New local variable.
26275 Use it to fill dispatch table gaps.
26276 Test for default_label before updating probabilities.
26277 (expand_case) <default_label>: Remove unneeded initialization.
26278 Test for unreachable default case statement and remove its edge.
26279 Set default_label accordingly.
26280 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26282 2017-05-10 Carl Love <cel@us.ibm.com>
26284 * config/rs6000/rs6000-c: Add support for built-in functions
26285 vector signed char vec_neg (vector signed char)
26286 vector signed short int vec_neg (vector short int)
26287 vector signed int vec_neg (vector signed int)
26288 vector signed long long vec_neg (vector signed long long)
26289 vector float vec_neg (vector float)
26290 vector double vec_neg (vector double)
26291 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26293 * config/rs6000/altivec.h: Add define for vec_neg
26294 * doc/extend.texi: Update the built-in documentation for the
26295 new built-in functions.
26297 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26299 PR tree-optimization/77644
26300 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26302 2017-05-10 Nathan Sidwell <nathan@acm.org>
26304 * dumpfile.h (TDI_lang_all): New.
26305 (TDF_KIND): New. Renumber others
26306 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26308 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26310 (get_dump_file_name): Adjust suffix generation.
26311 (dump_enable_all): Use TDF_KIND.
26312 * doc/invoke.texi (-fdump-lang-all): Document.
26314 * dumpfile.h: Tabify.
26316 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26319 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26320 Move member access before delete.
26322 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26324 * tree-inline.c (expand_call_inline): Split block at stmt
26327 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26330 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26331 are now unused after splitting mov{sf,sd}_hardfloat.
26332 (f32_lr2): Likewise.
26333 (f32_lm): Likewise.
26334 (f32_lm2): Likewise.
26335 (f32_li): Likewise.
26336 (f32_li2): Likewise.
26337 (f32_lv): Likewise.
26338 (f32_sr): Likewise.
26339 (f32_sr2): Likewise.
26340 (f32_sm): Likewise.
26341 (f32_sm2): Likewise.
26342 (f32_si): Likewise.
26343 (f32_si2): Likewise.
26344 (f32_sv): Likewise.
26345 (f32_dm): Likewise.
26346 (f32_vsx): Likewise.
26347 (f32_av): Likewise.
26348 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26349 For movsf, order stores so the VSX stores occur before the GPR
26350 store which encourages the register allocator to use a traditional
26351 FPR instead of a GPR. For movsd, order the stores so that the GPR
26352 store comes before the VSX stores to allow the power6 to work.
26353 This is due to the power6 not having a 32-bit integer store
26354 instruction from a FPR.
26355 (movsf_hardfloat): Likewise.
26356 (movsd_hardfloat): Likewise.
26358 2017-05-09 Martin Sebor <msebor@redhat.com>
26360 PR translation/80280
26361 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26364 PR translation/80280
26365 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26366 data member added in r247778.
26367 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26369 2017-05-09 Nathan Sidwell <nathan@acm.org>
26371 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26373 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26376 2017-05-09 Marek Polacek <polacek@redhat.com>
26378 * doc/invoke.texi: Fix typo.
26380 2017-05-09 Richard Biener <rguenther@suse.de>
26382 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26383 (vrp_val_is_min): Likewise.
26384 (set_value_range_to_value): Likewise.
26385 (set_value_range_to_nonnegative): Likewise.
26386 (gimple_assign_nonzero_p): Likewise.
26387 (gimple_stmt_nonzero_p): Likewise.
26388 (vrp_int_const_binop): Likewise. Remove unreachable case.
26389 (adjust_range_with_scev): Adjust comments.
26390 (compare_range_with_value): Likewise.
26391 (extract_range_from_phi_node): Likewise.
26392 (test_for_singularity): Likewise.
26394 2017-05-09 Richard Biener <rguenther@suse.de>
26396 * tree-vrp.c (get_single_symbol): Add assert that we don't
26397 get overflowed constants as invariant part.
26398 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26399 checks. Use wi::cmp instead of recursing for integer constants.
26400 (compare_values): Just ignore whether we assumed undefined
26401 overflow instead of failing the compare.
26402 (extract_range_for_var_from_comparison_expr): Add comment before the
26403 TREE_NO_WARNING sets.
26404 (test_for_singularity): Likewise.
26405 (extract_range_from_comparison): Do not disable optimization
26406 when we assumed undefined overflow.
26407 (extract_range_basic): Remove init of unused var.
26409 2017-05-09 Richard Biener <rguenther@suse.de>
26411 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26412 (extract_range_from_multiplicative_op_1): Adjust.
26413 (extract_range_from_binary_expr_1): Use int_const_binop.
26415 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26418 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26419 rs6000_store_data_bypass_p in seven define_bypass directives and
26420 in several comments.
26421 * config/rs6000/rs6000-protos.h: Add prototype for
26422 rs6000_store_data_bypass_p function.
26423 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26424 function implements slightly different (rs6000-specific) semantics
26425 than store_data_bypass_p, returning false rather than aborting
26426 with assertion error when arguments do not satisfy the
26427 requirements of store data bypass.
26428 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26429 rs6000_store_data_bypass_p.
26431 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26433 * config/xtensa/xtensa-protos.h
26434 (xtensa_initial_elimination_offset): New declaration.
26435 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26436 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26437 macro definition, add case for FRAME_POINTER_REGNUM when
26438 FRAME_GROWS_DOWNWARD.
26439 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26440 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26441 xtensa_initial_elimination_offset.
26443 2017-05-08 Nathan Sidwell <nathan@acm.org>
26445 * doc/invoke.texi: Alphabetize -fdump options.
26447 2017-05-08 Martin Sebor <msebor@redhat.com>
26449 PR translation/80280
26450 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26452 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26454 * target.def (compute_frame_layout): New optional target hook.
26455 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26456 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26457 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
26459 * reload1.c (verify_initial_elim_offsets): Likewise.
26460 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
26461 (use_simple_return_p): Call arm_compute_frame_layout if needed.
26462 (arm_get_frame_offsets): Split up into this ...
26463 (arm_compute_frame_layout): ... and this function.
26465 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
26467 * config/aarch64/constraints.md (Usa): New constraint.
26468 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
26470 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
26472 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
26473 with_multilib_list after it has been checked.
26475 2017-05-08 Richard Biener <rguenther@suse.de>
26477 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
26478 (bitmap_set_subtract_values): Likewise.
26480 2017-05-08 Richard Biener <rguenther@suse.de>
26482 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
26483 (gimple_assign_nonzero): ... this and remove strict_overflow_p
26485 (gimple_stmt_nonzero_warnv_p): Rename to ...
26486 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
26488 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
26489 (extract_range_basic): Adjust, do not disable propagation on
26490 strict overflow sensitive simplification.
26491 (vrp_visit_cond_stmt): Likewise.
26493 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26495 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26496 body size unconditionally.
26498 2017-05-07 Jeff Law <law@redhat.com>
26501 2017-05-06 Jeff Law <law@redhat.com>
26502 PR tree-optimization/78496
26503 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26506 PR tree-optimization/78496
26507 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26508 (simplify_stmt_using_ranges): Call it.
26509 (vrp_dom_walker::before_dom_children): Extract equivalences
26510 from an ASSERT_EXPR with an equality comparison against a
26513 2017-05-06 Jeff Law <law@redhat.com>
26515 PR tree-optimization/78496
26516 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26519 PR tree-optimization/78496
26520 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26521 (simplify_stmt_using_ranges): Call it.
26522 (vrp_dom_walker::before_dom_children): Extract equivalences
26523 from an ASSERT_EXPR with an equality comparison against a
26526 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26528 * lra-constraints.c (lra_copy_reg_equiv): New function.
26529 (split_reg): Use it to copy equivalence information from the
26530 original register to the spill register.
26532 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26534 PR rtl-optimization/75964
26535 * simplify-rtx.c (simplify_const_relational_operation): Remove
26536 invalid handling of comparisons of integer ABS.
26538 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26540 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26541 initialize to zero.
26542 (init_regs): Remove declaration.
26543 (function_arg_advance_32): Initialize error_p as boolean variable.
26545 2017-05-05 Nathan Sidwell <nathan@acm.org>
26547 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26548 lines. Use for (;;).
26550 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26552 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26553 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26554 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26555 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26556 VF=2 that require versioning.
26558 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26560 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26563 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26565 * diagnostic.h (diagnostic_override_option_index): Convert from
26566 macro to inline function.
26568 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26570 * diagnostic.c (last_module_changed_p): New function.
26571 (set_last_module): New function.
26572 (diagnostic_report_current_module): Convert macro usage to
26573 the above functions.
26574 * diagnostic.h (diagnostic_context::last_module): Strengthen
26575 from const line_map * to const line_map_ordinary *.
26576 (diagnostic_last_module_changed): Delete macro.
26577 (diagnostic_set_last_module): Delete macro.
26579 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26581 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26582 with diagnostic_report_diagnostic.
26583 (diagnostic_n_impl_richloc): Likewise.
26584 * diagnostic.h (report_diagnostic): Delete macro.
26585 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26586 with diagnostic_report_diagnostic.
26587 * substring-locations.c (format_warning_va): Likewise.
26589 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26591 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26592 save/restor of format_spec. Move option-printing code to...
26593 (print_option_information): ...this new function, and
26594 reimplement by simply printing to the pretty_printer,
26595 rather than appending to the format string.
26597 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26599 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26600 handling logic into...
26601 (update_effective_level_from_pragmas): ...this new function.
26603 2017-05-04 Andrew Waterman <andrew@sifive.com>
26605 * config/riscv/riscv.opt (mstrict-align): New option.
26606 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26607 (SLOW_UNALIGNED_ACCESS): Define.
26608 (riscv_slow_unaligned_access): Declare.
26609 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26611 (riscv_slow_unaligned_access): New variable.
26612 (rocket_tune_info): Set slow_unaligned_access to true.
26613 (optimize_size_tune_info): Set slow_unaligned_access to false.
26614 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26615 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26616 (riscv_option_override): Set riscv_slow_unaligned_access.
26617 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26619 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26621 * config/riscv/riscv.md: Unify indentation.
26623 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26628 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26630 (extendsi<mode>2): Add support for doing sign extension via
26631 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26632 don't have ISA 3.0 instructions.
26633 (extendsi<mode>2 splitter): Likewise.
26634 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26635 generate the normal insns since SImode can now go in vector
26636 registers. Disallow the special UNSPECs needed for previous
26637 machines to hide SImode being used. Add new insns
26638 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26639 (fix_trunc<mode>si2_stfiwx): Likewise.
26640 (fix_trunc<mode>si2_internal): Likewise.
26641 (fixuns_trunc<mode>si2): Likewise.
26642 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26643 (fctiw<u>z_<mode>_smallint): Likewise.
26644 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26645 of floating point to 32-bit integer from doing a direct move to
26646 the GPR registers to do a store.
26647 (fctiwz_<mode>): Break long line.
26649 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26651 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26652 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26653 (addr_list, addr_offset_valid_p): New.
26654 (split_address_groups): Check offset validity with above function.
26655 (gt-tree-ssa-loop-ivopts.h): Include header file.
26657 2017-05-05 Nathan Sidwell <nathan@acm.org>
26659 * config.gcc (arm*-*-*): Add missing 'fi'.
26661 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26663 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26664 included in -fopt-info does not matter.
26665 * doc/optinfo.texi (-fopt-info): Fix description of default
26666 behavour. Explicitly say order of options included in -fopt-info
26669 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26671 * config.gcc: Allow combinations of aprofile and rmprofile values for
26672 --with-multilib-list.
26673 * config/arm/t-multilib: New file.
26674 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26675 variables. Remove setting of ISA and floating-point ABI in
26676 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26677 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26678 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26679 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26681 * config/arm/t-rmprofile: Likewise except for the matches changes.
26682 * doc/install.texi (--with-multilib-list): Document the combination of
26683 aprofile and rmprofile values and warn about pitfalls in doing that.
26685 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26687 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26688 (movdi_aarch64): Likewise.
26690 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26692 PR tree-optimization/80632
26693 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26695 (build_arrays): Initialize it for virtual phis.
26696 (fix_phi_nodes): Use it for virtual phis.
26698 PR tree-optimization/80558
26699 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26700 [x, y] op z into [x op, y op z] for op & or | if conditions
26703 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26704 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26707 * config/arm/arm.md (use_literal_pool): Remove.
26708 (64-bit immediate split): No longer takes cost into consideration
26709 if arm_disable_literal_pool is enabled.
26710 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26711 used when arm_disable_literal_pool is enabled.
26712 (arm_max_const_double_inline_cost): Remove use of
26713 arm_disable_literal_pool.
26714 (push_minipool_fix): Add assert.
26715 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26716 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26717 (no_literal_pool_sf_immediate): New.
26719 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26721 PR tree-optimization/80613
26722 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26723 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26725 2017-05-05 Richard Biener <rguenther@suse.de>
26727 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26729 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26731 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26732 of this flag from insn conditions due to removal from r247495.
26734 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26736 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26738 (arm_early_store_addr_dep_ptr): Likewise.
26739 * config/arm/aarch-common-protos.h
26740 (arm_early_load_addr_dep_ptr): Add prototype.
26741 (arm_early_store_addr_dep_ptr): Likewise.
26742 * config/arm/cortex-a53.md: Add new bypasses.
26744 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26746 * tree.c (next_type_uid): Change type to unsigned.
26747 (type_hash_canon): Decrement back next_type_uid if
26748 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26749 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26752 2017-05-04 Martin Sebor <msebor@redhat.com>
26754 * builtins.c: Fix a trivial typo in a comment.
26756 PR middle-end/79234
26757 * builtins.c (check_sizes): Adjust to handle reading past the end.
26758 Avoid printing excessive upper bound of ranges. Use %E to print
26759 tree nodes instead of converting them to %wu.
26760 (expand_builtin_memchr): New function.
26761 (compute_dest_size): Rename...
26762 (compute_objsize): ...to this.
26763 (expand_builtin_memcpy): Adjust.
26764 (expand_builtin_mempcpy): Adjust.
26765 (expand_builtin_strcat): Adjust.
26766 (expand_builtin_strcpy): Adjust.
26767 (check_strncat_sizes): Adjust.
26768 (expand_builtin_strncat): Adjust.
26769 (expand_builtin_strncpy): Adjust and simplify.
26770 (expand_builtin_memset): Adjust.
26771 (expand_builtin_bzero): Adjust.
26772 (expand_builtin_memcmp): Adjust.
26773 (expand_builtin): Handle memcmp.
26774 (maybe_emit_chk_warning): Check strncat just once.
26776 2017-05-04 Martin Sebor <msebor@redhat.com>
26778 PR preprocessor/79214
26779 PR middle-end/79222
26780 PR middle-end/79223
26781 * builtins.c (check_sizes): Add inlining context and issue
26782 warnings even when -Wno-system-headers is set.
26783 (check_strncat_sizes): Same.
26784 (expand_builtin_strncat): Same.
26785 (expand_builtin_memmove): New function.
26786 (expand_builtin_stpncpy): Same.
26787 (expand_builtin): Handle memmove and stpncpy.
26789 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26791 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26792 which is not used any more.
26794 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26796 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26798 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26800 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26801 (cortexa53_tunings): Likewise.
26802 (cortexa57_tunings): Likewise.
26803 (cortexa72_tunings): Likewise.
26804 (cortexa73_tunings): Likewise.
26806 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26808 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26809 Set loop alignment to 8.
26811 2017-05-04 Martin Sebor <msebor@redhat.com>
26813 PR translation/80280
26814 * builtins.c (expand_builtin_object_size): Add missing quoting to
26815 %D and like directives.
26816 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26817 (hsa_type_for_tree_type): Same.
26818 (verify_function_arguments): Same.
26819 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26820 * varasm.c (get_section): Same.
26823 2017-05-04 Martin Sebor <msebor@redhat.com>
26825 PR translation/80280
26826 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26828 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26830 * config/aarch64/aarch64.c (generic_addrcost_table):
26831 Change HI/TI mode setting.
26833 2017-05-04 Martin Jambor <mjambor@suse.cz>
26835 PR tree-optimization/80622
26836 * tree-sra.c (comes_initialized_p): New function.
26837 (build_accesses_from_assign): Only set write lazily when
26838 comes_initialized_p is false.
26839 (analyze_access_subtree): Use comes_initialized_p.
26840 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26841 instead of testing for PARM_DECL.
26843 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26845 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26846 constraint on operand 0 to allow more general addressing modes.
26847 Adjust output template.
26848 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26850 * config/aarch64/aarch64-protos.h
26851 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26852 * config/aarch64/constraints.md (Dp): New address constraint.
26853 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26856 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26858 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26859 update use of estimate_ipcp_clone_size_and_time.
26860 (estimate_local_effects): Update use of
26861 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26862 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26863 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26864 Return nonspecialized time.
26866 2017-05-04 Richard Biener <rguenther@suse.de>
26868 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26869 for the last VUSE which def dominates the PHI. Directly call
26871 (get_continuation_for_phi_1): Remove.
26873 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26875 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26876 to explain the use of truncating division. Cap the number of
26877 iterations to the maximum given by nb_iterations_upper_bound,
26880 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26882 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26883 * configure: Regenerate.
26884 * config.in: Regenerate.
26885 * config/i386/driver-mingw32.c: new file.
26886 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26887 * config.host: Link driver-mingw32.o on MinGW host.
26888 * doc/install.texi: Document new --enable-mingw-wildcard configure
26891 2017-05-04 Marek Polacek <polacek@redhat.com>
26893 PR tree-optimization/80612
26894 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26896 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26897 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26899 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26900 (movt splitter): Likewise.
26901 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26902 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26903 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26904 block for Thumb-1 with MOVT.
26905 (thumb2_legitimate_address_p): Move code block ...
26906 (can_avoid_literal_pool_for_label_p): ... into this new function.
26907 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26909 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26910 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26911 "M-profile targets with the MOVT instruction".
26913 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26915 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26916 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26917 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26919 2017-05-04 Martin Liska <mliska@suse.cz>
26921 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26922 variable cond_code.
26924 2017-05-04 Richard Biener <rguenther@suse.de>
26926 * tree.c (array_at_struct_end_p): Handle arrays at struct
26927 end with flexarrays more conservatively. Refactor and treat
26928 arrays of arrays or aggregates more strict. Fix
26929 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26930 * tree.c (array_at_struct_end_p): Adjust prototype.
26931 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26932 * gimple-fold.c (get_range_strlen): Likewise.
26933 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26935 2017-05-04 Richard Biener <rguenther@suse.de>
26937 PR tree-optimization/31130
26938 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26940 (supports_overflow_infinity): Likewise.
26941 (is_negative_overflow_infinity): Likewise.
26942 (is_positive_overflow_infinity): Likewise.
26943 (is_overflow_infinity): Likewise.
26944 (stmt_overflow_infinity): Likewise.
26945 (overflow_infinity_range_p): Likewise.
26946 (usable_range_p): Remove as always returning true.
26947 (make_overflow_infinity): Remove.
26948 (negative_overflow_infinity): Likewise.
26949 (positive_overflow_infinity): Likewise.
26950 (avoid_overflow_infinity): Likewise.
26951 (set_value_range): Adjust accordingly.
26952 (set_value_range_to_nonnegative): Likewise, remove now unused
26953 overflow_infinity arg.
26954 (vrp_operand_equal_p): Adjust.
26955 (update_value_range): Likewise.
26956 (range_int_cst_singleton_p): Likewise.
26957 (operand_less_p): Likewise.
26958 (compare_values_warnv): Likewise.
26959 (extract_range_for_var_from_comparison_expr): Likewise.
26960 (vrp_int_const_binop): Likewise.
26961 (zero_nonzero_bits_from_vr): Likewise.
26962 (extract_range_from_multiplicative_op_1): Likewise.
26963 (extract_range_from_binary_expr_1): Likewise.
26964 (extract_range_from_unary_expr): Likewise.
26965 (extract_range_from_comparison): Likewise.
26966 (extract_range_basic): Likewise.
26967 (adjust_range_with_scev): Likewise.
26968 (compare_ranges): Likewise.
26969 (compare_range_with_value): Likewise.
26970 (dump_value_range): Likewise.
26971 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26973 (simplify_cond_using_ranges): Adjust.
26975 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26977 * brig-builtins.def: Added a builtin for class_f64.
26978 * builtin-types.def: Added a builtin type needed by class_f64.
26980 2017-05-03 Jason Merrill <jason@redhat.com>
26982 * timevar.def: Add TV_CONSTEXPR.
26984 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26986 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26988 2017-05-03 Martin Jambor <mjambor@suse.cz>
26990 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26991 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26992 * ipa-cp.c (ipcp_propagate_stage): Do not call
26993 ipa_update_after_lto_read.
26994 * ipa-inline.c (ipa_inline): Likewise.
26996 2017-05-03 Martin Jambor <mjambor@suse.cz>
26998 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26999 tag. Added a default constructor and a destructor.
27000 (ipa_edge_args_sum_t): New class;
27001 (ipa_edge_args_sum): Declare.
27002 (ipa_edge_args_vector): Remove declaration.
27003 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27004 (ipa_free_edge_args_substructures): Remove declaration.
27005 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27006 (ipa_edge_args_info_available_for_edge_p): Likewise.
27007 * ipa-prop.c (ipa_edge_args_vector): Removed.
27008 (edge_removal_hook_holder): Likewise.
27009 (edge_duplication_hook_holder): Likewise.
27010 (ipa_edge_args_sum): New variable.
27011 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27012 ipa_edge_args_vector.
27013 (ipa_free_edge_args_substructures): Likewise.
27014 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27015 ipa_edge_args_vector.
27016 (ipa_edge_removal_hook): Turned into method
27017 ipa_edge_args_sum_t::remove.
27018 (ipa_edge_duplication_hook): Turned into method
27019 ipa_edge_args_sum_t::duplicate.
27020 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27021 registering edge hooks.
27022 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27023 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27024 ipa_edge_args_sum instead of ipa_edge_args_vector.
27025 * ipa-profile.c (ipa_profile): Likewise.
27027 2017-05-03 Martin Jambor <mjambor@suse.cz>
27029 * symbol-summary.h (function_summary): New method exists.
27030 (function_summary::symtab_removal): Deallocate through release.
27031 (call_summary): New class.
27032 (gt_ggc_mx): New overload.
27033 (gt_pch_nx): Likewise.
27034 (gt_pch_nx): Likewise.
27036 2017-05-03 Jeff Law <law@redhat.com>
27038 PR tree-optimization/78496
27039 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27040 from simplify_cond_using_ranges. Split off code to walk
27041 backwards through casts into ...
27042 (simplify_cond_using_ranges_2): New function.
27043 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27044 (execute_vrp): After identifying jump threads, call
27045 simplify_cond_using_ranges_2.
27047 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27050 * ipa-inline.h (inline_summary): Add ctor.
27051 (create_ggc): Do not use ggc_cleared_alloc.
27053 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27054 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27056 * gcc.c (handle_braces): Support escaping in switch matching
27058 * doc/invoke.texi (Spec Files): Document it.
27059 Remove superfluous @code markup in items.
27061 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27063 * diagnostic-show-locus.c (struct column_range): New struct.
27064 (get_affected_columns): New function.
27065 (get_printed_columns): New function.
27066 (struct correction): New struct.
27067 (correction::ensure_capacity): New function.
27068 (correction::ensure_terminated): New function.
27069 (struct line_corrections): New struct.
27070 (line_corrections::~line_corrections): New dtor.
27071 (line_corrections::add_hint): New function.
27072 (layout::print_trailing_fixits): Reimplement in terms of the new
27074 (selftest::test_overlapped_fixit_printing): New function.
27075 (selftest::diagnostic_show_locus_c_tests): Call it.
27077 2017-05-03 Nathan Sidwell <nathan@acm.org>
27079 Canonicalize canonical type hashing
27080 * tree.h (type_hash_canon_hash): Declare.
27081 * tree.c (type_hash_list, attribute_hash_list): Move into
27082 type_hash_canon_hash.
27083 (build_type_attribute_qual_variant): Break out hash code calc into
27084 type_hash_canon_hash.
27085 (type_hash_canon_hash): New. Generic type hash computation.
27086 (build_range_type_1, build_array_type_1, build_function_type,
27087 build_method_type_directly, build_offset_type, build_complex_type,
27088 make_vector_type): Call it.
27090 2017-05-03 Richard Biener <rguenther@suse.de>
27092 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27093 When all DRs have unknown misaligned do not always peel
27094 when there is a store but apply the same costing model as if
27095 there were only loads.
27097 2017-05-03 Richard Biener <rguenther@suse.de>
27100 PR tree-optimization/80492
27101 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27102 compare_base_decls returning dont-know properly.
27104 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27106 * config/arm/iterators.md (CCSI): New mode iterator.
27107 (arch): New mode attribute.
27108 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27109 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27110 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27111 code iterator for success result mode.
27112 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27113 the corresponding new insn generators.
27115 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27118 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27119 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27121 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27123 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27124 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27125 (DDR_A): Wrap DDR argument in brackets.
27126 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27127 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27128 (DDR_REVERSED_P): Likewise.
27130 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27132 PR tree-optimization/79472
27133 * tree-switch-conversion.c (struct switch_conv_info): Add
27134 contiguous_range and default_case_nonstandard fields.
27135 (collect_switch_conv_info): Compute contiguous_range and
27136 default_case_nonstandard fields, don't clear final_bb if
27137 contiguous_range and only the default case doesn't have the required
27139 (check_all_empty_except_final): Set default_case_nonstandard instead
27140 of failing if contiguous_range and the default case doesn't have empty
27142 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27143 and only the default case doesn't have the required constants. Skip
27145 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27146 if default_case_nonstandard.
27147 (build_constructors): Build constant 1 just once. Assert that default
27148 values aren't inserted in between cases if contiguous_range. Skip
27150 (build_arrays): Skip virtual phis.
27151 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27152 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27153 Handle virtual phis.
27154 (gen_inbound_check): Handle default_case_nonstandard case.
27155 (process_switch): Adjust check_final_bb caller. Call
27156 gather_default_values with the first non-default case instead of
27157 default case if default_case_nonstandard.
27159 2017-05-02 Nathan Sidwell <nathan@acm.org>
27161 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27162 check. Fix formatting.
27164 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27166 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27167 errors when comparing specialized and unspecialized times.
27169 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27171 * diagnostic-show-locus.c
27172 (layout::should_print_annotation_line_p): Make private.
27173 (layout::print_annotation_line): Make private.
27174 (layout::annotation_line_showed_range_p): Make private.
27175 (layout::show_ruler): Make private.
27176 (layout::print_source_line): Make private. Pass in line and
27177 line_width, rather than calling location_get_source_line. Drop
27179 (layout::print_leading_fixits): New method.
27180 (layout::print_any_fixits): Rename to...
27181 (layout::print_trailing_fixits): ...this, and make private.
27182 Don't print newline fixits.
27183 (diagnostic_show_locus): Move logic for printing one row into...
27184 (layout::print_line): ...this new function. Move the
27185 location_get_source_line call and error-handling from
27186 print_source_line to here. Call print_leading_fixits, and rename
27187 print_any_fixits to print_trailing_fixits.
27188 (selftest::test_fixit_insert_containing_newline): Update now that
27189 newlines are partially supported.
27190 (selftest::test_fixit_insert_containing_newline_2): New test.
27191 (selftest::test_fixit_replace_containing_newline): Update comments.
27192 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27193 * edit-context.c (class added_line): New class.
27194 (class edited_line): Describe newline handling in comment.
27195 (edited_line::actually_edited_p): New method.
27196 (edited_line::print_content): Delete redundant decl.
27197 (edited_line::m_predecessors): New field.
27198 (edited_file::print_content): Call edited_line::print_content.
27199 (edited_file::print_diff): Update to support newlines.
27200 (edited_file::print_diff_hunk): Likewise.
27201 (edited_file::print_run_of_changed_lines): New function.
27202 (edited_file::print_diff_line): Convert to...
27203 (print_diff_line): ...this.
27204 (edited_file::get_effective_line_count): New function.
27205 (edited_line::edited_line): Initialize new field m_predecessors.
27206 (edited_line::~edited_line): Clean up m_predecessors.
27207 (edited_line::apply_fixit): Handle newlines.
27208 (edited_line::get_effective_line_count): New function.
27209 (edited_line::print_content): New function.
27210 (edited_line::print_diff_lines): New function.
27211 (selftest::test_applying_fixits_insert_containing_newline): New
27213 (selftest::test_applying_fixits_replace_containing_newline): New
27215 (selftest::insert_line): New function.
27216 (selftest::test_applying_fixits_multiple_lines): Add example of
27218 (selftest::edit_context_c_tests): Call the new tests.
27220 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27222 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27223 parameter cand. Update dump information.
27224 (get_computation_cost): Update uses.
27226 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27228 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27229 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27230 (get_computation_at, rewrite_use_address): Update use of
27231 get_computation_aff.
27233 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27235 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27236 (get_computation): Delete.
27237 (get_computation_cost): Implement like get_computation_cost_at.
27238 Use get_computation_at.
27239 (get_computation_cost_at): Delete.
27240 (rewrite_use_nonlinear_expr): Use get_computation_at.
27241 (rewrite_use_compare, remove_unused_ivs): Ditto.
27243 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27245 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27247 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27249 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27250 (ivopts_global_cost_for_size): Rename parameter and update uses.
27251 (iv_ca_recount_cost): Update uses.
27252 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27253 candidates seperately in n_invs and n_cands.
27254 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27256 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27258 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27259 (find_inv_vars_cb): New.
27260 (find_depends): Renamed to ...
27261 (find_inv_vars): ... this.
27262 (add_candidate_1, force_var_cost): Call find_inv_vars.
27263 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27265 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27267 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27268 inv_vars. Add inv_exprs.
27269 (struct iv_cand): Rename depends_on to inv_vars.
27270 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27271 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27272 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27273 (dump_cand): Dump inv_vars.
27274 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27275 (record_invariant, find_depends, add_candidate_1): Ditto.
27276 (set_group_iv_cost, force_var_cost): Ditto.
27277 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27278 (get_computation_cost_at, get_computation_cost): Ditto.
27279 (determine_group_iv_cost_generic): Ditto.
27280 (determine_group_iv_cost_address): Ditto.
27281 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27282 (determine_group_iv_costs): Ditto.
27283 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27284 (iv_ca_set_remove_invariants): Renamed to ...
27285 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27286 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27287 (iv_ca_set_add_invariants): Renamed to ...
27288 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27289 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27290 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27291 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27292 (create_new_ivs): Remove useless dump.
27294 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27296 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27298 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27299 (iv_ca_set_no_cp, create_new_iv): Ditto.
27301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27303 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27305 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27307 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27308 function tree_check2.
27310 2017-05-02 Martin Liska <mliska@suse.cz>
27312 * doc/gcov.texi: Add missing preposition.
27313 * gcov.c (function_info::function_info): Properly fill up
27314 all member variables.
27316 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27318 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27320 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27322 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27324 2017-05-02 Martin Liska <mliska@suse.cz>
27327 * lto-streamer-in.c (lto_read_tree_1): Remove
27328 LTO_STREAMER_DEBUG.
27329 * lto-streamer.c (struct tree_hash_entry): Likewise.
27330 (struct tree_entry_hasher): Likewise.
27331 (tree_entry_hasher::hash): Likewise.
27332 (tree_entry_hasher::equal): Likewise.
27333 (lto_streamer_init): Likewise.
27334 (lto_orig_address_map): Likewise.
27335 (lto_orig_address_get): Likewise.
27336 (lto_orig_address_remove): Likewise.
27337 * lto-streamer.h: Likewise.
27338 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27339 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27341 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27343 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27344 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27345 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27346 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27347 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27348 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27349 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27350 (mm_maskz_sub_ss): New intrinsics.
27351 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27352 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27353 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27354 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27355 (__builtin_ia32_subss_mask_round): New builtins.
27356 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27357 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27358 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27360 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27361 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27363 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27366 2017-05-02 Martin Jambor <mjambor@suse.cz>
27368 PR tree-optimization/78687
27369 * tree-sra.c (access): New field parent.
27370 (process_subtree_disqualification): New function.
27371 (disqualify_candidate): Call it.
27372 (build_accesses_from_assign): Reset write flag if creating an
27374 (build_access_subtree): Fill in parent field and also prpagate
27375 down grp_write flag.
27376 (create_artificial_child_access): New parameter set_grp_write, set
27377 grp_write to its value.
27378 (propagate_subaccesses_across_link): Also propagate grp_write flag
27380 (propagate_all_subaccesses): Push the closest parent back to work
27381 queue if add_access_to_work_queue returned true.
27383 2017-05-02 Richard Biener <rguenther@suse.de>
27385 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27386 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27387 -fstrict-overflow documentation.
27388 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27389 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27390 flag_strict_overflow.
27391 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27392 * lto-opts.c (lto_write_options): Do not stream it.
27393 * lto-wrapper.c (merge_and_complain): Do not handle it.
27394 * opts.c (default_options_table): Do not set -fstrict-overflow.
27395 (finish_options): Likewise do not clear it when sanitizing.
27396 * simplify-rtx.c (simplify_const_relational_operation): Do not
27397 test flag_strict_overflow.
27399 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27401 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27402 using enabled attribute.
27403 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27404 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27405 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27406 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27407 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27408 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27409 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27410 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27411 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27412 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27414 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27416 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27418 2017-05-02 Richard Biener <rguenther@suse.de>
27420 PR tree-optimization/80591
27422 2017-04-10 Richard Biener <rguenther@suse.de>
27424 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27427 2017-05-02 Richard Biener <rguenther@suse.de>
27429 PR tree-optimization/80549
27430 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27431 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27432 headers if they do not have a preheader.
27434 2017-05-02 Martin Liska <mliska@suse.cz>
27437 * common.opt: Fix typo.
27438 * doc/invoke.texi: Likewise.
27440 2017-05-01 Jan Beulich <jbeulich@suse.com>
27442 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27443 swapping, add (x,x,m,x,n) alternative.
27445 2017-05-01 Nathan Sidwell <nathan@acm.org>
27447 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27448 unnecessary unadjusted_alignment check.
27450 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27453 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27455 * gimplify.c (gimplify_cilk_detach): New function.
27456 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27457 * tree-core.h: Document EXPR_CILK_SPAWN.
27458 * tree.h (EXPR_CILK_SPAWN): Define.
27460 2017-05-01 David Malcolm <dmalcolm@redhat.com>
27462 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
27463 to use new fixit_hint representation, using the "replace" logic.
27464 (get_line_span_for_fixit_hint): Likewise.
27465 (layout::print_any_fixits): Likewise.
27466 (selftest::test_one_liner_many_fixits): Rename to...
27467 (selftest::test_one_liner_many_fixits_1): ...this, and update
27468 comment and expected output to reflect that the multiple fix-it
27469 hints are now consolidated into one insertion.
27470 (selftest::test_one_liner_many_fixits_2): New test.
27471 (selftest::test_diagnostic_show_locus_one_liner): Update for
27473 (selftest::test_fixit_consolidation): Update for fix-it API
27475 * diagnostic.c (print_parseable_fixits): Likewise.
27476 * edit-context.c (edited_line::m_line_events): Convert from
27477 auto_vec <line_event *> to auto_vec <line_event>.
27478 (class line_event): Convert from abstract base class to a concrete
27479 class, taking over the role of replace_event.
27480 (class insert_event): Delete.
27481 (class replace_event): Rename to class line_event. Convert to
27483 (edit_context::add_fixits): Reimplement.
27484 (edit_context::apply_insert): Delete.
27485 (edit_context::apply_replace): Rename to...
27486 (edit_context::apply_fixit): ...this. Convert to half-open range.
27487 (edited_file::apply_insert): Delete.
27488 (edited_file::apply_replace): Rename to...
27489 (edited_file::apply_fixit): ...this.
27490 (edited_line::~edited_line): Drop deletion of events.
27491 (edited_line::apply_insert): Delete.
27492 (edited_line::apply_replace): Rename to...
27493 (edited_line::apply_fixit): ...this. Convert to half-open range.
27494 Update for change to type of m_line_events.
27495 * edit-context.h (edit_context::apply_insert): Delete.
27496 (edit_context::apply_replace): Rename to...
27497 (edit_context::apply_fixit): ...this.
27499 2017-05-01 Martin Sebor <msebor@redhat.com>
27501 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27504 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27507 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27508 __get_cpuid_max returns 0.
27509 (__get_cpuid_count): Ditto.
27511 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27513 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27514 replacement expression is another instance of one of its arguments.
27516 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27519 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27520 check for stack push/pop autoinc.
27521 * config/i386/i386.c (ix86_agi_dependent): Return false
27522 if the only reason why modified_in_p returned true is that
27523 addr is SP based and set_insn is a push or pop.
27525 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27527 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27530 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27533 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27534 (account_size_time): Use two predicates - exec_pred and
27536 (evaluate_conditions_for_known_args): Compute both clause and
27538 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27539 (inline_summary_t::duplicate): Update.
27540 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27542 (compute_inline_parameters): Likewise.
27543 (estimate_edge_size_and_time): Update caluclation of time.
27544 (estimate_node_size_and_time): Compute both time and nonspecialized
27546 (estimate_ipcp_clone_size_and_time): Update.
27547 (inline_merge_summary): Update.
27548 (do_estimate_edge_time): Update.
27549 (do_estimate_edge_size): Update.
27550 (do_estimate_edge_hints): Update.
27551 (inline_read_section, inline_write_summary): Stream both new predicates.
27552 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27554 (compute_inlined_call_time): Cleanup.
27555 (big_speedup_p): Update.
27556 (edge_badness): Update.
27557 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27558 (size_time_entry): Replace predicate by exec_predicate and
27559 nonconst_predicate.
27560 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27561 (estimate_edge_time): Return also nonspec_time.
27562 (reset_edge_growth_cache): Update.
27564 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27566 PR rtl-optimization/80491
27567 * ifcvt.c (noce_process_if_block): When looking for x setter
27568 with missing else_bb, don't check only the insn right before
27569 cond_earliest, but look for the last insn that x is modified in
27570 within the same bb.
27572 PR rtl-optimization/80491
27573 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27575 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27577 PR tree-optimization/80487
27578 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27580 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27582 PR tree-optimization/79697
27583 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27584 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27585 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27587 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27588 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27590 2017-04-28 Martin Sebor <msebor@redhat.com>
27592 PR tree-optimization/80523
27593 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27594 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27596 (maybe_warn, format_directive, parse_directive): Use new functions.
27597 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27599 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27601 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27603 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27605 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27606 target_header_dir): Set correctly.
27607 * configure: Regenerated.
27608 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27609 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27610 instead of SYSTEM_HEADER_DIR.
27612 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27614 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27615 (estimate_local_effects): Likewise.
27616 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27617 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27618 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27619 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27620 do_estimate_edge_time, estimate_edge_time): Likewise.
27621 * ipa-inline-analysis.c (estimate_node_size_and_time,
27622 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27623 (estimate_time_after_inlining): Remove.
27625 2017-04-28 Martin Liska <mliska@suse.cz>
27627 * doc/gcov.texi: Enhance documentation of gcov.
27629 2017-04-28 Martin Liska <mliska@suse.cz>
27631 * doc/gcov.texi: Sort options in alphabetic order.
27632 * doc/gcov-dump.texi: Likewise.
27633 * doc/gcov-tool.texi: Likewise.
27634 * gcov.c (print_usage): Likewise.
27635 * gcov-dump.c (print_usage): Likewise.
27636 * gcov-tool.c (print_merge_usage_message): Likewise.
27637 (print_rewrite_usage_message): Likewise.
27638 (print_overlap_usage_message): Likewise.
27640 2017-04-28 Martin Liska <mliska@suse.cz>
27642 PR gcov-profile/53915
27643 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27645 2017-04-28 Martin Liska <mliska@suse.cz>
27647 PR gcov-profile/79891
27648 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27649 is marked by compiler as living on a line.
27650 (get_cycles_count): Remove usage of the union.
27651 (output_intermediate_file): Likewise.
27652 (find_source): Fix GNU coding style.
27653 (accumulate_line_counts): Remove old non-all block mode.
27654 (output_lines): Remove usage of the union.
27655 * profile.c (output_location): Include all BBs, even if
27656 belonging to a same line (and file) as a previous BB.
27658 2017-04-28 Martin Liska <mliska@suse.cz>
27660 * gcov.c (process_args): Handle new argument 'w'.
27661 (read_graph_file): Assign ID to BBs.
27662 (output_branch_count): Display BB # if verbose flag is set.
27663 (output_lines): Likewise for arcs.
27664 (print_usage): Add '--verbose' option help.
27665 * doc/gcov.texi: Document --verbose (-w) option.
27667 2017-04-28 Martin Liska <mliska@suse.cz>
27669 * gcov.c (struct block_location_info): New struct.
27670 (process_file): Fill up the new structure.
27671 (read_graph_file): Replace usage of encoding by the newly added
27673 (add_line_counts): Likewise.
27674 (accumulate_line_counts): Remove usage of the union.
27675 (function_info::function_info): New function.
27676 (function_info::~function_info): Likewise.
27677 (process_file): Call delete instead of release_function.
27678 (release_function): Release the function.
27679 (release_structures): Call delete instead of release_function.
27680 (solve_flow_graph): Replace usage of num_blocks.
27681 (find_exception_blocks): Likewise.
27682 (output_lines): Fix GNU coding style.
27684 2017-04-28 Martin Liska <mliska@suse.cz>
27687 * coverage.c (coverage_remove_note_file): New function.
27688 * coverage.h: Declare the function.
27689 * toplev.c (finalize): Clean if an error has been seen.
27691 2017-04-28 Martin Liska <mliska@suse.cz>
27693 PR gcov-profile/80031
27694 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27695 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27696 * gcov.c (read_graph_file): Read just number of blocks.
27697 * profile.c (branch_prob): Do not stream 0 flags per a basic
27700 2017-04-28 Martin Liska <mliska@suse.cz>
27702 * gcov-dump.c (tag_*): Add new argument to declarations.
27703 (dump_gcov_file): Likewise.
27704 (tag_blocks): Add and use new argument depth.
27705 (tag_arcs): Likewise.
27706 (tag_lines): Likewise.
27707 (tag_counters): Likewise.
27708 (tag_summary): Likewise.
27709 (dump_working_sets): Use depth to do a proper indentation.
27711 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27714 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27715 * symtab.c (symtab_node::debug_symtab): Move definition here.
27717 2017-04-28 Richard Biener <rguenther@suse.de>
27719 * lto-streamer.h (LTO_major_version): Bump to 7.
27721 2017-04-28 Richard Biener <rguenther@suse.de>
27723 * tree-vrp.c (assert_info): New struct.
27724 (add_assert_info): New helper.
27725 (register_edge_assert_for_2): Refactor to add asserts to a vector
27727 (register_edge_assert_for_1): Likewise.
27728 (register_edge_assert_for): Likewise.
27729 (finish_register_edge_assert_for): New helper actually registering
27730 asserts where live on edge.
27731 (find_conditional_asserts): Adjust.
27732 (find_switch_asserts): Likewise.
27733 (evrp_dom_walker::try_find_new_range): Generalize.
27734 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27736 2017-04-27 Marek Polacek <polacek@redhat.com>
27739 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27740 arg10 and arg11 to itype.
27742 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27744 * doc/extend.texi (Object Size Checking): Improve grammar.
27746 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27749 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27750 that the logic for permitting reciprocal estimates matches that
27753 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27756 * tree.c (type_cache_hasher::equal): Only compare
27757 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27758 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27759 non-aggregate element types.
27760 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27761 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27763 2017-04-27 Richard Biener <rguenther@suse.de>
27765 PR middle-end/80533
27766 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27767 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27768 keeping a reference to a trailing array.
27770 2017-04-27 Richard Biener <rguenther@suse.de>
27772 PR middle-end/80539
27773 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27774 being in loop-closed SSA form conservatively.
27775 (chrec_fold_multiply_poly_poly): Likewise.
27777 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27779 PR middle-end/79665
27780 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27781 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27783 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27786 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27787 (aarch64_function_arg_alignment): Return unsigned int again, but still
27788 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27789 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27790 Don't emit -Wpsabi note.
27791 (aarch64_function_arg_boundary): Likewise.
27792 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27795 2017-04-26 Nathan Sidwell <nathan@acm.org>
27797 * tree.h (crc32_unsigned_n): Declare.
27798 (crc32_unsigned, crc32_unsigned): Make inline.
27799 * tree.c (crc32_unsigned_bits): Replace with ...
27800 (crc32_unsigned_n): ... this.
27801 (crc32_unsigned, crc32_byte): Remove.
27802 (crc32_string): Remove unnecessary braces.
27804 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27806 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27807 * ipa-inline-analysis.c (MAX_TIME): Remove.
27808 (account_size_time): Use sreal for time.
27809 (dump_inline_summary): Update.
27810 (estimate_function_body_sizes): Update.
27811 (estimate_edge_size_and_time): Update.
27812 (estimate_calls_size_and_time): Update.
27813 (estimate_node_size_and_time): Update.
27814 (inline_merge_summary): Update.
27815 (inline_update_overall_summary): Update.
27816 (estimate_time_after_inlining): Update.
27817 (inline_read_section): Update.
27818 (inline_write_summary): Update.
27819 * ipa-inline.c (compute_uninlined_call_time): Update.
27820 (compute_inlined_call_time): Update.
27821 (recursive_inlining): Update.
27822 (inline_small_functions): Update.
27823 (dump_overall_stats): Update.
27824 * ipa-inline.h: Include sreal.h.
27825 (size_time_entry): Turn time to sreal.
27826 (inline_summary): Turn self_time nad time to sreal.
27828 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27830 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27832 (sreal::stream_out, sreal::stream_in): New.
27833 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27835 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27837 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27840 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27843 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27845 (dimode_scalar_chain::compute_convert_gain): Ditto.
27846 (dimode_scalar_chain::make_vector_copies): Ditto.
27847 (dimode_scalar_chain::convert_reg): Ditto.
27848 (dimode_scalar_chain::convert_insn): Ditto.
27849 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27850 (VI248_AVX512BW_1): New mode iterator.
27851 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27852 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27855 2017-04-25 Martin Sebor <msebor@redhat.com>
27857 PR tree-optimization/80497
27858 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27859 constants are representable in HOST_WIDE_INT.
27860 (parse_directive): Ditto.
27862 2017-04-25 Martin Sebor <msebor@redhat.com>
27865 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27866 (new_zero_array): Adjust signature.
27867 (dom_info::dom_init): Used unsigned rather that size_t.
27868 (dom_info::dom_info): Same.
27870 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27871 Jakub Jelinek <jakub@redhat.com>
27874 * config/arm/arm.c: Include gimple.h.
27875 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27876 returns negative, increment ncrn only if it returned positive.
27877 (arm_needs_doubleword_align): Return int instead of bool,
27878 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27879 members, but if there is any such non-FIELD_DECL
27880 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27881 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27882 returns negative, increment nregs only if it returned positive.
27883 (arm_setup_incoming_varargs): Likewise.
27884 (arm_function_arg_boundary): Emit -Wpsabi note if
27885 arm_needs_doubleword_align returns negative, return
27886 DOUBLEWORD_ALIGNMENT only if it returned positive.
27888 2017-04-25 Marek Polacek <polacek@redhat.com>
27891 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27892 first argument to type.
27894 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27897 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27898 type checks to test for compatibility instead of equality.
27900 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27901 Jakub Jelinek <jakub@redhat.com>
27904 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27906 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27907 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27908 the alignment computation, but return their maximum in warn_alignment.
27909 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27910 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27912 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27913 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27916 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27918 * config/arc/simdext.md (dmpyh): Fix typo.
27920 2017-04-25 Richard Biener <rguenther@suse.de>
27922 PR tree-optimization/80492
27923 * alias.c (compare_base_decls): Handle registers with asm
27924 specification conservatively.
27925 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27926 compare_base_decls returning dont-know properly.
27928 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27930 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27931 (legitimate_offset_address_p): New function.
27932 (arc_legitimate_address_p): Use above function.
27934 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27936 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27938 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27940 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27941 ACCH registers whenever they are available.
27943 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27945 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27946 double regs fix when not used.
27948 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27950 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27952 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27953 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27955 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27957 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27958 when emitting short ADD instructions.
27960 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27962 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27964 (cmpsi_cc_c_insn): Likewise.
27965 (cbranchsi4_scratch): Compute proper instruction length using
27966 compact_hreg_operand.
27967 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27969 2017-04-25 Richard Biener <rguenther@suse.de>
27971 PR middle-end/80509
27972 * passes.c (pass_manager::pass_manager): Initialize
27973 m_name_to_pass_map.
27975 2017-04-25 Richard Biener <rguenther@suse.de>
27977 PR tree-optimization/79201
27978 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27980 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27983 * config/s390/vector.md: Split MEM->GPR vector moves for
27984 non-s_operand addresses.
27986 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27989 * config/s390/predicates.md (reload_const_wide_int_operand): New
27991 * config/s390/s390.md ("movti"): Remove d/P alternative.
27992 ("movti_bigconst"): New pattern definition.
27994 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27997 * s390-protos.h (s390_expand_cs_hqi): Removed.
27998 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27999 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28000 modes as well as CCZ1mode and CCZmode.
28001 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28002 signature of s390_emit_compare_and_swap.
28003 (s390_expand_cs_hqi): Likewise, make static.
28004 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28005 compare-and-swap, in some cases.
28006 (s390_expand_cs): Wrapper function.
28007 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28009 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28010 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28011 patterns for small and large integers. Forbid symref memory operands.
28012 Move expander to s390.c. Require cc register.
28013 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28014 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28015 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28016 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28017 symref memory operands. Remove CC mode and call s390_match_ccmode
28019 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28021 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28023 * config/s390/s390.md (define_peephole2): New peephole to help
28024 combining the load-and-test pattern with volatile memory.
28026 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28028 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28029 with CCZmode for TARGET_Z196.
28031 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28033 PR rtl-optimization/80501
28034 * combine.c (make_compound_operation_int): Set subreg_code to SET
28035 even for AND with mask of the sign bit of mode.
28037 PR rtl-optimization/80500
28038 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28039 sum's initial value.
28041 2017-04-25 Julian Brown <julian@codesourcery.com>
28042 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28044 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28046 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28048 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28050 2017-04-25 Julian Brown <julian@codesourcery.com>
28051 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28053 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28054 (thunderx2t99_sha): New Reservation.
28056 2017-04-25 Julian Brown <julian@codesourcery.com>
28057 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28059 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28060 type for 1-element load.
28062 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28064 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28066 2017-04-24 Martin Jambor <mjambor@suse.cz>
28068 PR tree-optimization/80293
28069 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28070 char arrays not totally scalarizable if it is false.
28071 (analyze_all_variable_accesses): Pass correct value in the new
28072 parameter. Add a statistics counter.
28074 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28076 PR middle-end/79931
28077 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28079 2017-04-24 Richard Biener <rguenther@suse.de>
28081 PR tree-optimization/80494
28082 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28083 out for complex types.
28085 2017-04-24 Richard Biener <rguenther@suse.de>
28087 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28088 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28089 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28090 (DFS): Adjust and never fail.
28091 (sccvn_dom_walker::fail): Remove.
28092 (sccvn_dom_walker::before_dom_children): Adjust.
28093 (run_scc_vn): Likewise and never fail.
28094 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28095 (pass_fre::execute): Likewise.
28097 2017-04-24 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/79725
28100 * tree-ssa-sink.c (statement_sink_location): Return whether
28101 failure reason was zero uses. Move that check later.
28102 (sink_code_in_bb): Deal with zero uses by removing the stmt
28105 2017-04-24 Richard Biener <rguenther@suse.de>
28108 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28109 pointer-based references.
28111 2017-04-24 Richard Biener <rguenther@suse.de>
28114 * pass_manager.h (pass_manager::operator new): Remove.
28115 (pass_manager::operator delete): Likewise.
28116 * passes.c (pass_manager::operator new): Remove.
28117 (pass_manager::operator delete): Likewise.
28118 (pass_manager::pass_manager): Zero individual pass members.
28120 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28123 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28124 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28125 Check "XEXP (src, 1)" operand here.
28126 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28127 Check "XEXP (src, 1)" operand here.
28128 (dimode_scalar_chain::make_vector_copies): Detect count register
28129 of a shift instruction. Zero extend count register from QImode
28130 to DImode to satisfy vector shift pattern count operand predicate.
28131 Substitute vector shift count operand with a DImode copy.
28132 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28135 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28137 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28138 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28139 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28140 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28141 (UNSPEC_NOREX_MEM): Remove definition.
28143 2017-04-21 Richard Biener <rguenther@suse.de>
28145 PR tree-optimization/79547
28146 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28147 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28148 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28149 without any constraints.
28151 2017-04-21 Richard Biener <rguenther@suse.de>
28153 PR tree-optimization/78847
28154 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28156 2017-04-21 Richard Biener <rguenther@suse.de>
28158 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28159 (build_distinct_type_copy): Likewise.
28160 (build_variant_type_copy): Likewise.
28161 * tree.c (build_qualified_type): Pass down mem-stat info.
28162 (build_distinct_type_copy): Likewise.
28163 (build_variant_type_copy): Likewise.
28165 2017-04-21 Richard Biener <rguenther@suse.de>
28167 PR tree-optimization/80237
28168 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28170 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28171 for a simplified result.
28173 2016-04-21 Richard Biener <rguenther@suse.de>
28175 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28176 sth as strict as a simple_iv but a chrec without symbols and an
28177 operand defined in the loop we are peeling (and not some subloop).
28178 (propagate_constants_for_unrolling): Propagate all constants.
28180 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28183 * config/i386/i386.c (print_reg): Remove assert for disalowed
28184 regno values, call output_operand_lossage instead.
28186 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28189 * config/i386/constraints.md (Yc): New register constraint.
28190 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28191 Use Yc constraint for alternative 2 of operand 0. Remove
28192 preferred_for_speed attribute.
28194 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28196 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28197 lastprivate clauses in SIMT case.
28199 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28201 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28203 2017-04-20 Richard Biener <rguenther@suse.de>
28205 PR tree-optimization/57796
28206 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28207 as N scalar stores.
28208 (vect_model_load_cost): Cost gathers as N scalar loads.
28210 2017-04-20 Richard Biener <rguenther@suse.de>
28212 * ggc-page.c (ggc_allocated_p): Rename to ...
28213 (safe_lookup_page_table_entry): ... this and return the lookup
28215 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28217 2017-04-20 Richard Biener <rguenther@suse.de>
28219 PR tree-optimization/80453
28220 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28221 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28222 from the conditions.
28223 (vn_phi_eq): Pass them down.
28224 (vn_phi_lookup): Record them.
28225 (vn_phi_insert): Likewise.
28227 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28229 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28230 uninitialized variable warning to avoid buffer overrun.
28232 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28235 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28236 is suppressed for '{ 0 }' in C.
28238 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28240 * BASE-VER: Set to 8.0.0.
28242 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28244 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28245 priority .init_array and .fini_array section with SECTION_NOTYPE
28248 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28250 PR middle-end/80423
28251 * tree.h (build_array_type): Add typeless_storage default argument.
28252 * tree.c (type_cache_hasher::equal): Also compare
28253 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28254 (build_array_type): Add typeless_storage argument, set
28255 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28257 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28258 (build_array_type): Likewise. Add typeless_storage argument.
28260 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28261 Jakub Jelinek <jakub@redhat.com>
28263 PR tree-optimization/80426
28264 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28265 operation on symbolic operands, also compute the overflow for the
28266 invariant part when the operation degenerates into a negation.
28268 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28271 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28272 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28275 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28277 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28280 * config/avr/avr.c (tree.h): Include it.
28281 (cgraph.h): Include it.
28282 (avr_encode_section_info): Don't warn for uninitialized progmem
28283 variable if it's just an alias.
28285 2017-04-19 Richard Biener <rguenther@suse.de>
28288 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28289 when needed by AutoPGO.
28291 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28294 * doc/lto.texi: Remove an extra 'that'.
28296 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28298 PR rtl-optimization/80429
28299 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28300 are only used in debug insns.
28302 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28303 Vladimir Makarov <vmakarov@redhat.com>
28305 * config/sparc/predicates.md (input_operand): Add comment. Return
28306 true for any memory operand when LRA is in progress.
28307 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28309 2017-04-18 Jeff Law <law@redhat.com>
28312 * mips.md ({return,simple_return}_internal): Do not overwrite
28315 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28317 PR tree-optimization/80443
28318 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28319 instead of adding 1, subtract -1 and similarly instead of subtracting
28322 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28324 PR rtl-optimization/80357
28325 * haifa-sched.c (tmp_bitmap): New variable.
28326 (model_recompute): Handle duplicate use records.
28327 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28328 (free_global_sched_pressure_data): Free it.
28330 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28333 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28334 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28335 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28336 instead of SYSTEM_HEADER_DIR.
28338 2017-04-18 Jeff Law <law@redhat.com>
28340 PR middle-end/80422
28341 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28342 predecessors after walking up the insn chain.
28344 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28347 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28348 sizetype type into debug info.
28350 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28353 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28354 unneeded test for TARGET_UPPER_REGS_SF.
28355 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28357 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28360 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28361 instead of gsi_after_labels.
28363 2017-04-18 Jeff Law <law@redhat.com>
28365 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28369 2017-04-13 Jeff Law <law@redhat.com>
28370 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28371 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28373 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28376 * config/avr/avr.c (intl.h): Include it.
28377 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28379 2017-04-18 Martin Liska <mliska@suse.cz>
28381 PR gcov-profile/78783
28382 * gcov-tool.c (gcov_output_files): Validate that destination
28383 file is either removed by the tool or by a user.
28385 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28386 Guy Benyei <guybe@mellanox.com>
28388 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28389 block, and do not negate it, the stored id is already negative.
28391 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28393 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28395 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28398 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28399 masks of options that should be turned off if the VSX vector
28400 options are turned off.
28401 (OTHER_P8_VECTOR_MASKS): Likewise.
28402 (OTHER_VSX_VECTOR_MASKS): Likewise.
28403 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28404 rs6000_disable_incompatible_switches to validate no type switches
28406 (rs6000_incompatible_switch): New function to disallow turning on
28407 other vector options if -mno-vsx, -mno-power8-vector, or
28408 -mno-power9-vector are specified.
28410 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28412 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28414 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28416 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28417 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28418 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28419 (ARG_POINTER_CFA_OFFSET): Likewise.
28421 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28423 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28424 conditions to take advantage of various optimizations.
28426 2017-04-13 Jeff Law <law@redhat.com>
28428 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28429 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28430 (zero_extendsidi2_dext): Likewise.
28432 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28435 * fold-const.c (fold_ternary_loc): Revert
28436 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28439 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28441 PR rtl-optimization/80343
28442 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28443 new scratch pseudo.
28445 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28448 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28449 to ubsan_encode_value.
28451 2017-04-13 Jeff Law <law@redhat.com>
28453 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28454 appearing in DEBUG_INSNs.
28456 2017-04-13 Martin Liska <mliska@suse.cz>
28458 PR gcov-profile/80413
28459 * gcov-io.c (gcov_write_string): Copy to buffer just when
28460 allocated size is greater than zero.
28462 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28465 * dwarf2out.c (decls_for_scope): Ignore declarations of
28466 current_function_decl in BLOCK_NONLOCALIZED_VARS.
28468 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
28471 * ipa-visibility.c (non_local_p): Fix typos.
28472 (localize_node): When localizing symbol in same comdat group,
28473 dissolve the group only when we know external symbols are going
28475 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
28477 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28479 PR tree-optimization/79390
28480 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
28481 order does not result in usable sequence, retry with reversed operand
28487 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
28488 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
28489 op0 instead of fold_convert_loc (loc, type, arg0).
28491 2017-04-12 Jeff Law <law@redhat.com>
28493 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28494 has a delay slot in the generated code.
28496 * config/cris/cris.md (cris_preferred_reload_class): Return
28497 GENNONACR_REGS rather than GENERAL_REGS.
28499 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28502 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28503 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28504 signedness of the result type.
28506 2017-04-12 Richard Biener <rguenther@suse.de>
28507 Jeff Law <law@redhat.com>
28509 PR tree-optimization/80359
28510 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28511 trim stores to TARGET_MEM_REFs.
28513 2017-04-12 Richard Biener <rguenther@suse.de>
28515 PR tree-optimization/79390
28516 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28517 threading case even more.
28519 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28522 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28523 for quad_address_p for TImode, instead of just not indexed_address.
28525 2017-04-12 Richard Biener <rguenther@suse.de>
28526 Bernd Edlinger <bernd.edlinger@hotmail.de>
28528 PR middle-end/79671
28529 * alias.c (component_uses_parent_alias_set_from): Handle
28530 TYPE_TYPELESS_STORAGE.
28531 (get_alias_set): Likewise.
28532 * tree-core.h (tree_type_common): Add typeless_storage flag.
28533 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28534 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28535 for types containing members with TYPE_TYPELESS_STORAGE.
28536 (place_field): Likewise.
28537 (layout_type): Likewise for ARRAY_TYPE.
28538 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28539 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28540 TYPE_TYPELESS_STORAGE.
28541 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28543 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28546 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28547 first argument to type.
28549 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28553 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28554 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28555 (rs6000_expand_binop_builtin): Likewise.
28556 (rs6000_expand_ternop_builtin): Likewise; also add missing
28557 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28558 vshasigma built-ins.
28559 * doc/extend.texi: Document that vec_xxpermdi's third argument
28560 must be a constant.
28562 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28564 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28565 Use shift_const cost parameter when calculating gain of STV shifts.
28567 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28569 PR rtl-optimization/70478
28570 * lra-constraints.c (process_alt_operands): Check memory for
28571 disfavoring memory insn operand.
28573 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28575 PR middle-end/80100
28576 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28577 left shift in unsigned HOST_WIDE_INT type.
28579 PR rtl-optimization/80385
28580 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28581 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28584 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28585 if they have any depend clauses.
28587 2017-04-11 Martin Liska <mliska@suse.cz>
28590 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28591 * ipa-split.c (split_function): Create a local comdat symbol
28592 if caller is in a comdat group.
28594 2017-04-11 Martin Liska <mliska@suse.cz>
28597 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28600 2017-04-11 Martin Sebor <msebor@redhat.com>
28602 PR middle-end/80364
28603 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28604 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28606 (directive::set_width, directive::set_precision, format_character):
28608 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28611 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28614 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28615 conflict, set target->arch_name instead of target->cpu_name.
28617 2017-04-11 Richard Biener <rguenther@suse.de>
28619 PR tree-optimization/80374
28620 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28621 build_zero_cst, remove fold_convertible_p check again.
28623 2017-04-11 Martin Liska <mliska@suse.cz>
28626 * ubsan.c (instrument_object_size): Do not instrument register
28629 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28632 * config/i386/i386-builtin-types.def
28633 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28634 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28635 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28636 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28637 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28638 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28639 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28640 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28641 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28642 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28643 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28644 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28645 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28646 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28647 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28648 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28649 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28650 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28651 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28652 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28653 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28654 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28655 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28656 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28657 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28658 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28659 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28660 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28661 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28662 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28663 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28664 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28665 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28666 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28667 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28668 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28669 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28670 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28671 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28672 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28673 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28674 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28675 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28676 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28677 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28678 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28680 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28681 flag to second_arg_count, handle 4 argument function type _COUNT
28682 aliases, handle second_arg_count on second argument rather than last.
28684 2017-04-10 Jeff Law <law@redhat.com>
28686 PR tree-optimization/80374
28687 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28688 record anything if we can not convert integer_zero_node to the
28691 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28694 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28695 Enhance special handling given to the TARGET_P9_MINMAX option in
28696 relation to certain other options.
28698 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28700 PR tree-optimization/80153
28701 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28702 remove POINTER_PLUS_EXPR's base part directly, rather than through
28705 2017-04-10 Richard Biener <rguenther@suse.de>
28706 Bin Cheng <bin.cheng@arm.com>
28708 PR tree-optimization/80153
28709 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28710 the first element of pointer type aff_tree. Build result expr in
28712 (add_elt_to_tree): Convert to type unconditionally. Remove other
28713 fold_convert calls.
28714 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28715 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28717 2017-04-10 Richard Biener <rguenther@suse.de>
28719 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28722 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28724 PR rtl-optimization/70478
28725 * lra-constraints.c (curr_small_class_check): New.
28726 (update_and_check_small_class_inputs): New.
28727 (process_alt_operands): Update curr_small_class_check. Disfavor
28728 alternative insn memory operands. Check available regs for small
28731 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28734 * config/mips/mips.opt (-mvirt): Update description.
28735 * doc/invoke.texi (-mvirt): Likewise.
28737 2017-04-10 Richard Biener <rguenther@suse.de>
28739 PR middle-end/80362
28740 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28741 looking for NEGATE_EXPR in -A / -B to A / B folding.
28743 2017-04-10 Martin Liska <mliska@suse.cz>
28745 PR gcov-profile/80224
28746 * gcov.c (print_usage): Fix usage string.
28747 (get_gcov_intermediate_filename): Remove.
28748 (output_gcov_file): Use both for normal and intermediate format.
28749 (generate_results): Do not initialize special file for
28750 intermediate format.
28752 2017-04-10 Richard Biener <rguenther@suse.de>
28754 PR tree-optimization/80304
28755 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28758 2017-04-10 Nathan Sidwell <nathan@acm.org>
28761 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28762 (rs6000_init_builtins): Use it.
28764 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28766 * config/arm/arm.md (<mrc>): Add mode to SET source.
28767 (<mrrc>): Likewise.
28769 2017-04-10 Richard Biener <rguenther@suse.de>
28771 PR middle-end/80344
28772 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28774 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28777 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28778 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28779 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28780 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28781 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28782 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28783 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28784 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28785 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28786 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28787 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28788 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28789 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28790 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28791 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28792 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28793 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28794 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28795 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28796 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28797 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28798 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28799 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28801 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28803 PR rtl-optimization/70478
28804 * lra-constraints.c: Reverse the last patch.
28806 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28808 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28809 Add comment for WCHAR_T.
28811 2017-04-08 Martin Liska <mliska@suse.cz>
28814 2017-04-07 Martin Liska <mliska@suse.cz>
28817 * ipa-split.c (split_function): Add function part to a same comdat
28820 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28823 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28825 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28827 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28828 (vec_pairing): ...to this.
28829 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28830 (rs6000_sched_init): Adjust for name change.
28831 (struct rs6000_sched_context): Likewise.
28832 (rs6000_init_sched_context): Likewise.
28833 (rs6000_set_sched_context): Likewise.
28835 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28841 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28843 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28844 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28845 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28847 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28849 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28851 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28853 PR rtl-optimization/70703
28854 * ira-color.c (update_conflict_hard_regno_costs): Use
28855 int64_t instead of HOST_WIDE_INT.
28857 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28859 PR rtl-optimization/70478
28860 * lra-constraints.c (process_alt_operands): Disfavor alternative
28861 insn memory operands.
28863 2017-04-07 Jeff Law <law@redhat.com>
28865 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28866 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28868 2017-04-07 Martin Liska <mliska@suse.cz>
28871 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28872 Show error message instead of an ICE.
28874 2017-04-07 Martin Liska <mliska@suse.cz>
28877 * ipa-split.c (split_function): Add function part to a same comdat
28880 2017-04-07 Richard Biener <rguenther@suse.de>
28882 PR middle-end/80341
28883 * tree.c (get_unwidened): Also handle ! for_type case for
28885 * convert.c (do_narrow): Split out from ...
28886 (convert_to_integer_1): ... here. Do not pass final truncation
28887 type to get_unwidened for TRUNC_DIV_EXPR.
28889 2017-04-07 Richard Biener <rguenther@suse.de>
28891 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28893 (aff_combination_const): Adjust.
28894 (aff_combination_scale): Likewise.
28895 (aff_combination_add_elt): Likewise.
28896 (aff_combination_add_cst): Likewise.
28897 (aff_combination_convert): Likewise.
28898 (add_elt_to_tree): Likewise. Remove unused argument.
28899 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28901 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28903 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28905 * config/arm/arm.c (arm_default_short_enums): Use
28906 ARM_DEFAULT_SHORT_ENUMS.
28907 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28909 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28912 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28913 members with redundant out-of-class redeclaration.
28915 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28918 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28919 * config/i386/i386.md (*zero_extendsidi2):
28920 Add (?*x,*x) and (?*v,*v) alternatives.
28922 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28925 * config/i386/i386.c (ix86_expand_builtin)
28926 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28927 mode from insn data. Convert operands to insn operand mode.
28928 Copy operands that don't satisfy insn predicate to a register.
28930 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28932 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28935 2017-04-06 Richard Biener <rguenther@suse.de>
28937 PR tree-optimization/80334
28938 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28939 preserve alignment of accesses.
28941 2017-04-06 Richard Biener <rguenther@suse.de>
28943 PR tree-optimization/80262
28944 * tree-sra.c (build_ref_for_offset): Preserve address-space
28946 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28947 Drop useless address-space information on MEM_REF offsets.
28949 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28951 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28953 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28955 PR rtl-optimization/70703
28956 * ira-color.c (update_conflict_hard_regno_costs): Use
28957 HOST_WIDE_INT instead of long.
28959 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28962 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28963 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28965 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28966 for x86_64 target. Handle -m3dnowa option.
28968 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28970 PR rtl-optimization/70703
28971 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28972 (update_conflict_hard_regno_costs): Use long instead of unsigned
28973 arithmetic for cost calculation.
28975 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28976 Bernd Edlinger <bernd.edlinger@hotmail.de>
28979 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28982 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28985 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28986 ptr_mode with Pmode throughout.
28987 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28988 into probe_stack_range and use DImode.
28990 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28993 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28994 call_eh_return is true.
28996 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28998 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28999 Initialize last_match_fntype_index.
29001 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29004 * tree-nvr.c: Include internal-fn.h.
29005 (pass_return_slot::execute): Ignore internal calls without
29008 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29009 Richard Biener <rguenther@suse.de>
29012 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29013 captures used multiple times, except for the last use.
29014 * generic-match-head.c: Include gimplify.h.
29016 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29018 PR tree-optimization/79390
29019 * target.h (struct noce_if_info): Declare.
29020 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29021 * target.def (noce_conversion_profitable_p): New target hook.
29022 * ifcvt.h (struct noce_if_info): New type, moved from ...
29023 * ifcvt.c (struct noce_if_info): ... here.
29024 (noce_conversion_profitable_p): Renamed to ...
29025 (default_noce_conversion_profitable_p): ... this. No longer
29027 (noce_try_store_flag_constants, noce_try_addcc,
29028 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29029 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29030 instead of noce_conversion_profitable_p.
29031 * config/i386/i386.c: Include ifcvt.h.
29032 (ix86_option_override_internal): Don't override
29033 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29034 (ix86_noce_conversion_profitable_p): New function.
29035 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29036 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29037 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29038 * doc/tm.texi: Regenerated.
29040 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29042 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29045 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29048 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29049 instructions for small multiply cores.
29051 2017-04-04 Jeff Law <law@redhat.com>
29053 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29055 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29056 that are not set by the loop over the elements.
29058 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29061 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29062 int mode, convert_modes it to mode as unsigned, otherwise use
29063 lowpart_subreg to mode rather than SImode.
29064 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29065 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29066 Use DImode instead of SImode for the shift count operand.
29067 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29070 2017-04-04 Richard Biener <rguenther@suse.de>
29072 PR middle-end/80281
29073 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29074 arithmetic done for the negate or the plus. Simplify.
29075 (A - (-B) -> A + B): Likewise.
29076 * fold-const.c (split_tree): Make sure to not negate pointers.
29078 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29080 PR rtl-optimization/60818
29081 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29082 a compare of comparisons with the thing compared if this results
29083 in a different machine mode.
29085 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29087 * alias.c (base_alias_check): Fix typo in comment.
29088 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29089 * cgraphunit.c (symbol_table::compile): Likewise.
29090 * collect2.c (maybe_run_lto_and_relink): Likewise.
29091 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29092 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29093 * config/avr/avr.c (avr_map_op_t): Likewise.
29094 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29095 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29096 * config/epiphany/epiphany.md (movcc): Likewise.
29097 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29098 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29100 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29101 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29102 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29103 * config/sh/sh.c (sh_rtx_costs): Likewise.
29104 * fold-const.c (fold_truth_andor): Likewise.
29105 * genautomata.c (collapse_flag): Likewise.
29106 * gengtype.h (struct type::u::s): Likewise.
29107 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29108 * input.c (FORMAT_AMOUNT): Likewise.
29109 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29110 (known_aggs_to_agg_replacement_list): Likewise.
29111 * ipa-inline-analysis.c: Likewise.
29112 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29113 * ipa-polymorphic-call.c
29114 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29115 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29116 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29118 * modulo-sched.c (apply_reg_moves): Likewise.
29119 * omp-expand.c (build_omp_regions_1): Likewise.
29120 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29121 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29122 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29123 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29124 * value-prof.c: Likewise.
29125 * var-tracking.c (val_reset): Likewise.
29127 2017-04-03 Richard Biener <rguenther@suse.de>
29129 PR tree-optimization/80275
29130 * fold-const.c (split_address_to_core_and_offset): Handle
29133 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29135 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29136 descriptors is at least equal to that of functions.
29138 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29140 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29142 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29145 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29146 (mov<IMOD4:mode>): New expander.
29147 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29149 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29151 PR rtl-optimization/79405
29152 * fwprop.c (propagations_left): New variable.
29153 (forward_propagate_into): Decrement it.
29154 (fwprop_init): Initialize it.
29155 (fw_prop): If the variable has reached zero, stop propagating.
29156 (fwprop_addr): Ditto.
29158 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29161 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29162 a FUNCTION_DECL, pass it as decl instead of origin to
29165 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29167 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29170 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29173 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29174 TARGET_VSX_SMALL_INTEGER.
29176 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29178 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29179 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29181 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29183 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29184 extraction from odd-numbered MSA register.
29186 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29188 PR middle-end/80173
29189 * expmed.c (store_bit_field_1): Don't attempt to create
29190 a word subreg out of hard registers wider than word if they
29191 have HARD_REGNO_NREGS of 1 for their mode.
29193 PR middle-end/80163
29194 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29195 conversions to integer types wider than word and pointer.
29198 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29199 (rtx_equal_for_cselib_p): Pass 0 to it.
29200 * cselib.c (cselib_hasher::equal): Likewise.
29201 (rtx_equal_for_cselib_1): Add depth argument. If depth
29202 is 128, don't look up VALUE locs and punt. Increment
29203 depth in recursive calls when walking VALUE locs.
29205 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29207 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29208 (make_gcov_file_name): Use the canonical path name for generating
29210 (read_line): Fix handling of files with ascii null bytes.
29212 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29214 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29215 to initialise a vector register instead
29216 of using a const_int.
29218 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29220 PR translation/80189
29221 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29222 diagnostic messages.
29224 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29227 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29228 (dfp_diex_<mode>): Update mode of operand 1.
29229 * doc/extend.texi (dxex, dxexq): Document change to return type.
29230 (diex, diexq): Document change to argument type.
29232 2017-03-30 Martin Jambor <mjambor@suse.cz>
29235 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29236 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29237 it reflects the signature changes performed at the callee side.
29238 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29239 to cgraph_build_function_type_skip_args.
29240 (build_function_decl_skip_args): Adjust call to the above function.
29242 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29245 * config/i386/sse.md
29246 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29247 register as dest whenever it is a MEM not rtx_equal_p to the
29248 corresponding dup operand, and when forcing into reg move the
29249 reg into the memory afterwards.
29250 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29251 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29252 for the force_reg mode.
29253 (avx512vl_vextractf128<mode>): Use register as dest either
29254 always when a MEM, or when it is a MEM not rtx_equal_p to the
29255 corresponding dup operand, or even not when it is a CONST_VECTOR
29256 depending on the mode and lo vs. hi.
29257 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29259 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29260 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29261 Likewise. Require that operands[2] is even.
29262 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29263 Remove extraneous parens. Require that operands[2] is a multiple
29265 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29266 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29267 disallow memory then.
29269 2017-03-30 Richard Biener <rguenther@suse.de>
29271 PR tree-optimization/77498
29272 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29273 to non-constants over backedges.
29275 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29277 PR rtl-optimization/80233
29278 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29279 as last_combined_insn. Do not test for BARRIER_P separately.
29281 2017-03-29 Andreas Schwab <schwab@suse.de>
29284 * calls.c (prepare_call_address): Convert funexp to Pmode before
29285 copying to temp reg.
29287 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29289 PR tree-optimization/80158
29290 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29291 Handle possible future case of more than one alternate
29293 (replace_rhs_if_not_dup): Likewise.
29294 (replace_one_candidate): Likewise.
29296 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29298 PR rtl-optimization/80193
29299 * ira.c (ira): Do not check allocation for LRA.
29301 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29303 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29304 (nvptx_output_simt_exit): Declare.
29305 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29306 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29307 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29308 (nvptx_declare_function_name): ...here. Emit declaration of local
29309 memory space buffer for omp_simt_enter insn.
29310 (nvptx_output_unisimt_switch): New.
29311 (nvptx_output_softstack_switch): New.
29312 (nvptx_output_simt_enter): New.
29313 (nvptx_output_simt_exit): New.
29314 * config/nvptx/nvptx.h (struct machine_function): New fields
29315 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29316 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29317 (UNSPECV_SIMT_EXIT): Ditto.
29318 (omp_simt_enter_insn): New insn.
29319 (omp_simt_enter): New expansion.
29320 (omp_simt_exit): New insn.
29321 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29323 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29324 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29325 (expand_GOMP_SIMT_EXIT): New.
29326 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29327 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29328 (GOMP_SIMT_EXIT): Ditto.
29329 * target-insns.def (omp_simt_enter): New insn.
29330 (omp_simt_exit): Ditto.
29331 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29333 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29334 (lower_rec_input_clauses): Likewise.
29335 (lower_lastprivate_clauses): Handle SIMT privatization.
29337 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29338 (ompdevlow_adjust_simt_enter): New.
29339 (find_simtpriv_var_op): New.
29340 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29341 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29343 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29344 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29345 (copy_decl_for_dup_finish): Ditto.
29347 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29349 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29352 * config/i386/i386.c (ix86_option_override_internal): Always
29353 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29355 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29357 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29359 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29361 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29362 mark new edge's irreducible flag accordign to it.
29363 (vect_do_peeling): Check loop preheader edge's irreducible flag
29364 and pass it to function slpeel_add_loop_guard.
29366 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29368 PR tree-optimization/80218
29369 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29370 Update block frequencies and counts.
29372 2017-03-28 Richard Biener <rguenther@suse.de>
29374 PR tree-optimization/78644
29375 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29376 of a simplification result we may not use it at all.
29378 2017-03-28 Richard Biener <rguenther@suse.de>
29381 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29382 without arguments, generate default definition of a SSA name.
29384 2017-03-28 Richard Biener <rguenther@suse.de>
29386 PR middle-end/80222
29387 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29388 TYPE_REF_CAN_ALIAS_ALL references.
29389 * fold-const.c (fold_indirect_ref_1): Likewise.
29391 2017-03-28 Martin Liska <mliska@suse.cz>
29394 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29395 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29397 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29398 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29400 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29401 (EXTRA_SPECS): Define.
29402 (SUBTARGET_EXTRA_SPECS): Likewise.
29403 (SUBTARGET_CPP_SPEC): Likewise.
29404 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29405 SUBTARGET_EXTRA_SPECS.
29406 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29408 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29410 * config/arc/simdext.md (vst64_insn): Update pattern.
29411 (vld32wh_insn): Likewise.
29412 (vld32wl_insn): Likewise.
29413 (vld64_insn): Likewise.
29414 (vld32_insn): Likewise.
29416 2017-03-28 Marek Polacek <polacek@redhat.com>
29419 * fold-const.c (fold_comparison): Use protected_set_expr_location
29420 instead of SET_EXPR_LOCATION.
29422 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29424 * tree.c (add_expr): Avoid name lookup warning.
29426 2017-03-27 Jeff Law <law@redhat.com>
29428 PR tree-optimization/80216
29429 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29430 function name. Limit recursion depth.
29431 (record_temporary_equivalences): Corresponding changes.
29433 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29435 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29438 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29441 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29443 * cfgcleanup.c (reg_note_cfa_p): New array.
29444 (insns_have_identical_cfa_notes): New function.
29445 (old_insns_match_p): Don't cross-jump in between /f
29446 and non-/f instructions. If both i1 and i2 are frame related,
29447 verify all CFA notes, their order and content.
29449 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29452 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29453 HImode and SImode with zero extend to DImode to one insn.
29454 (bswap<mode>2_extenddi): Likewise.
29455 (bswapsi2_extenddi): Likewise.
29456 (bswaphi2_extendsi): Likewise.
29457 (bswaphi2): Combine bswap HImode and SImode into one insn.
29458 Separate memory insns from swapping register.
29459 (bswapsi2): Likewise.
29460 (bswap<mode>2): Likewise.
29461 (bswaphi2_internal): Delete, no longer used.
29462 (bswapsi2_internal): Likewise.
29463 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
29464 store, and gpr<-gpr swap insns.
29465 (bswap<mode>2_store): Likewise.
29466 (bswaphi2_reg): Register only splitter, combine with the splitter.
29467 (bswaphi2 splitter): Likewise.
29468 (bswapsi2_reg): Likewise.
29469 (bswapsi2 splitter): Likewise.
29470 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
29471 the insns into load, store, and register/register insns.
29472 (bswapdi2_ldbrx): Likewise.
29473 (bswapdi2_load): Likewise.
29474 (bswapdi2_store): Likewise.
29475 (bswapdi2_reg): Likewise.
29477 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
29479 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
29480 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
29482 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
29485 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
29487 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29488 special handling for target option conflicts between dform
29489 options (-mpower9-dform, -mpower9-dform-vector,
29490 -mpower9-dform-scalar) and -mno-direct-move.
29492 2017-03-27 Richard Biener <rguenther@suse.de>
29494 PR tree-optimization/80181
29495 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29497 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29499 * config/arc/predicates.md (move_double_src_operand): Replace the
29500 call to move_double_src_operand with a call to address_operand.
29502 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29504 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29505 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29506 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29508 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29510 * config/arc/predicates.md (long_immediate_loadstore_operand):
29511 Consider scaled addresses cases.
29513 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29515 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29516 restored when in interrupt.
29517 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29518 doesn't have delay slot.
29520 2017-03-27 Richard Biener <rguenther@suse.de>
29523 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29524 inlined thunk clones.
29526 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29529 * asan.c (instrument_derefs): Copy over last operand from
29530 original COMPONENT_REF to the new COMPONENT_REF with
29531 DECL_BIT_FIELD_REPRESENTATIVE.
29532 * ubsan.c (instrument_object_size): Likewise.
29534 2017-03-27 Richard Biener <rguenther@suse.de>
29536 PR tree-optimization/80170
29537 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29538 sure DR/SCEV didnt fold in constants we do not see when looking
29539 at the reference base alignment.
29541 2017-03-27 Richard Biener <rguenther@suse.de>
29543 PR middle-end/80171
29544 * gimple-fold.c (fold_ctor_reference): Properly guard against
29545 NULL return value from canonicalize_constructor_val.
29547 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29550 * config/i386/i386.c (ix86_expand_builtin)
29551 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29552 flags reg setting and flags reg using instructions.
29553 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29554 clobbering instructions to zero extend op2.
29556 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29558 * doc/install.texi (Configuration) <--with-aix-soname>:
29559 Update link to AIX ld.
29561 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29563 PR rtl-optimization/80160
29564 PR rtl-optimization/80159
29565 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29566 reg_alternate_class into account.
29568 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29571 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29572 to consider in curr_insn_transform.
29574 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29576 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29577 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29578 and emit_mode_inner.
29580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29582 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29583 argument to the overloaded builtin variants. Use the new flag to
29584 deprecate certain builtin variants.
29585 * config/s390/s390-builtin-types.def: Add new builtin types.
29586 * config/s390/s390-builtins.h: Support new flags field for
29587 overloaded builtins.
29588 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29589 (s390_macro_to_expand): Enable vector float data type.
29590 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29591 builtins by incrementing the __VEC__ version number.
29592 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29594 (s390_resolve_overloaded_builtin): Emit error messages depending
29595 on the builtin flags.
29596 * config/s390/s390.c (s390_expand_builtin): Support additional
29597 flags argument. Change error message to match the messages
29598 emitted in s390-c.c.
29599 * config/s390/s390.md: New UNSPEC_* constants.
29600 (op_type): Add new instruction types.
29601 * config/s390/vecintrin.h: Add new builtins and test data class
29603 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29604 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29605 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29606 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29607 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29608 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29610 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29611 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29612 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29613 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29615 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29616 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29617 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29618 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29620 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29621 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29622 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29623 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29624 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29625 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29626 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29628 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29629 ("vec_scatter_element<V_HW_4:mode>_DI")
29630 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29631 ("vec_fpint<mode>", "vflls")
29632 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29633 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29634 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29635 ("*vec_cmphe<mode>_cc"): ... these.
29637 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29638 mode constant instead of magic value.
29640 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29642 * config/s390/s390.c (s390_expand_vec_compare): Support other
29643 vector floating point modes than just V2DF.
29644 (s390_expand_vcond): Likewise.
29645 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29646 (s390_cannot_change_mode_class): Prevent mode changes between TF
29647 and V1TF in vector registers.
29648 * config/s390/s390.md (DF, SF): New mode attributes.
29649 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29650 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29651 SFmode support for VRs.
29652 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29654 (VFT, VF_HW): New mode iterators.
29655 (vw, sdx): New mode attributes.
29656 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29657 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29658 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29659 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29660 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29661 also the new vector floating point modes. Renaming to ...
29663 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29664 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29665 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29666 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29667 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29668 ("vec_unordered<mode>"): ... these.
29670 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29671 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29672 ("*vec_extendv2df"): New insn definitions.
29674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29676 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29677 ("mulditi3_2", "*muldi3_sign"): New patterns.
29678 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29679 rename the pattern definition.
29681 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29683 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29685 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29687 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29689 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29690 instruction if possible.
29691 * config/s390/vector.md (vec_halfnumelts): New mode
29693 ("*vec_vllezlf<mode>"): New pattern.
29695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29697 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29698 ("popcountv4si2", "popcountv2di2"): Rename to ...
29699 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29700 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29702 ("popcount<mode>2_vxe"): New pattern.
29704 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29706 * common/config/s390/s390-common.c (processor_flags_table): Add
29708 * config.gcc: Add arch12.
29709 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29710 Default to arch12 for unknown CPU model numbers.
29711 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29712 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29713 PROCESSOR_max sanity check.
29714 * config/s390/s390-opts.h (enum processor_type): Add
29716 * config/s390/s390.c (processor_table): Add arch12.
29717 (s390_expand_builtin): Add check for B_VXE flag.
29718 (s390_issue_rate): Add PROCESSOR_ARCH12.
29719 (s390_get_sched_attrmask): Likewise.
29720 (s390_get_unit_mask): Likewise.
29721 (s390_sched_score): Enable z13 scheduling for arch12.
29722 (s390_sched_reorder): Likewise.
29723 (s390_sched_variable_issue): Likewise.
29724 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29726 (s390_tune_attr): Use z13 scheduling also for arch12.
29727 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29728 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29729 (TARGET_VXE_P): New macros.
29730 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29731 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29732 * config/s390/s390.opt: Add arch12 as processor_type.
29734 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29736 * config/s390/s390.md
29737 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29738 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29739 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29741 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29742 Rename expanders to ...
29744 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29745 ("fixuns_truncdddi2_emu"): ... these.
29747 ("fixuns_trunc<mode>si2_emu"): New expander.
29749 ("*fixuns_truncdfdi2_z13"): Rename to ...
29750 ("*fixuns_truncdfdi2_vx"): ... this.
29752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29754 * config/s390/2964.md: Remove the single element vector compare
29755 instructions which are no longer used.
29756 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29758 (s390_canonicalize_comparison): Remove handling of DFmode
29760 (s390_expand_vec_compare_scalar): Remove function.
29761 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29762 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29764 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29766 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29768 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29770 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29771 will anyway by matched by mov<mode>_64dfp.
29773 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29775 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29776 vlef/vstef. Add missing operand to vleif.
29778 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29780 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29781 pair for all vector types with 64 bit elements.
29782 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29783 * config/s390/vector.md (V_HW_64): ... here.
29784 (V_128_NOSINGLE): New mode iterator.
29785 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29786 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29787 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29788 ("*vec_load_pairv2di"): Change to ...
29789 ("*vec_load_pair<mode>"): ... this one.
29791 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29793 * config/s390/constraints.md: Add comments.
29794 (jKK): Reject element sizes > 8 bytes.
29795 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29797 * config/s390/s390.md: Add the s_operand checks formerly in
29798 s390_split_ok_p to various splitters where they are still
29800 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29801 for 128 bit vectors. Plus two splitters.
29803 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29805 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29808 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29811 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29812 error if the boundary argument is not constant.
29814 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29816 PR rtl-optimization/80112
29817 * loop-doloop.c (doloop_condition_get): Don't check condition
29818 if cmp isn't SET with IF_THEN_ELSE src.
29820 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29822 PR tree-optimization/80158
29823 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29824 replacing a candidate statement, also replace it for the
29825 candidate's alternate interpretation.
29826 (replace_rhs_if_not_dup): Likewise.
29827 (replace_one_candidate): Likewise.
29829 2017-03-24 Richard Biener <rguenther@suse.de>
29831 PR tree-optimization/80167
29832 * graphite-isl-ast-to-gimple.c
29833 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29835 (translate_isl_ast_to_gimple::get_rename): Likewise.
29837 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29839 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29840 handling of certain combinations of target options, including the
29841 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29842 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29844 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29847 * config/arm/arm.md (*load_multiple): Add reload_completed to
29848 matching condition.
29850 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29851 Richard Biener <rguenth@suse.de>
29853 PR tree-optimization/79908
29854 PR tree-optimization/80136
29855 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29856 been cast away, gimplify_and_add suffices.
29858 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29860 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29862 2017-03-23 Richard Biener <rguenther@suse.de>
29864 PR tree-optimization/80032
29865 * gimplify.c (gimple_push_cleanup): Forced unconditional
29866 cleanups still have to go to the conditional_cleanups
29869 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29871 PR tree-optimization/80072
29872 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29874 (next_operand_entry_id): Change type to unsigned int.
29875 (sort_by_operand_rank): Make sure to return the right return value
29876 even if unsigned fields are bigger than INT_MAX.
29877 (struct oecount): Change cnt and id type to unsigned int.
29878 (oecount_hasher::equal): Formatting fix.
29879 (oecount_cmp): Make sure to return the right return value
29880 even if unsigned fields are bigger than INT_MAX.
29881 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29884 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29885 TREE_READONLY on result if writing it more than once.
29888 * doc/invoke.texi (-fsanitize=thread): Document that with
29889 -fnon-call-exceptions atomics are not able to throw
29893 * tsan.c: Include tree-eh.h.
29894 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29895 maybe_clean_or_replace_eh_stmt where needed.
29896 (instrument_memory_accesses): Add cfg_changed argument.
29897 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29898 if it returned true.
29899 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29901 PR rtl-optimization/63191
29902 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29903 wrapper function, moved the whole old content into ...
29904 (ix86_delegitimize_address_1): ... this. New inline function.
29905 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29906 true as last argument instead of ix86_delegitimize_address.
29908 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29910 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29911 cortexa57_branch_cost.
29913 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29915 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29917 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29920 * doc/md.texi (Constraints): Document wA constraint.
29921 * config/rs6000/constraints.md (wA): New.
29922 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29923 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29924 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29925 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29927 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29930 * gimplify.c (is_oacc_declared): New function.
29931 (oacc_default_clause): Use it to set default flags for acc declared
29932 variables inside parallel regions.
29933 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29934 declared variables.
29935 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29936 declare attribute to any decl as necessary.
29938 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29941 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29942 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29943 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29944 (arm_arch_lpae): This.
29945 * config/arm/arm.c (arm_arch7ve): Rename into ...
29946 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29947 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29950 2017-03-22 Martin Liska <mliska@suse.cz>
29953 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29954 error message instead of an ICE.
29956 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29958 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29960 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29962 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29964 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29967 2017-03-21 Martin Sebor <msebor@redhat.com>
29969 * doc/extend.texi: Use "cannot" instead of "can't."
29970 * doc/hostconfig.texi: Same.
29971 * doc/install.texi: Same.
29972 * doc/invoke.texi: Same.
29973 * doc/loop.texi: Same.
29974 * doc/md.texi: Same.
29975 * doc/objc.texi: Same.
29976 * doc/rtl.texi: Same.
29977 * doc/tm.texi: Same.
29978 * doc/tm.texi.in: Same.
29979 * doc/trouble.texi: Same.
29981 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29984 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29985 (collect_checksum_attributes): Set it.
29986 (die_checksum_ordered): Use it.
29988 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29990 PR tree-optimization/79908
29991 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29992 change: For a VA_ARG whose LHS has been cast away, use
29993 force_gimple_operand to construct the side effects.
29995 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29997 PR translation/80001
29998 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29999 more amenable to translation.
30000 (oacc_loop_auto_partitions): Likewise.
30002 2017-03-21 Marek Polacek <polacek@redhat.com>
30003 Martin Sebor <msebor@redhat.com>
30005 PR tree-optimization/80109
30006 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30007 on INTEGRAL_TYPE_P.
30009 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30010 Segher Boessenkool <segher@kernel.crashing.org>
30013 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30014 check reg_used_between_p between insn and one of succ or succ2
30015 depending on if succ is artificial insn not inserted into insn
30018 2017-03-21 Martin Liska <mliska@suse.cz>
30020 PR gcov-profile/80081
30021 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30022 * doc/gcc.texi: Include gcov-dump stuff.
30023 * doc/gcov-dump.texi: New file.
30025 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30027 PR rtl-optimization/79150
30028 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30029 conditional jump, if the jump is the last insn of the loop.
30031 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30032 Richard Biener <rguenth@suse.de>
30034 PR tree-optimization/79908
30035 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30036 been cast away, use force_gimple_operand to construct the side
30039 2017-03-21 Martin Liska <mliska@suse.cz>
30041 PR libfortran/79956
30042 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30045 2017-03-21 Brad Spengler <spender@grsecurity.net>
30048 * plugin.c (htab_hash_plugin): New function.
30049 (add_new_plugin): Use it and adjust.
30050 (parse_plugin_arg_opt): Adjust.
30051 (init_one_plugin): Likewise.
30053 2017-03-21 Richard Biener <rguenther@suse.de>
30055 PR tree-optimization/80032
30056 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30057 if set force the cleanup to happen unconditionally.
30058 (gimplify_target_expr): Push inserted clobbers with force_uncond
30059 to avoid them being removed by control-dependent DCE.
30061 2017-03-21 Richard Biener <rguenther@suse.de>
30063 PR tree-optimization/80122
30064 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30065 va_arg_pack_len when the inlined call stmt requires pack
30067 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30069 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30072 * tsan.c (instrument_builtin_call): If the memory model argument
30073 is not a constant, assume it is valid.
30076 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30079 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30081 PR rtl-optimization/79910
30082 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30083 if its dest is used by an insn before I2 (other than the combined
30084 insns themselves, which are properly handled already).
30086 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30089 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30091 * combine.c (record_used_regs): New static function.
30092 (try_combine): Handle situations where there is an additional
30093 instruction between I2 and I3 which needs to have a LOG_LINK
30097 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30099 * combine.c (try_combine): Delete redundant i1 test. Call
30100 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30102 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30105 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30106 alternatives 13/14.
30108 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30110 PR tree-optimization/80054
30111 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30112 the optimization if a PHI or any of its arguments is not dominated
30113 by the candidate's basis. Use gphi* rather than gimple* as
30115 (replace_profitable_candidates): Clean up a gimple* variable that
30116 should be a gphi* variable.
30118 2017-03-20 Martin Sebor <msebor@redhat.com>
30121 * doc/extend.texi (attribute constructor): Document present limitation.
30123 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30126 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30127 __POWER9_VECTOR__ #ifdef control, change template definition to
30128 use Power9-specific built-in function.
30129 (vec_any_eq): Likewise.
30130 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30131 to control outcomes from this test.
30132 (vector_ae_<mode>p): For VEC_F modes, likewise.
30134 2017-03-20 Ian Lance Taylor <iant@google.com>
30136 * config/i386/i386.c (ix86_function_regparm): Save an extra
30137 register for -fsplit-stack with DECL_STATIC_CHAIN.
30139 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30142 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30143 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30145 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30147 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30149 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30152 2017-03-20 Marek Polacek <polacek@redhat.com>
30155 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30157 2017-03-20 Richard Biener <rguenther@suse.de>
30159 PR tree-optimization/80113
30160 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30161 allocate extra SSA name for PHI def.
30162 (add_close_phis_to_outer_loops): Likewise.
30163 (add_close_phis_to_merge_points): Likewise.
30164 (copy_loop_close_phi_args): Likewise.
30165 (copy_cond_phi_nodes): Likewise.
30167 2017-03-20 Martin Liska <mliska@suse.cz>
30169 PR middle-end/79753
30170 * tree-chkp.c (chkp_build_returned_bound): Do not build
30171 returned bounds for a LHS that's not a BOUNDED_P type.
30173 2017-03-20 Martin Liska <mliska@suse.cz>
30177 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30178 COMPLEX_CST and VECTOR_CST.
30180 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30183 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30184 target operand. A new splitter adds the clobber statement in case
30185 the target operand is dead anyway.
30187 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30189 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30190 to age-old versions of binutils and glibc.
30192 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30194 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30196 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30198 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30200 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30202 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30203 requirement for binutils 2.13.
30205 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30207 * combine.c (try_combine): Delete redundant i1 test. Call
30208 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30210 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30212 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30213 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30215 <riscv64-*-elf>: Re-arrange section
30216 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30217 <riscv32-*-linux>: Likewise.
30218 <riscv64-*-elf>: Likewise
30219 <riscv64-*-linux>: Likewise.
30221 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30224 * aarch64.opt(verbose-cost-dump): Fix typo.
30226 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30229 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30230 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30232 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30234 * reload.c (find_reloads): When reloading a nonoffsettable address,
30235 use RELOAD_OTHER for it and its address reloads.
30237 PR rtl-optimization/79910
30238 * combine.c (record_used_regs): New static function.
30239 (try_combine): Handle situations where there is an additional
30240 instruction between I2 and I3 which needs to have a LOG_LINK
30243 2017-03-17 Jeff Law <law@redhat.com>
30245 PR tree-optimization/71437
30246 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30247 conditional in the hash table first.
30248 (vrp_dom_walker::before_dom_children): Extract condition from
30249 ASSERT_EXPR. Record condition, its inverion and any implied
30250 conditions as well.
30252 2017-03-17 Marek Polacek <polacek@redhat.com>
30253 Markus Trippelsdorf <markus@trippelsdorf.de>
30255 PR tree-optimization/80079
30256 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30259 2017-03-17 Richard Biener <rguenther@suse.de>
30261 PR middle-end/80075
30262 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30263 Properly verify the LHS before the RHS possibly claims to be
30265 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30268 2017-03-17 Martin Jambor <mjambor@suse.cz>
30270 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30271 (List of -O2 options): Likewise.
30272 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30275 2017-03-17 Tom de Vries <tom@codesourcery.com>
30277 * gcov-dump.c (print_usage): Print bug_report_url.
30279 2017-03-17 Richard Biener <rguenther@suse.de>
30281 PR middle-end/80050
30282 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30283 (parser::peek): Likewise.
30285 2017-03-17 Richard Biener <rguenther@suse.de>
30287 PR tree-optimization/80048
30288 * sese.c (free_sese_info): Properly release rename_map and
30289 copied_bb_map elements.
30291 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30293 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30294 Add linked-list forward and backlinks. Insert on
30295 construction, remove on destruction.
30296 (class pass_store_merging): Add m_stores_head field.
30297 (pass_store_merging::terminate_and_process_all_chains):
30298 Iterate over m_stores_head list.
30299 (pass_store_merging::terminate_all_aliasing_chains):
30301 (pass_store_merging::execute): Check for debug stmts first.
30302 Push new chains onto the m_stores_head stack.
30304 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30307 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30308 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30309 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30311 2017-03-16 Jeff Law <law@redhat.com>
30313 PR tree-optimization/71437
30314 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30315 member function. Implementation moved into after_dom_children
30316 member function and into the threader's thread_outgoing_edges
30318 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30319 some code into new thread_outgoing_edges.
30320 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30321 definition. Simplify marker handling (do it here). Assume we always
30322 have the available expression and the const/copies tables.
30323 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30325 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30326 * tree-vrp.c (equiv_stack): No longer file scoped.
30327 (vrp_dom_walker): New class.
30328 (vrp_dom_walker::before_dom_children): New member function.
30329 (vrp_dom_walker::after_dom_children): Likewise.
30330 (identify_jump_threads): Setup domwalker. Use it rather than
30331 walking edges in a random order by hand. Simplify setup/finalization.
30332 (finalize_jump_threads): Remove.
30333 (vrp_finalize): Do not call identify_jump_threads here.
30334 (execute_vrp): Do it here instead and call thread_through_all_blocks
30337 PR tree-optimization/71437
30338 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30340 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30342 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30343 (dom_opt_dom_walker::thread_across_edge): Remove
30344 handle_dominating_asserts argument. All callers changed.
30345 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30346 changes. Remove calls to lhs_of_dominating_assert. Other
30347 uses of handle_dominating_asserts turn into unconditional code
30348 (simplify_control_stmt_condition_1): Likewise.
30349 (simplify_control_stmt_condition): Likewise.
30350 (thread_through_normal_block, thread_across_edge): Likewise.
30351 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30352 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30353 object if it is not an SSA_NAME.
30354 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30355 before calling into the VRP specific simplifiers.
30356 (identify_jump_threads): Remove handle_dominating_asserts
30359 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30362 * tree-diagnostic.c (default_tree_printer): No longer static.
30363 * tree-diagnostic.h (default_tree_printer): New prototype.
30365 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30367 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30368 Change ins into fmov.
30370 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30372 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30373 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30374 Use h_con constraint for operand 1.
30375 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30376 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30378 2017-03-15 Jeff Law <law@redhat.com>
30380 PR tree-optimization/71437
30381 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30382 (record_temporary_equivalences): Use it.
30384 PR tree-optimization/71437
30385 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30386 tree-ssa-scopedtables.
30387 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30388 (record_conditions, record_cond, vuse_eq): Likewise.
30389 (record_edge_info): Adjust to API tweak of record_conditions.
30390 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30391 (record_temporary_equivalences, optimize_stmt): Likewise.
30392 (eliminate_redundant_computations): Likewise.
30393 (record_equivalences_from_stmt): Likewise.
30394 * tree-ssa-scopedtables.c: Include options.h and params.h.
30395 (vuse_eq): New function, moved from tree-ssa-dom.c
30396 (build_and_record_new_cond): Likewise.
30397 (record_conditions): Likewise. Accept vector of conditions rather
30398 than edge_equivalence structure for first argument.
30399 for the first argument.
30400 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30401 from tree-ssa-dom.c.
30402 (avail_exprs_stack::record_cond): Likewise.
30403 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30404 from tree-ssa-dom.c.
30405 (avail_exprs_stack): Add new member functions lookup_avail_expr
30407 (record_conditions): Declare.
30409 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30412 * lra-constraints.c (process_alt_operands): Increase reject for
30413 reloading an input/output operand.
30415 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30418 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30419 insns to convert from signed/unsigned char/short to IEEE 128-bit
30421 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30423 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30426 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30427 subreg of inner mode for values already in registers.
30429 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30431 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30432 iteration reg is used after the loop.
30434 2017-03-14 Martin Sebor <msebor@redhat.com>
30436 PR tree-optimization/79800
30437 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30438 precision in negative-positive range.
30439 (format_floating): Call non-const overload with adjusted precision.
30441 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30444 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30447 2017-03-14 Martin Sebor <msebor@redhat.com>
30449 PR middle-end/80020
30450 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30451 * builtins.def (aligned_alloc): Use it.
30454 * Makefile.in (GTFILES): Add calls.c.
30455 * calls.c: Include "gt-calls.h".
30457 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
30459 PR rtl-optimization/79728
30460 * regs.h (struct target_regs): New field
30461 x_contains_allocatable_regs_of_mode.
30462 (contains_allocatable_regs_of_mode): New macro.
30463 * reginfo.c (init_reg_sets_1): Initialize it, and change
30464 contains_reg_of_mode so it includes global regs as well.
30465 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
30466 rather than contains_regs_of_mode.
30468 2017-03-14 Martin Liska <mliska@suse.cz>
30470 * doc/invoke.texi: Document options that can't be combined with
30471 -fcheck-pointer-bounds.
30473 2017-03-14 Martin Liska <mliska@suse.cz>
30475 PR middle-end/79831
30476 * doc/invoke.texi (-Wchkp): Document the option.
30478 2017-03-14 Martin Liska <mliska@suse.cz>
30480 * Makefile.in: Install gcov-dump.
30482 2017-03-14 Martin Liska <mliska@suse.cz>
30484 * multiple_target.c (expand_target_clones): Bail out for
30485 an invalid attribute.
30487 2017-03-14 Richard Biener <rguenther@suse.de>
30489 * alias.c (struct alias_set_entry): Pack properly.
30490 * cfgloop.h (struct loop): Likewise.
30491 * cse.c (struct set): Likewise.
30492 * ipa-utils.c (struct searchc_env): Likewise.
30493 * loop-invariant.c (struct invariant): Likewise.
30494 * lra-remat.c (struct cand): Likewise.
30495 * recog.c (struct change_t): Likewise.
30496 * rtl.h (struct address_info): Likewise.
30497 * symbol-summary.h (function_summary): Likewise.
30498 * tree-loop-distribution.c (struct partition): Likewise.
30499 * tree-object-size.c (struct object_size_info): Likewise.
30500 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30501 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30502 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30503 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30504 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30505 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30506 (struct _stmt_vec_info): Likewise.
30508 2017-03-14 Martin Liska <mliska@suse.cz>
30511 * multiple_target.c (create_dispatcher_calls): Check that
30512 a target can create a function dispatcher.
30514 2017-03-14 Martin Liska <mliska@suse.cz>
30517 * multiple_target.c (expand_target_clones): Drop local.local
30518 flag for default implementation.
30520 2017-03-14 Richard Biener <rguenther@suse.de>
30522 PR tree-optimization/80030
30523 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30525 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30527 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30528 gcc_fallthrough() instead of __attribute__((fallthrough));
30530 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30532 * doc/gcc.texi: Remove "up" link to (DIR).
30533 * doc/gccint.texi: Ditto.
30535 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30537 * doc/install.texi (Specific) <avr>: Remove reference to
30540 2017-03-13 Jeff Law <law@redhat.com>
30542 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30543 attribute rather than comments.
30545 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30546 match_scratch operand is highest.
30548 2017-03-13 Martin Liska <mliska@suse.cz>
30550 PR middle-end/78339
30551 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30552 is a CHKP clone, use original declaration.
30554 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30556 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30557 (arc_conditional_register_usage): Use a different allocation order
30558 when optimizing for size.
30559 * common/config/arc/arc-common.c (arc_option_optimization_table):
30560 Section anchors default on when optimizing for size.
30562 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30564 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30566 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30568 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30569 * config/arc/arc.md (cpu_facility): Add cd variant.
30570 (*movqi_insn): Add code density variant.
30571 (*movhi_insn): Likewise.
30572 (*movqi_insn): Likewise.
30573 (*addsi3_mixed): Likewise.
30574 (subsi3_insn): Likewise.
30576 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30578 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30580 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30582 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30583 expressions with MINUS and UNARY ops.
30585 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30588 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30590 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30591 between vec_select and vector argument.
30592 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30593 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30594 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30595 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30596 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30597 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30599 2017-03-13 Richard Biener <rguenther@suse.de>
30602 * params.def (vect-max-peeling-for-alignment): Fix typo.
30604 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30606 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30607 issue that only occurred with binutils below 2.18.
30609 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30611 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30612 refer to binutils 2.11/2.12 minimum.
30614 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30616 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30617 ftp.kernel.org and simplify binutils requirement.
30619 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30621 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30623 (Optimize Options): Ditto. Also remove redundancy.
30625 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30627 PR translation/79848
30628 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30630 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30631 to G_ to avoid double translation.
30633 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30635 PR translation/79923
30636 * auto-profile.c (get_combined_location): Convert leading
30637 character of diagnostics to lower case and remove trailing period.
30638 (read_profile): Likewise for various diagnostics.
30639 * config/arm/arm.c (arm_option_override): Remove trailing period
30640 from various diagnostics.
30641 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30642 (msp430_expand_delay_cycles): Likewise.
30644 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30647 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30648 full command-line argument, rather than just "str".
30649 (aarch64_validate_march): Likewise.
30650 (aarch64_validate_mtune): Likewise.
30652 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30654 PR rtl-optimization/78911
30655 * lra-assigns.c (must_not_spill_p): New function.
30656 (spill_for): Use it.
30658 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30660 PR tree-optimization/79981
30661 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30662 ATOMIC_COMPARE_EXCHANGE ifn result.
30663 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30664 IFN_ATOMIC_COMPARE_EXCHANGE.
30666 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30669 * opts.c (parse_sanitizer_options): Add missing question mark to
30670 "did you mean" message.
30672 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30674 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30676 (VMULEUH_UNS): Likewise.
30677 (VMULOUB_UNS): Likewise.
30678 (VMULOUH_UNS): Likewise.
30679 * config/rs6000/rs6000.c (builtin_function_type): Remove
30680 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30682 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30685 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30686 x with result of extra_parsing_for_operand_code_0.
30687 (function_reader::extra_parsing_for_operand_code_0): Convert
30688 return type from void to rtx, returning x. When reading
30689 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30690 larger size containing struct block_symbol.
30692 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30694 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30695 -mfloat128-hardware without -m64.
30697 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30700 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30701 entries to the case statement that marks unsigned arguments to
30702 overloaded functions.
30704 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30706 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30707 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30709 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30712 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30713 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30715 2017-03-10 Martin Liska <mliska@suse.cz>
30719 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30720 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30723 2017-03-10 Olivier Hainque <hainque@adacore.com>
30725 * tree-switch-conversion (array_value_type): Start by resetting
30726 candidate type to it's main variant.
30728 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30730 PR rtl-optimization/79909
30731 * combine.c (try_combine): Use simplify_replace_rtx on individual
30732 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30733 of the whole CALL_INSN_FUNCTION_USAGE.
30735 PR tree-optimization/79972
30736 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30737 get_range_info on SSA_NAMEs. Formatting fixes.
30739 2017-03-10 Richard Biener <rguenther@suse.de>
30740 Jakub Jelinek <jakub@redhat.com>
30742 PR tree-optimization/77975
30743 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30744 edge to be constant.
30745 (get_val_for): For constant x return it. Formatting fix.
30746 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30747 would use the same bases as the current one.
30749 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30751 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30752 instead of vec_select for V1TImode.
30753 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30755 (VSX_LE_128): Add V1TI to this mode iterator.
30756 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30757 (*vsx_le_perm_store_<mode>): Likewise.
30758 (pre-reload splitter for VSX stores): Likewise.
30759 (post-reload splitter for VSX stores): Likewise.
30760 (*vsx_xxpermdi2_le_<mode>): Likewise.
30761 (*vsx_lxvd2x2_le_<mode>): Likewise.
30762 (*vsx_stxvd2x2_le_<mode>): Likewise.
30764 2017-03-09 Michael Eager <eager@eagercon.com>
30766 Correct failures with --enable-checking=yes,rtl.
30768 * config/microblaze/microblaze.c (microblaze_expand_shift):
30769 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30770 test for const0_rtx.
30771 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30772 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30774 2017-03-09 Richard Biener <rguenther@suse.de>
30776 PR tree-optimization/79977
30777 * graphite-scop-detection.c (scop_detection::merge_sese):
30778 Handle the case of extra exits to blocks dominating the entry.
30780 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30782 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30785 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30787 PR rtl-optimization/79949
30788 * lra-constraints.c (process_alt_operands): Check memory when
30789 trying to predict a cycle. Print about the overall increase.
30791 2017-03-09 Richard Biener <rguenther@suse.de>
30793 PR middle-end/79971
30794 * gimple-expr.c (useless_type_conversion_p): Preserve
30795 TYPE_SATURATING for fixed-point types.
30797 2017-03-09 Richard Biener <rguenther@suse.de>
30800 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30801 alignment of BLKmode params.
30803 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30806 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30807 (VALL_NO_V2Q): Likewise.
30809 * config/aarch64/aarch64-simd.md
30810 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30812 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30813 VALL_NO_V2Q mode iterator.
30814 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30816 2017-03-09 Martin Liska <mliska@suse.cz>
30818 PR tree-optimization/79631
30819 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30820 tree_int_cst_sign_bit just for INTEGER constants.
30822 2017-03-09 Martin Liska <mliska@suse.cz>
30826 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30829 2017-03-09 Marek Polacek <polacek@redhat.com>
30832 * tree.c (inchash::add_expr): Handle TREE_VEC.
30834 2017-03-09 Martin Liska <mliska@suse.cz>
30837 (chkp_narrow_size_and_offset): New function.
30838 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30839 (void chkp_parse_bit_field_ref): New function.
30840 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30841 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30843 2017-03-09 Martin Liska <mliska@suse.cz>
30846 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30847 (chkp_find_bounds_1): Remove gcc_unreachable.
30849 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30852 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30853 BUILT_IN_SYNC*, determine the access type from the size suffix and
30854 always build a MEM_REF with that type. Handle forgotten
30855 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30858 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30859 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30860 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30861 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30862 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30863 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30864 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30865 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30866 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30867 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30868 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30869 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30870 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30871 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30872 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30873 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30874 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30875 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30876 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30877 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30878 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30879 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30880 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30881 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30882 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30883 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30884 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30885 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30886 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30887 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30888 definitions outside of __OPTIMIZE__ guarded section.
30891 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30892 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30893 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30894 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30897 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30899 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30900 ("vfenez<mode>"): Add missing constraints.
30902 2017-03-08 Martin Sebor <msebor@redhat.com>
30905 * config/nds32/nds32.c (nds32_option_override):
30906 Fix misspelled diagnostic.
30908 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30911 * gimplify.c (gimplify_omp_for): Replace index var in outer
30912 taskloop statement with an artificial variable and add
30913 OMP_CLAUSE_PRIVATE clause for it.
30915 2017-03-08 Richard Biener <rguenther@suse.de>
30917 PR tree-optimization/79955
30918 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30919 for accesses that are completely outside of the variable.
30921 2017-03-08 Andrew Haley <aph@redhat.com>
30923 PR tree-optimization/79943
30924 * tree-ssa-loop-split.c (compute_new_first_bound): When
30925 calculating the new upper bound, (END-BEG) should be added, not
30928 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30930 * config/avr/avr.md (setmemhi): Make sure match_dup
30931 operand number comes before match_scratch.
30933 2017-03-08 Richard Biener <rguenther@suse.de>
30935 PR tree-optimization/79920
30936 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30937 with ncopies == 1 to ...
30938 (vect_transform_slp_perm_load): ... here. Properly compute
30939 all element loads by iterating VF times over the group. Do
30940 not handle ncopies (computed in a broken way) in
30941 vect_create_mask_and_perm.
30943 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30946 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30947 is a uniform vector, use uniform_vector_p return value instead of
30948 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30950 2017-03-07 Marek Polacek <polacek@redhat.com>
30952 PR middle-end/79809
30953 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30954 (alloca_call_type): Likewise.
30956 2017-03-07 Martin Liska <mliska@suse.cz>
30958 * gcov.c (process_args): Put comment to correct location.
30960 2017-03-07 Martin Liska <mliska@suse.cz>
30962 PR middle-end/68270
30963 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30964 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30965 (chkp_narrow_bounds_for_field): Likewise.
30966 (chkp_parse_array_and_component_ref): Pass one more argument to
30969 2017-03-07 Richard Biener <rguenther@suse.de>
30971 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30974 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30976 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30977 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30979 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30982 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30983 to end of description.
30984 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30986 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30988 PR rtl-optimization/79901
30989 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30991 (*avx512f_<code><mode>3<mask_name>): ... this.
30992 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30993 iterator instead of VI8_AVX2_AVX512BW.
30995 PR rtl-optimization/79901
30996 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30997 min/max expander, expand it using expand_vec_cond_expr.
31000 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31003 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31006 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31007 to void * for PCH reasons.
31008 * dwarf2out.c (output_loc_operands, output_die): Cast
31009 v.val_vec.array to unsigned char *.
31011 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31014 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31017 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31019 PR rtl-optimization/79571
31020 * lra-constraints.c (process_alt_operands): Calculate static
31021 reject and subtract it from overall when only addresses will be
31024 2017-03-06 Julia Koval <julia.koval@intel.com>
31027 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31028 incoming stack boundary to 128 for 64-bit targets.
31030 2017-03-06 Richard Biener <rguenther@suse.de>
31032 PR tree-optimization/79894
31033 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31034 to NULL after folding it.
31036 2017-03-06 Richard Biener <rguenther@suse.de>
31038 PR tree-optimization/79824
31039 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31040 check disabling peeling for gaps.
31042 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31044 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31045 attributes): Document gettimeofday.
31047 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31049 * config/s390/s390.c (s390_option_override_internal): Set
31050 PARAM_MIN_VECT_LOOP_BOUND
31052 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31054 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31055 * config/s390/s390.md: Likewise.
31057 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31060 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31061 (<avx2_avx512>_perm<mode>): Rename to ...
31062 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31064 (<avx512>_perm<mode>_mask): Rename to ...
31065 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31067 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31068 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31069 instead of VI8F_256_512.
31070 (avx512f_perm<mode>): New define_expand.
31071 (avx512f_perm<mode>_mask): Likewise.
31072 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31073 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31075 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31077 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31078 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31080 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31082 2017-03-06 Martin Liska <mliska@suse.cz>
31085 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31086 when having a SSA NAME w/o VAR_DECL assigned to it.
31088 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31090 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31091 msa_dpsub_<su>_d): Fix MODE for vec_select.
31093 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31095 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31097 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31099 2017-03-06 Richard Biener <rguenther@suse.de>
31101 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31102 * plugin.c (register_plugin_info): Likewise.
31103 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31105 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31107 * config/i386/sse.md (sse_storehps, sse_storelps,
31108 avx_<castmode><avxsizesuffix>_<castmode>,
31109 avx512f_<castmode><avxsizesuffix>_<castmode>,
31110 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31111 in condition that at least one operand is not a MEM.
31113 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31115 PR middle-end/79805
31116 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31117 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31119 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31120 gimple_call_nothrow_p flag based on whether original builtin can throw.
31121 If it can, emit following stmts on the fallthrough edge.
31122 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31123 don't create new bb if inserting just debug stmts on the edge, try to
31124 insert them on the fallthru bb or just reset debug stmts.
31126 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31129 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31130 restore recog_data (including the operand rtxes inside it) around
31131 the call to get_insn_template.
31133 2017-03-03 Martin Sebor <msebor@redhat.com>
31135 PR tree-optimization/79699
31136 * context.c (context::~context): Free MPFR caches to avoid
31137 a memory leak on program exit.
31139 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31141 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31142 Use wide_int::ulow () instead of .elt (0).
31144 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31146 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31147 (*pushxf): Limit oF constraint to 32bit targets and add oC
31148 constraint for 64bit targets.
31149 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31150 (*pushdf): Change rmF constraint to rmC.
31152 2017-03-03 Martin Liska <mliska@suse.cz>
31154 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31155 Remove unused variable.
31157 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31160 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31161 is a memory operand, increase num_memory.
31162 (ix86_expand_args_builtin): Likewise.
31164 2017-03-03 Jan Hubicka <jh@suse.cz>
31167 * ipa-devirt.c (maybe_record_node): Properly handle
31168 __cxa_pure_virtual visibility.
31170 2017-03-03 Martin Liska <mliska@suse.cz>
31172 PR tree-optimization/79803
31173 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31175 (pass_loop_prefetch::execute): Disabled optimization if an
31176 assumption about L1 cache size is not met.
31178 2017-03-03 Martin Liska <mliska@suse.cz>
31180 PR rtl-optimization/79574
31181 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31182 (hash_scan_set): Likewise.
31183 (dump_hash_table): Likewise.
31184 (hoist_code): Likewise.
31186 2017-03-03 Richard Biener <rguenther@suse.de>
31188 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31189 in place of uhigh ().
31190 (fixed_convert_from_real): Likewise.
31192 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31195 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31197 2017-03-03 Richard Biener <rguenther@suse.de>
31199 PR middle-end/79818
31200 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31201 TYPE_OVERFLOW_UNDEFINED check.
31203 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31205 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31207 (vector_ae_<mode>_p): Likewise.
31208 (vector_nez_<mode>_p): Likewise.
31209 (vector_ne_v2di_p): Likewise.
31210 (vector_ae_v2di_p): Likewise.
31211 (vector_ne_<mode>_p): Likewise.
31212 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31214 (vsx_tsqrt<mode>2_fe): Likewise.
31216 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31219 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31221 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31223 PR rtl-optimization/79780
31224 * cprop.c (one_cprop_pass): When second and further conditional trap
31225 in a single basic block is turned into an unconditional trap, turn it
31226 into a deleted note to avoid RTL verification failures.
31228 2017-03-02 Richard Biener <rguenther@suse.de>
31230 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31232 2017-03-02 Richard Biener <rguenther@suse.de>
31234 PR tree-optimization/79345
31236 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31237 param and abort the walk, returning -1 if it is hit.
31238 (walk_aliased_vdefs): Take a limit param and pass it on.
31239 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31240 defaulting to 0 and return a signed int.
31241 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31242 (check_defs): New helper.
31243 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31244 about uninitialized memory.
31245 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31246 bogus uninitialized warning.
31247 (fixed_convert_from_real): Likewise.
31249 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31251 PR tree-optimization/66768
31252 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31253 iv_use if base object can't be determined.
31255 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31257 PR tree-optimization/79345
31258 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31259 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31260 (get_pattern_stats): Initialize it.
31261 * genemit.c (gen_expand): Verify match_scratch numbers come after
31262 match_operand/match_dup numbers.
31263 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31264 match_scratch numbers.
31265 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31267 * config/s390/s390.md (trunctdsd2): Likewise.
31269 2017-03-02 Richard Biener <rguenther@suse.de>
31271 * wide-int.h (wide_int_storage::operator=): Implement in terms
31274 2017-03-02 Richard Biener <rguenther@suse.de>
31276 PR tree-optimization/79777
31277 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31278 the to insert expression to sth existing.
31280 2017-03-01 Martin Sebor <msebor@redhat.com>
31282 PR middle-end/79692
31283 * gimple-ssa-sprintf.c
31284 (directive::known_width_and_precision): New function.
31285 (format_integer): Use it.
31286 (get_mpfr_format_length): Consider the full range of precision
31287 when computing %g output with the # flag. Set the likely byte
31288 count to 3 rather than 1 when precision is indeterminate.
31289 (format_floating): Correct the lower bound of precision.
31291 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31293 * doc/invoke.texi: Document default code model for 64-bit Linux.
31295 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31298 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31299 udiv rather than div since input pattern is unsigned.
31301 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31303 * config/i386/i386.c (print_reg): Warn for values of
31304 unsupported size in integer register.
31306 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31309 * config/rs6000/predicates.md (current_file_function_operand): Do
31310 not allow self calls to be local if the function is replaceable.
31312 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31315 * config/rs6000/altivec.h (vec_ctz and others): Change the
31316 preprocessor macro that controls conditional compilation from
31317 _ARCH_PWR9 to __POWER9_VECTOR__.
31318 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31319 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31320 control (instead of _ARCH_PWR9 control) so that template
31321 definition uses power9-specific function.
31322 (vec_any_eq): Likewise.
31323 (vec_all_ne): Change macro definition to use a power9-specific
31324 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31325 _ARCH_PWR9 control).
31326 (vec_any_eq) Likewise.
31327 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31328 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31329 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31330 support for xvcmpnedp instruction.
31331 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31332 macro expansion so that Power9 implementation of vec_all_ne does
31333 not use the AltiVec predicate framework.
31334 (VCMPNEH_P): Likewise.
31335 (VCMPNEW_P): Likewise.
31336 (VCMPNED_P): Likewise.
31337 (VCMPNEFP_P): Likewise.
31338 (VCMPNEDP_P): Likewise.
31339 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31340 implementation of vec_any_eq to not use AltiVec predicate
31342 (VCMPAEH_P): Likewise.
31343 (VCMPAEW_P): Likewise.
31344 (VCMPAED_P): Likewise.
31345 (VCMPAEFP_P): Likewise.
31346 (VCMPAEDP_P): Likewise.
31347 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31348 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31349 not use the AltiVec predicate framework.
31350 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31351 of vec_any_eq to not use AltiVec predicate framework.
31352 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31353 support for predefined __POWER9_VECTOR__ macro to indicate that
31354 Power9 instruction selection is enabled.
31355 (altivec_overloaded_builtins): Remove extraneous
31356 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31357 function argument types RS6000_BTI_bool_V16QI and
31358 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31359 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31360 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31361 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31362 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31363 Power9 for implementations of vec_cmpne. Change the signature for
31364 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31365 (representing vec_all_ne) to remove the previously described first
31366 argument of type RS6000_BTI_INTSI, as this was an artifact of
31367 reliance on the AltiVec predicate framework, which is no longer
31368 used in the implementation of these functions. Add
31369 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31370 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31371 since, unlike the AltiVec predicate framework implementation, we
31372 do not share function descriptors between vec_alle and vec_anyeq.
31373 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31374 set of modes that receive special treatment even when
31375 TARGET_P9_VECTOR is true. The special treatment emits code that
31376 does not depend on Power9 instructions.
31377 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31378 define_expand to not rely on AltiVec predicate framework.
31379 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31381 (vector_ne_v2di_p): Change this define_expand to not rely on
31382 AltiVec predicate framework.
31383 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31385 (vector_ne_<mode>_p): Change this define_expand to not rely on
31386 AltiVec predicate framework.
31387 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31389 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31390 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31391 define_insn pattern.
31392 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31393 define_insn pattern because the xvcmpne<VSs>. instruction is not
31395 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31396 instruction is not supported.
31398 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31400 * config/nvptx/nvptx.c: Include intl.h.
31402 2017-03-01 Martin Jambor <mjambor@suse.cz>
31405 * ipa-prop.h (ipa_bits): Removed field known.
31406 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31407 to pointers. Adjusted their comments to warn about their sharing.
31408 (ipcp_transformation_summary): Change bits to a vector of pointers.
31409 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31410 (ipa_get_ipa_bits_for_value): Declare.
31411 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31412 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31413 (ipa_bits_hash_table): Likewise.
31414 (ipa_vr_ggc_hash_traits): Likewise.
31415 (ipa_vr_hash_table): Likewise.
31416 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31417 being pointers and vr_known being removed.
31418 (ipa_set_jf_unknown): Likewise.
31419 (ipa_get_ipa_bits_for_value): New function.
31420 (ipa_set_jfunc_bits): Likewise.
31421 (ipa_get_value_range): New overloaded functions.
31422 (ipa_set_jfunc_vr): Likewise.
31423 (ipa_compute_jump_functions_for_edge): Use the above functions to
31424 construct bits and vr parts of jump functions.
31425 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31426 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31428 (ipcp_grow_transformations_if_necessary): Also allocate
31429 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31431 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31432 them. Fix too long lines.
31433 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31434 vr_known being removed.
31435 (ipa_read_jump_function): Use new setter functions to construct bits
31436 and vr parts of jump functions or set them to NULL.
31437 (write_ipcp_transformation_info): Adjust for bits being pointers.
31438 (read_ipcp_transformation_info): Likewise.
31439 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31441 Include gt-ipa-prop.h.
31442 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31444 (ipcp_store_bits_results): Likewise.
31445 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31446 Do not write to existing jump functions but use a temporary instead.
31448 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31451 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31452 attempt to use its first operand as BIT_FIELD_REF base.
31454 2017-03-01 Richard Biener <rguenther@suse.de>
31456 PR middle-end/79721
31457 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
31458 interpolating formula in wrapping arithmetic.
31459 (chrec_apply): Convert chrec_evaluate return value to wanted type.
31461 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31463 PR tree-optimization/79734
31464 * tree-vect-generic.c (expand_vector_condition): Optimize
31465 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
31466 Handle VEC_COND_EXPR where comparison has different inner width from
31467 type's inner width.
31469 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
31471 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
31472 markup, and similar issues. Remove @opindex entries for things
31473 that aren't options. Add missing -mmpy-option entries.
31475 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31477 PR tree-optimization/79737
31478 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
31479 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
31480 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
31481 instead of byte_size. Formatting fix.
31482 (shift_bytes_in_array_right): Formatting fix.
31484 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
31487 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
31488 condition on optimize for the leaf function test.
31490 2017-02-28 Martin Liska <mliska@suse.cz>
31493 * read-rtl-function.c (function_reader::handle_unknown_directive):
31494 Bail out when one uses -flto.
31496 2017-02-28 Martin Liska <mliska@suse.cz>
31498 * common.opt: Replace space with tabular for options of <number>
31500 * config/i386/i386.opt: Show <number> value for
31501 -mlarge-data-threshold.
31502 * opts.c (print_filtered_help): Do not display number in hexadecimal
31505 2017-02-28 Martin Liska <mliska@suse.cz>
31507 * common.opt: Fix --help=option -Q for options which are of
31510 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31512 * config/i386/i386.c (print_reg): Error out for values
31513 of 8-bit size in invalid integer register.
31515 2017-02-28 Martin Sebor <msebor@redhat.com>
31517 PR tree-optimization/79691
31518 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31520 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31523 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31524 gcc_unreachable with output_operand_lossage.
31526 2017-02-28 Richard Biener <rguenther@suse.de>
31528 PR tree-optimization/79740
31529 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31531 (visit_nary_op): Insert the nary into the hashtable if we
31532 pattern-matched sth.
31533 * tree-ssa-pre.c (eliminate_insert): Robustify.
31535 2017-02-28 Richard Biener <rguenther@suse.de>
31537 PR middle-end/79731
31538 * fold-const.c (decode_field_reference): Reject out-of-bound
31541 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31543 * config/i386/i386.c: Include intl.h.
31544 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31545 instead of just cond ? "..." : "...".
31546 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31547 * coverage.c (read_counts_file): Likewise.
31548 * omp-offload.c: Include intl.h.
31549 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31550 of just cond ? "..." : "...".
31551 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31552 of just cond ? "..." : "...".
31554 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31557 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31559 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31560 'tune for' CPU name.
31561 * config/arm/arm-cpu-data.h: Regenerated.
31563 2017-02-28 Richard Biener <rguenther@suse.de>
31565 PR tree-optimization/79732
31566 * tree-inline.c (expand_call_inline): Do not shadow var.
31568 2017-02-28 Richard Biener <rguenther@suse.de>
31570 PR tree-optimization/79723
31571 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31572 address-space properly.
31574 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31576 * doc/optinfo.texi (Optimization groups): Fix option used for
31578 * doc/invoke.texi (-fopt-info): Document "omp".
31579 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31580 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31581 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31582 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31583 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31585 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31587 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31588 OPTGROUP_OMP with "omp".
31590 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31593 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31594 for arithmetic shift of unsigned V2DI.
31596 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31598 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31599 arc/linux.h headers.
31600 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31601 (LINK_SPEC): Likewise.
31602 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31603 (EXTRA_SPECS): Likewise.
31604 (STARTFILE_SPEC): Likewise.
31605 (ENDFILE_SPEC): Likewise.
31606 (LIB_SPEC): Likewise.
31607 (TARGET_SDATA_DEFAULT): Likewise.
31608 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31609 (MULTILIB_DEFAULTS): Likewise.
31610 (DWARF2_UNWIND_INFO): Likewise.
31611 * config/arc/big.h: New file.
31612 * config/arc/elf.h: Likewise.
31613 * config/arc/linux.h: Likewise.
31614 * config/arc/t-uClibc: Remove.
31616 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31618 PR tree-optimization/77536
31619 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31620 (tree_transform_and_unroll_loop): Use above function to compute the
31621 estimated niter of unrolled loop and use it when scaling profile.
31622 Also use count info rather than frequency if it's non-zero.
31623 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31624 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31625 (vect_transform_loop): Call above function.
31627 2017-02-27 Richard Biener <rguenther@suse.de>
31629 PR tree-optimization/45397
31630 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31631 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31632 (visit_nary_op): Add pattern matching for CSEing sign-changed
31633 or truncated operations with wider ones.
31635 2017-02-27 Richard Biener <rguenther@suse.de>
31637 PR tree-optimization/79690
31638 * tree-vect-stmts.c (vectorizable_store): Use vector type
31639 built from the DR with address-space.
31641 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31643 * doc/invoke.texi (Optimize Options): Refine the description
31644 of asan-use-after-return.
31646 2017-02-25 Alan Modra <amodra@gmail.com>
31648 PR rtl-optimization/79584
31649 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31650 base, not ad->base_term, the reg within base. Remove assertion
31651 that ad->base == ad->base_term. Replace gen_int_mode using
31652 bogus mode with const0_rtx.
31654 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31656 PR middle-end/79396
31657 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31658 FMA_EXPR like tcc_binary or tcc_unary.
31660 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31663 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31665 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31666 (output_loc_operands): Handle DW_OP_call_ref and
31667 DW_OP_GNU_variable_value.
31668 (struct variable_value_struct): New type.
31669 (struct variable_value_hasher): Likewise.
31670 (variable_value_hash): New variable.
31671 (string_types): Remove.
31672 (copy_loc_descr): New function.
31673 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31674 (prepend_loc_descr_to_each): New function.
31675 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31676 instead of add_loc_descr_to_each if the first argument is single
31677 location list and the second has multiple.
31678 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31679 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31680 when looking for variable value which doesn't have other location info.
31681 (loc_list_from_tree): Formatting fix.
31682 (gen_array_type_die): Simplify DW_AT_string_length handling.
31683 (adjust_string_types): Remove.
31684 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31685 string_types. Call resolve_variable_values.
31686 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31687 (resolve_addr_in_expr): Likewise. Add A argument.
31688 (copy_deref_exprloc): Remove deref argument. Adjust for the
31689 original expression being DW_OP_GNU_variable_value with optionally
31690 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31691 optionally after it.
31692 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31693 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31694 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31695 (variable_value_hasher::hash, variable_value_hasher::equal): New
31697 (resolve_variable_value_in_expr, resolve_variable_value,
31698 resolve_variable_values, note_variable_value_in_expr,
31699 note_variable_value): New functions.
31700 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31703 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31706 * opts.h (handle_generated_option): Add GENERATED_P argument.
31707 * opts-common.c (handle_option): Adjust function comment.
31708 (handle_generated_option): Add GENERATED_P argument, pass it to
31710 (control_warning_option): Pass false to handle_generated_option
31712 * opts.c (maybe_default_option): Pass true to handle_generated_option
31714 * optc-gen.awk: Likewise.
31716 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31718 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31719 a REG, look at the REG it is a SUBREG of.
31720 (splitter for cmpeqsi_t): Ditto.
31722 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31724 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31725 the special USEs with the pattern of the insn, not the insn itself.
31727 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31730 * doc/invoke.texi: Document -mload-store-pairs.
31732 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31733 Sandra Loosemore <sandra@codesourcery.com>
31735 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31736 argument isn't a CONST_INT.
31737 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31738 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31739 (nios2_validate_compare): Bypass alternate compare logic if *op2
31740 is not a CONST_INT.
31741 (ldstwm_operation_p): Return false if first_base is not a REG or
31742 if first_offset is not a CONST_INT.
31744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31746 * config/cris/cris.md: Use correct operand in a define_peephole2.
31748 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31750 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31752 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31754 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31755 this_insn if it is an INSN or JUMP_INSN.
31756 (force_offsettable): Look at base, not at addr.
31757 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31758 on things that aren't necessarily CONST_INTs.
31760 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31762 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31763 -mfpmath=sse is the default also for x86-32 targets with SSE2
31764 instruction set when @option{-ffast-math} is enabled
31766 2017-02-24 Jeff Law <law@redhat.com>
31768 PR rtl-optimizatoin/79286
31769 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31772 2017-02-24 Richard Biener <rguenther@suse.de>
31774 PR tree-optimization/79389
31775 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31778 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31780 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31781 function comment to reflect reality.
31782 (loop_exits_before_overflow): Fix typo in function description.
31784 2017-02-24 Richard Biener <rguenther@suse.de>
31786 PR tree-optimization/79389
31787 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31788 properly that a threading opportunity exists. Detect conditional
31789 copy/constant propagation opportunities.
31791 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31793 * config/visium/visium.md (type): Add trap.
31794 (b): New mode attribute.
31795 (*btst): Rename into...
31796 (*btst<mode>): ...this and adjust.
31797 (*cbranchsi4_btst_insn): Rename into...
31798 (*cbranch<mode>4_btst_insn): ...this and adjust.
31799 (trap): New define_insn.
31801 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31803 PR tree-optimization/79389
31804 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31805 (noce_reversed_cond_code): New function.
31806 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31807 reversed_comparison_code. Formatting fix.
31808 (noce_try_store_flag): Test rev_cond != NULL in addition to
31809 reversed_comparison_code.
31810 (noce_try_store_flag_constants): Likewise.
31811 (noce_try_store_flag_mask): Likewise.
31812 (noce_try_addcc): Use rev_cond if non-NULL instead of
31813 reversed_comparison_code.
31814 (noce_try_cmove_arith): Likewise. Formatting fixes.
31815 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31816 (noce_find_if_block): Initialize rev_cond.
31817 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31818 instead of false as last argument never attempt to reverse it
31821 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31823 PR tree-optimization/79663
31824 * tree-predcom.c (combine_chains): Process refs in reverse order
31825 only for ZERO length chains, and add explaining comment.
31827 2017-02-23 Jeff Law <law@redhat.com>
31829 PR tree-optimization/79578
31830 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31831 in call to operand_equal_p.
31833 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31836 * config/i386/cpuid.h: Fix another undefined behavior.
31838 2017-02-23 Richard Biener <rguenther@suse.de>
31840 PR tree-optimization/79683
31841 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31842 vector types for data-refs.
31844 2017-02-23 Martin Liska <mliska@suse.cz>
31846 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31848 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31850 PR middle-end/79665
31851 * internal-fn.c (get_range_pos_neg): Moved to ...
31852 * tree.c (get_range_pos_neg): ... here. No longer static.
31853 * tree.h (get_range_pos_neg): New prototype.
31854 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31855 are known to be in between 0 and signed maximum inclusive, try to
31856 expand both unsigned and signed divmod and use the cheaper one from
31859 2017-02-22 Jeff Law <law@redhat.com>
31861 PR tree-optimization/79578
31862 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31863 to compare base operands.
31865 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31868 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31869 gpc_reg_operand instead of fpr_reg_operand.
31871 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31873 * config/mips/mips.c (mips_return_in_memory): Force FP
31874 vector types to be returned in memory for o32 ABI.
31876 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31878 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31879 instead of DW_TAG_member for static data member declarations and don't
31880 set no_linkage_name for static inline data members.
31881 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31884 2017-02-22 Martin Liska <mliska@suse.cz>
31886 * doc/invoke.texi: Replace inequality signs with square brackets
31889 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31892 * lra-constraints.c (simplify_operand_subreg): Handle
31893 WORD_REGISTER_OPERATIONS targets.
31895 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31898 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31899 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31900 elimination by swapping fld*.
31902 2017-02-22 Richard Biener <rguenther@suse.de>
31904 PR tree-optimization/79673
31905 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31906 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31907 irrelevant address-space qualifiers and avoiding a
31908 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31910 2017-02-22 Richard Biener <rguenther@suse.de>
31912 PR tree-optimization/79666
31913 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31914 to not symbolically negate if that may introduce undefined
31917 2017-02-22 Martin Liska <mliska@suse.cz>
31920 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31921 * data-streamer-out.c (streamer_write_gcov_count_stream):
31923 * value-prof.c (stream_out_histogram_value): Make assert more
31924 precise based on type of counter.
31926 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31929 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31930 Use nonimmediate_operand instead of memory_operand for operand 1.
31931 (float-extend standard_x87sse_constant_load splitter): Ditto.
31933 2017-02-21 Jeff Law <law@redhat.com>
31935 PR tree-optimization/79621
31936 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31937 blocks with edges to themselves.
31939 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31942 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31943 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31944 Use gimple_call_builtin_p.
31947 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31948 on temporarily removed DEBUG_INSNs.
31950 PR tree-optimization/79649
31951 * tree-loop-distribution.c (classify_partition): Give up on
31952 non-generic address space loads/stores.
31954 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31956 * doc/loop.texi (Loop manipulation): Remove nonexistent
31957 tree_ssa_loop_version from the documentation.
31958 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31960 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31963 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31964 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31965 * config/rs6000/rs6000.c: Include except.h.
31966 (rs6000_expand_split_stack_prologue): Call
31967 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31969 2017-02-21 Martin Jambor <mjambor@suse.cz>
31972 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31973 have been analyzed.
31975 2017-02-21 Martin Jambor <mjambor@suse.cz>
31977 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31978 for backward compatibility only.
31979 * doc/invoke.texi (Option Summary): Remove all references to
31980 -fipa-cp-alignment.
31982 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31986 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31988 * lra-constraints.c (curr_insn_transform): Handle
31989 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31991 2017-02-21 Martin Liska <mliska@suse.cz>
31993 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31995 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31997 PR translation/79638
31998 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32000 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32003 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32004 (arm_function_ok_for_sibcall): Return false for an indirect call by
32005 descriptor if all the argument registers are used.
32006 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32007 alignment of the function.
32009 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32011 PR tree-optimization/61441
32012 * simplify-rtx.c (simplify_const_unary_operation): For
32013 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32014 the sNaN unmodified.
32016 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32018 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32019 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32020 instead of SYSTEM_HEADER_DIR.
32022 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32023 Martin Liška <mliska@suse.cz>
32025 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32026 Fix typos and grammar, use active voice, and clarify.
32028 2017-02-20 Marek Polacek <polacek@redhat.com>
32030 PR middle-end/79537
32031 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32034 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32036 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32039 * config/i386/i386.c (ix86_expand_builtin): Handle
32040 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32041 ix86_builtins_isa[fcode].isa as a requirement of those
32042 flags and any other flag in the bitmask.
32043 (ix86_init_mmx_sse_builtins): Use 0 instead of
32044 ~OPTION_MASK_ISA_64BIT as mask.
32045 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32046 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32047 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32048 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32050 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32053 * lra-constraints.c (split_reg): Check requested split mode
32054 is supported by the register.
32056 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32058 * lra-constraints.c (simplify_operand_subreg): Remove early
32061 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32064 * lra-constraints.c (curr_insn_transform): Tighten condition
32065 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32067 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32070 * lra-constraints.c (curr_insn_transform): Handle
32071 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32073 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32076 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32078 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32080 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32083 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32086 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32088 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32091 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32093 * final.c (last_columnnum, override_columnnum): New variables.
32094 (final_start_function): Set last_columnnum, pass it to begin_prologue
32095 hook and pass 0 to dwarf2out_begin_prologue.
32096 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32097 to source_line debug hook.
32098 (notice_source_line): Compute last_columnnum and for debug_column_info
32099 return true on column changes.
32100 * debug.h (struct gcc_debug_hooks): Add column argument to
32101 source_line and begin_prologue hooks.
32102 (debug_nothing_int_charstar_int_bool): Remove prototype.
32103 (debug_nothing_int_int_charstar,
32104 debug_nothing_int_int_charstar_int_bool): New prototypes.
32105 (dwarf2out_begin_prologue): Add column argument.
32106 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32107 begin_prologue hooks.
32108 (debug_nothing_int_charstar_int_bool): Remove.
32109 (debug_nothing_int_int_charstar,
32110 debug_nothing_int_int_charstar_int_bool): New functions.
32111 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32112 through to dwarf2out_source_line.
32113 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32114 (dwarf2out_source_line): Add column argument, emit it if requested.
32115 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32117 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32118 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32119 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32120 through to dwarf2out_begin_prologue.
32121 (vmsdbgout_source_line): Add column argument, pass it through to
32122 dwarf2out_source_line.
32123 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32124 dbxout_source_line caller.
32125 (dbxout_source_line): Add column argument.
32127 * common.opt (gno-column-info, gcolumn-info): New options.
32128 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32129 (check_die): Also test for multiple DW_AT_decl_column attributes.
32130 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32131 DW_AT_decl_column if requested.
32132 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32134 (gen_variable_die): Likewise.
32135 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32136 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32139 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32140 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32141 (ix86_handle_option): Handle OPT_m3dnowa.
32142 * doc/invoke.texi (-m3dnowa): Document.
32143 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32144 -m3dnowa instead of -m3dnow -march=athlon.
32147 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32148 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32150 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32153 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32154 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32155 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32156 generator for vsx_xxpermdi_<mode>_be.
32157 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32158 force big-endian semantics.
32159 (vsx_xxpermdi_<mode>_be): New define_expand with same
32160 implementation as previous version of vsx_xxpermdi_<mode>.
32162 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32164 PR tree-optimization/79327
32165 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32166 variable, its initialization and use.
32168 2017-02-17 Julia Koval <julia.koval@intel.com>
32170 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32171 (OPTION_MASK_ISA_PKU_UNSET): New.
32172 (ix86_handle_option): Handle -mrdpid.
32173 * config/i386/cpuid.h (bit_RDPID): New.
32174 * config/i386/driver-i386.c (host_detect_local_cpu):
32175 Detect RDPID feature.
32176 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32177 * config/i386/i386-c.c (ix86_target_macros_internal):
32179 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32180 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32181 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32182 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32183 * config/i386/i386.md (define_insn "rdpid"): New.
32184 * config/i386/i386.opt Add -mrdpid.
32185 * config/i386/immintrin.h (_rdpid_u32): New.
32187 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32189 PR rtl-optimization/79541
32190 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32191 instead of transforming it into USE.
32193 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32195 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32196 If HONOR_SNANS (SFmode) force the input to a register.
32197 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32198 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32199 an frsp or similar insn.
32201 2017-02-17 Martin Liska <mliska@suse.cz>
32203 PR rtl-optimization/79577
32204 * params.def (selsched-max-sched-times): Increase minimum to 1.
32206 2017-02-17 Martin Liska <mliska@suse.cz>
32208 PR rtl-optimization/79574
32209 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32211 2017-02-17 Martin Liska <mliska@suse.cz>
32213 PR tree-optimization/79529
32214 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32215 ssa_defined_default_def_p to handle cases which are implicitly
32217 * tree-ssa.c (ssa_defined_default_def_p): New function.
32218 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32219 which are implicitly defined.
32220 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32222 2017-02-17 Richard Biener <rguenther@suse.de>
32224 PR middle-end/79576
32225 * params.def (max-ssa-name-query-depth): Limit to 10.
32227 2017-02-17 Richard Biener <rguenther@suse.de>
32229 PR tree-optimization/79552
32230 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32233 2017-02-17 Richard Biener <rguenther@suse.de>
32236 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32238 2017-02-17 Marek Polacek <polacek@redhat.com>
32240 PR middle-end/79536
32241 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32242 (fold_negate_expr): New wrapper.
32244 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32246 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32247 Correct terminology and de-emphasize pre-standard behavior.
32249 2017-02-16 Alan Modra <amodra@gmail.com>
32251 PR rtl-optimization/79286
32252 * ira.c (def_dominates_uses): New function.
32253 (update_equiv_regs): Don't create an equivalence for insns that
32254 may trap where the register def does not dominate the use.
32256 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32258 PR rtl-optimization/78127
32259 * lra.c (lra): Call lra_eliminate before finish the loop after
32262 2017-02-16 Richard Biener <rguenther@suse.de>
32264 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32266 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32267 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32268 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32269 (isl_val_int_from_wi): New function.
32270 (extract_affine_gmp): Rename to ...
32271 (extract_affine_wi): ... this, take a widest_int.
32272 (extract_affine_int): Just wrap extract_affine_wi.
32273 (add_param_constraints): Use isl_val_int_from_wi.
32274 (add_loop_constraints): Likewise, and extract_affine_wi.
32276 2017-02-15 Jeff Law <law@redhat.com>
32278 PR middle-end/79521
32279 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32280 ira_init_register_move_cost_if_necessary.
32282 2017-02-15 Martin Sebor <msebor@redhat.com>
32284 PR middle-end/32003
32285 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32286 removed in a prior commit.
32288 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32290 PR tree-optimization/79347
32291 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32292 counters during peeling.
32294 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32296 * Makefile.in (site.exp): Remove "set ISLVER".
32298 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32301 * real.c (real_from_integer): Call real_convert even for decimal.
32303 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32306 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32308 2017-02-14 Andrew Pinski <apinski@cavium.com>
32310 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32311 cores and change the partno/implementer to be correct.
32312 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32313 the 'B" as the implementer.
32314 * config/aarch64/aarch64-tune.md: Regenerate.
32316 2017-02-14 Carl Love <cel@us.ibm.com>
32318 * config/rs6000/rs6000.c: Add case statement entry to make the
32319 xvcvuxdsp built-in argument unsigned.
32320 * config/rs6000/vsx.md: Fix the source and return operand types so they
32321 match the instruction definitions from the ISA document. Fix typo
32322 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32325 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32328 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32329 member early_clobber_alts.
32330 * lra-lives.c (reg_early_clobber_p): New.
32331 (process_bb_lives): Use it.
32332 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32333 (debug_operand_data): Initialize early_clobber_alts.
32334 (setup_operand_alternative): Set up early_clobber_alts.
32335 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32336 alternatives to new_insn_reg.
32337 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32339 (lra_update_insn_regno_info): Pass the new arg.
32341 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32343 PR middle-end/79505
32344 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32345 (new_oacc_loop_raw): Don't clear already cleared fields.
32348 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32349 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32350 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32352 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32355 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32357 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32360 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32361 the extra instruction to the right place to store 128-bit constant
32364 2017-02-14 Martin Sebor <msebor@redhat.com>
32366 PR middle-end/79448
32367 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32368 warning for strings of unknown length.
32370 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32372 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32374 2017-02-14 Jeff Law <law@redhat.com>
32377 * ira-costs.c (scan_one_insn): Initialize register move costs
32378 for pseudos seen in USE/CLOBBER insns.
32380 PR tree-optimization/79095
32381 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32382 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32383 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32384 if the operands are known to be not equal, then the resulting range
32386 (intersect_ranges): If the new range is ~[0,0] and the old range is
32387 wide, then prefer ~[0,0].
32388 * tree-vrp.c (overflow_comparison_p_1): New function.
32389 (overflow_comparison_p): New function.
32390 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32391 if NAME is used in an overflow test.
32392 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32393 overflow check that can be expressed as an equality test, then adjust
32394 ops to be that equality test.
32396 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32398 * config/s390/s390-builtin-types.def: Remove flags argument.
32399 * config/s390/s390.c (s390_init_builtins): Likewise.
32401 2017-02-14 Martin Liska <mliska@suse.cz>
32403 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32404 vector. Fix trailing white spaces.
32406 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32408 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32411 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32413 PR rtl-optimization/68664
32414 * config/arm/arm.c (arm_sched_can_speculate_insn):
32415 New function. Declare prototype.
32416 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32418 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32420 PR rtl-optimization/68664
32421 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32423 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32425 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32427 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32428 max skip bytes for function, loop and jump.
32430 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32432 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32433 ABS_EXPR for gimple dump.
32435 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32438 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32440 PR tree-optimization/79408
32441 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32442 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32443 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32444 also if rhs1 is INTEGER_CST.
32446 2017-02-14 Richard Biener <rguenther@suse.de>
32448 PR middle-end/79432
32449 * tree-into-ssa.c (insert_phi_nodes): When the function can
32450 have abnormal edges rewrite SSA names with broken use-def
32451 dominance out of SSA and register them for PHI insertion.
32453 2017-02-13 Martin Sebor <msebor@redhat.com>
32455 PR middle-end/79496
32456 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32457 clearing info.nowrite flag when snprintf size argument is a range.
32459 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32461 * cprop.c (cprop_jump): Add missing space in string literal.
32462 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
32463 (get_constraint_for_component_ref): Likewise.
32464 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
32465 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
32466 * lra-constraints.c (process_alt_operands): Likewise.
32467 * ipa-inline.c (inline_small_functions): Likewise.
32468 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
32469 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
32470 * trans-mem.c (diagnose_tm_1_op): Likewise.
32471 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
32472 (grid_parallel_clauses_gridifiable): Likewise.
32474 * config/nvptx/mkoffload.c (process): Add space in between
32477 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
32478 "MOD4_SSE_REGS" and "ALL_REGS".
32480 * spellcheck.c (test_data): Add , in between "foo" and "food".
32482 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32485 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
32486 boundary crossing check and subsequent code generation agree.
32488 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32490 * config/aarch64/aarch64.c (has_memory_op): Delete.
32491 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32494 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32496 PR rtl-optimization/79388
32497 PR rtl-optimization/79450
32498 * combine.c (distribute_notes): When removing TEM_INSN for which
32499 corresponding dest has last value recorded, invalidate that last
32502 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32504 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32505 of explicit '@'. Add missing assembly comment marker on branch costs
32508 2017-02-13 Nathan Sidwell <nathan@acm.org>
32510 * gengtype-lex.l (<in_struct>): Add '/'.
32512 2017-02-13 Martin Liska <mliska@suse.cz>
32515 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32517 2017-02-13 Richard Biener <rguenther@suse.de>
32519 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32521 * configure: Re-generate.
32522 * config.in: Likewise.
32523 * graphite-dependences.c: Simplify as if
32524 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32525 * graphite-isl-ast-to-gimple.c: Likewise.
32526 * graphite-optimize-isl.c: Likewise.
32527 * graphite-poly.c: Likewise.
32528 * graphite-sese-to-poly.c: Likewise.
32529 * graphite.h: Likewise.
32530 * toplev.c: Include isl/version.h and use isl_version () for
32531 printing the ISL version.
32532 * doc/install.texi: Update ISL requirement.
32534 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32536 * doc/standards.texi (Standards): Update reference to
32539 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32541 * doc/extend.texi (Named Address Spaces): sourceware.org now
32543 * doc/install.texi (Binaries): Ditto.
32546 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32548 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32549 standard terminology "stringize"/"stringizing" throughout.
32550 * doc/cppinternals.texi: Likewise.
32552 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32554 * doc/extend.texi: Fix some spelling mistakes and typos.
32555 * doc/invoke.texi: Likewise.
32557 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32560 * params.def (inline-min-speedup) Change from 10 to 8.
32562 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32564 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32567 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32570 * ipa-inline-analysis.c (get_minimal_bb): New function.
32571 (record_modified): Use it.
32572 (remap_edge_change_prob): Handle also ancestor functions.
32574 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32576 * doc/contrib.texi (Contributors): Remove broken link into
32577 the Mauve CVS repository.
32579 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32581 PR middle-end/79454
32582 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32583 result computation whenever lhs doesn't have vector mode, not
32584 just when it has BLKmode.
32586 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32588 * doc/makefile.texi (profiledbootstrap): Refer to the
32589 installation instructions only in textual form.
32591 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32594 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32596 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32598 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32599 (Specific): Update mingw-w64 reference.
32601 (Specific): Remove broken link to Renesas RX processor.
32603 2017-02-10 Richard Biener <rguenther@suse.de>
32605 * toplev.c (process_options): Do not mention obsolete graphite
32606 options when printing sorry message about missing graphite support.
32607 Mention -floop-nest-optimize.
32609 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32611 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32612 (vtst_p16): Likewise.
32613 (vtstq_p8): Likewise.
32614 (vtstq_p16): Likewise.
32616 (vtstq_p64): Likewise.
32617 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32618 (vset_lane_p64): New.
32619 (vsetq_lane_p64): New.
32621 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32623 PR tree-optimization/79411
32624 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32625 stmt operands are SSA_NAMEs used in abnormal phis.
32626 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32629 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32632 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32635 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32637 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32639 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32641 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32645 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32646 not arbitrary TREE_CONSTANT.
32649 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32650 "omp declare target link" attribute unless is_global_var.
32651 * omp-offload.c (find_link_var_op): Likewise.
32653 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32654 Chung-Lin Tang <cltang@codesourcery.com>
32656 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32658 (gimplify_adjust_omp_clauses): Don't delete TILE.
32659 (gimplify_omp_for): Deal with TILE.
32660 * internal-fn.c (expand_GOACC_TILE): New function.
32661 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32663 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32664 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32666 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32667 avoid DIV for outermost collapse var.
32668 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32669 Remove out of date comments, fix whitespace.
32670 * omp-general.c (omp_extract_for_data): Deal with tiling.
32671 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32672 adjust OLF_DIM_BASE value.
32673 (struct omp_for_data): Add tiling field.
32674 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32675 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32676 for auto loops. Remove default auto determining, moved to
32677 oacc_loop_fixed_partitions.
32678 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32679 stmts, add e_mask field.
32680 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32681 (oacc_thread_numbers): Use oacc_dim_call.
32682 (oacc_xform_tile): New.
32683 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32684 (finish_oacc_loop): Adjust for ifns vector.
32685 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32686 add case for GOACC_TILE fns.
32687 (oacc_loop_xform_loop): Delete.
32688 (oacc_loop_process): Iterate over call list directly, and add
32689 handling for GOACC_TILE fns.
32690 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32692 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32693 vector partitioning to outer loops. Assign 2 partitions to loops
32694 when available. Add TILE handling.
32695 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32696 (execite_oacc_device_lower): Process GOACC_TILE fns,
32697 ignore unknown specs.
32698 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32699 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32700 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32702 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32704 * configure.ac (ACX_BUGURL): Update.
32705 * configure: Regenerate.
32707 2017-02-09 Richard Biener <rguenther@suse.de>
32709 PR tree-optimization/69823
32710 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32711 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32713 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32715 * config/arc/arc-c.def: Add __NPS400__ definition.
32716 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32717 (TARGET_NPS400): Define.
32719 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32721 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32723 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32724 pointer, arch_info.
32725 (arc_cpu_types): Fill the arch_info field with a pointer into the
32726 arc_arch_types table.
32727 (arc_selected_cpu): Declare.
32728 * config/arc/arc.c (arc_selected_cpu): Make global.
32729 (arc_selected_arch): Delete.
32730 (arc_base_cpu): Delete.
32731 (arc_override_options): Remove references to deleted variables,
32732 update access to arch information.
32733 (ARC_OPT): Update access to arch information.
32734 (ARC_OPTX): Likewise.
32735 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32736 (TARGET_ARC600): Update access to arch information.
32737 (TARGET_ARC601): Likewise.
32738 (TARGET_ARC700): Likewise.
32739 (TARGET_EM): Likewise.
32740 (TARGET_HS): Likewise.
32741 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32744 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32747 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32748 condition/operands for integer GE/LE/GEU/LEU operations.
32750 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32752 PR translation/79397
32753 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32756 2017-02-08 Martin Jambor <mjambor@suse.cz>
32759 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32760 whether allocation happened.
32761 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32762 nothing was allocated.
32764 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32766 PR tree-optimization/79408
32767 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32768 constant, but SSA_NAME with a known integer range, use the minimum
32769 of that range instead of op1 to determine if modulo can be replaced
32770 with its first operand.
32772 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32774 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32776 2017-02-08 Richard Biener <rguenther@suse.de>
32778 PR tree-optimization/71824
32779 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32780 Check all loops contained in the merged region.
32782 2017-02-07 Andrew Pinski <apinski@cavium.com>
32784 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32786 2017-02-07 Andrew Pinski <apinski@cavium.com>
32788 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32789 (thunderxt88): Likewise.
32790 (thunderxt81): Disable LSE and change v8.1 to v8.
32791 (thunderxt83): Likewise.
32793 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32794 Richard Biener <rguenther@suse.de>
32796 PR middle-end/79399
32797 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32798 type from int to size_t.
32799 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32801 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32803 PR rtl-optimization/79386
32804 * cprop.c (bypass_conditional_jumps): Initialize
32805 bypass_last_basic_block already before splitting bbs after
32806 unconditional traps...
32807 (bypass_conditional_jumps): ... rather than here.
32810 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32811 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32812 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32813 fix -masm=intel patterns.
32815 2017-02-07 Richard Biener <rguenther@suse.de>
32817 PR tree-optimization/79256
32818 PR middle-end/79278
32819 * builtins.c (get_object_alignment_2): Use min_align_of_type
32820 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32821 and ADJUST_FIELD_ALIGN.
32823 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32825 * doc/tm.texi: Regenerate.
32826 * stor-layout.c (layout_decl): Adjust.
32827 (update_alignment_for_field): Likewise.
32828 (place_field): Likewise.
32829 (min_align_of_type): Likewise.
32830 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32831 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32832 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32833 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32834 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32835 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32836 * config/i386/i386.c (x86_field_alignment): Likewise.
32837 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32838 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32839 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32840 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32841 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32842 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32846 2017-01-30 Richard Biener <rguenther@suse.de>
32848 PR tree-optimization/79256
32849 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32850 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32853 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32855 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32856 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32857 builtins to SImode and emit a zero-extend, if necessary.
32859 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32861 * docs/invoke.texi (RISC-V Options): Alphabetize.
32863 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32865 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32868 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32870 * config/riscv/riscv.c: New file.
32871 * common/config/riscv/riscv-common.c: Likewise.
32872 * config.gcc: Likewise.
32873 * config/riscv/constraints.md: Likewise.
32874 * config/riscv/elf.h: Likewise.
32875 * config/riscv/generic.md: Likewise.
32876 * config/riscv/linux.h: Likewise.
32877 * config/riscv/multilib-generator: Likewise.
32878 * config/riscv/peephole.md: Likewise.
32879 * config/riscv/pic.md: Likewise.
32880 * config/riscv/predicates.md: Likewise.
32881 * config/riscv/riscv-builtins.c: Likewise.
32882 * config/riscv/riscv-c.c: Likewise.
32883 * config/riscv/riscv-ftypes.def: Likewise.
32884 * config/riscv/riscv-modes.def: Likewise.
32885 * config/riscv/riscv-opts.h: Likewise.
32886 * config/riscv/riscv-protos.h: Likewise.
32887 * config/riscv/riscv.h: Likewise.
32888 * config/riscv/riscv.md: Likewise.
32889 * config/riscv/riscv.opt: Likewise.
32890 * config/riscv/sync.md: Likewise.
32891 * config/riscv/t-elf-multilib: Likewise.
32892 * config/riscv/t-linux: Likewise.
32893 * config/riscv/t-linux-multilib: Likewise.
32894 * config/riscv/t-riscv: Likewise.
32895 * configure.ac: Likewise.
32896 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32897 Waterman as RISC-V maintainers.
32898 * doc/install.texi: Add RISC-V entries.
32899 * doc/invoke.texi: Add RISC-V options section.
32900 * doc/md.texi: Add RISC-V constraints section.
32901 * configure: Regenerated.
32903 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32906 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32907 false values to be constant vectors with all 0 or all 1 bits set.
32908 (vcondu<mode><mode>): Likewise.
32909 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32911 (fpmask_comparison_operator): Update comment.
32912 (vecint_comparison_operator): New predicate.
32913 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32914 vector conditionals when the true and false values are constant
32915 vectors with all 0 bits or all 1 bits set.
32917 2017-02-06 Martin Sebor <msebor@redhat.com>
32919 PR tree-optimization/79376
32920 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32922 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32924 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32925 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32926 to simplify split condition.
32928 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32930 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32931 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32934 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32936 PR rtl-optimization/68664
32937 * target.def (can_speculate_insn): New hook.
32938 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32939 * doc/tm.texi: Regenerate.
32940 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32941 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32942 (rs6000_sched_can_speculate_insn): New function.
32944 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32946 PR tree-optimization/79284
32947 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32948 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32949 vectorizable_mask_load_store, vectorizable_operation,
32950 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32951 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32952 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32953 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32954 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32955 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32956 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32957 is_gimple_assign (stmt). Replace another such test with
32958 is_gimple_assign (stmt).
32960 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32963 * config/avr/avr.c (rtl-iter.h): Include it.
32964 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32965 (avr_legitimate_combined_insn): ...and implementation.
32967 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32969 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32970 * config/s390/s390.c (s390_const_operand_ok)
32971 (s390_canonicalize_comparison, s390_extract_part)
32972 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32973 (s390_contiguous_bitmask_p, s390_rtx_costs)
32974 (legitimize_pic_address): Likewise.
32975 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32976 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32977 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32978 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32979 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32981 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32983 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32984 REGNO($0) == REGNO($1).
32986 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32988 * config/s390/linux.h(SIZE_TYPE): Add comment.
32990 2017-02-06 Julian Brown <julian@codesourcery.com>
32991 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32992 Virendra Pathak <virendra.pathak@broadcom.com>
32994 * config/aarch64/aarch64-cores.def: Change the scheduler
32996 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32997 * config/aarch64/thunderx2t99.md: New file.
32999 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33001 * doc/standards.texi (Go Language): Update link to language
33004 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33006 * tree-eh.c (lower_resx): Sanitize profile.
33007 (cleanup_empty_eh_move_lp): Likewise.
33009 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33012 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33014 * cfgloopmanip.h (loop_version): Update prototype.
33015 * modulo-sched.c (sms_schedule): Update call of loop_version.
33016 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33017 * tree-parloops.c (gen_parallel_loop): Likewise.
33018 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33019 * tree-ssa-loop-split.c (split_loop): Likewise.
33020 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33021 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33023 2017-02-05 Martin Liska <mliska@suse.cz>
33026 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33028 (print_operand_address): Initialize a struct to zero.
33030 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33032 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33033 garbage collector only in textual form.
33035 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33037 * doc/extend.texi (x86 specific memory model extensions for
33038 transactional memory): Simplify a phrase.
33040 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33043 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33044 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33045 (atomic_storedi_1): Likewise.
33047 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33049 PR tree-optimization/79338
33050 * tree-parloops.c (gather_scalar_reductions): Don't call
33051 vect_analyze_loop_form for loop->inner before destroying loop's
33054 2017-02-03 Martin Sebor <msebor@redhat.com>
33056 PR tree-optimization/79327
33057 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33058 when precision has resulted in leading zeros.
33059 (format_integer): Adjust the likely counter to assume an unknown
33060 argument that may be zero is non-zero.
33062 2017-02-03 Jason Merrill <jason@redhat.com>
33065 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33066 avoid copying non-taken branch.
33068 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33070 PR tree-optimization/79340
33071 * tree-vect-loop.c (vectorizable_reduction): Release
33072 vec_defs elements after safe_splicing them into other vectors.
33075 PR tree-optimization/79327
33076 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33077 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33079 (format_integer): Use wide_int_to_tree instead of build_int_cst
33080 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33081 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33082 of shortest and longest sequence.
33084 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33086 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33087 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33089 2017-02-03 Walter Lee <walt@tilera.com>
33092 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33093 after initial stackframe link reg save.
33094 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33096 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33099 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33100 wu for stxssp alternative.
33102 2017-02-03 Martin Sebor <msebor@redhat.com>
33104 PR tree-optimization/79352
33105 * gimple-fold.c (get_range_strlen): Add argument.
33106 (get_range_strlen): Change return type to bool.
33107 (get_maxval_strlen): Pass in a dummy argument.
33108 * gimple-fold.h (get_range_strlen): Change return type to bool.
33109 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33110 * tree.h (array_at_struct_end_p): Add argument.
33111 * tree.c (array_at_struct_end_p): Handle it.
33113 2017-02-03 Martin Liska <mliska@suse.cz>
33116 * multiple_target.c (create_dispatcher_calls): Redirect edge
33117 from a caller of a dispatcher.
33118 (expand_target_clones): Make the clones local.
33119 (ipa_target_clone): Do both target clones and resolvers.
33120 (ipa_dispatcher_calls): Remove the pass.
33121 (pass_dispatcher_calls::gate): Likewise.
33122 (make_pass_dispatcher_calls): Likewise.
33123 * passes.def (pass_target_clone): Put as very first IPA early
33126 2017-02-03 Martin Liska <mliska@suse.cz>
33128 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33129 in case of a function with ifunc attribute.
33131 2017-02-03 Martin Liska <mliska@suse.cz>
33133 * cgraph.c (cgraph_node::dump): Dump function version info.
33134 * symtab.c (symtab_node::dump_base): Add missing new line.
33136 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33138 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33139 (ifcombine_ifandif): Use it.
33141 2017-02-03 Martin Liska <mliska@suse.cz>
33143 * doc/invoke.texi: Document default value for
33144 use-after-scope-direct-emission-threshold.
33146 2017-02-03 Martin Liska <mliska@suse.cz>
33148 PR tree-optimization/79339
33149 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33150 (format_floating): Likewise.
33152 2017-02-03 Martin Liska <mliska@suse.cz>
33155 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33157 (ipa_node_params_t::remove): Likewise.
33158 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33159 initialization from removed ipa_node_params_t::insert.
33160 (ipa_node_params::~ipa_node_params): Move from removed
33161 ipa_node_params_t::release.
33162 * symbol-summary.h (symbol_summary::m_released): New member.
33163 Do not release a summary twice. Do not allow to call finalizer
33164 for types of a summary that live in GGC memory.
33166 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33168 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33171 2017-02-02 Martin Sebor <msebor@redhat.com>
33173 PR middle-end/79275
33174 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33175 (format_string): Tighten up the range of output for non-constant
33176 strings and correct the expected range for wide non-constant strings.
33178 2017-02-02 Martin Sebor <msebor@redhat.com>
33180 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33182 PR middle-end/32003
33183 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33185 (-fdump-tree-@var): Add to index and document how to come up
33186 with pass-specific option and dump file names.
33187 (-fdump-passes): Clarify where to look for output.
33189 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33191 PR middle-end/77445
33192 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33193 statistics of the analyzed path; allow threading for speed when
33194 any of BBs along the path are optimized for speed.
33196 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33198 PR middle-end/78468
33199 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33200 settings of the virtual registers.
33203 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33205 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33206 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33209 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33211 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33212 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33214 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33216 * config/s390/s390.md: Add missing comments with the expanded
33218 * config/s390/vector.md: Likewise.
33219 * config/s390/vx-builtins.md: Likewise.
33221 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33224 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33225 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33226 conditions on a single line.
33228 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33230 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33231 __S390_VX__ to __VX__.
33233 2017-02-01 Andrew Pinski <apinski@cavium.com>
33235 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33236 stmt_info to record_stmt_cost.
33237 (vect_get_known_peeling_cost): Pass stmt_info if known to
33239 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33240 cpu_vector_cost field into
33241 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33242 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33243 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33244 splitting of scalar_stmt_cost and vec_stmt_cost.
33245 (thunderx_vector_cost): Likewise.
33246 (cortexa57_vector_cost): LIkewise.
33247 (exynosm1_vector_cost): Likewise.
33248 (xgene1_vector_cost): Likewise.
33249 (thunderx2t99_vector_cost): Improve after the splitting of the two
33251 (aarch64_builtin_vectorization_cost): Update for the splitting of
33252 scalar_stmt_cost and vec_stmt_cost.
33254 2017-02-01 Torvald Riegel <triegel@redhat.com>
33255 Richard Henderson <rth@redhat.com>
33257 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33258 conditional on existance of a fast atomic load.
33259 * optabs-query.c (can_atomic_load_p): New function.
33260 * optabs-query.h (can_atomic_load_p): Declare it.
33261 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33262 no fast atomic load is available for the particular size of access.
33263 (expand_atomic_compare_and_swap): Likewise.
33264 (expand_atomic_load): Likewise.
33265 (expand_atomic_store): Likewise.
33266 (expand_atomic_fetch_op): Likewise.
33267 * testsuite/lib/target-supports.exp
33268 (check_effective_target_sync_int_128): Remove x86 because it provides
33269 no fast atomic load.
33270 (check_effective_target_sync_int_128_runtime): Likewise.
33272 2017-02-01 Richard Biener <rguenther@suse.de>
33274 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33275 (graphite_transform_loops): Provide opt-info for optimized nests.
33276 * tree-parloop.c (parallelize_loops): Provide opt-info for
33277 parallelized loops.
33279 2017-02-01 Richard Biener <rguenther@suse.de>
33281 PR middle-end/79315
33282 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33283 was not set before.
33285 2017-02-01 Richard Biener <rguenther@suse.de>
33287 PR tree-optimization/71824
33288 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33289 Verify the loops are valid in the merged SESE region.
33290 (scop_detection::can_represent_loop_1): Check analyzing the
33291 evolution of the number of iterations in the region succeeds.
33293 2017-01-31 Ian Lance Taylor <iant@golang.org>
33295 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33296 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33298 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33300 PR preprocessor/79210
33301 * input.c (get_substring_ranges_for_loc): Replace line_width
33302 assertion with error-handling.
33304 2017-01-31 Richard Biener <rguenther@suse.de>
33306 PR tree-optimization/77318
33307 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33308 (create_pw_aff_from_tree): Take loop parameter.
33309 (add_condition_to_pbb): Pass loop of the condition to
33310 create_pw_aff_from_tree.
33312 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33314 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33315 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33317 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33321 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33323 (convert_int_to_float128): Likewise.
33324 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33325 (convert_int_to_float128): Likewise.
33326 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33327 (UNSPEC_IEEE128_CONVERT): Likewise.
33328 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33329 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33330 Use local variables for IBM extended format.
33331 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33332 (fix_trunc<mode>si2_fprs): Likewise.
33333 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33334 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33335 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33336 to know that we can now have integers of all sizes in vector
33338 (fix<uns>_<mode>di2_hw): Likewise.
33339 (float<uns>_<mode>si2_hw): Likewise.
33340 (fix_<mode>si2_hw): Likewise.
33341 (fixuns_<mode>si2_hw): Likewise.
33342 (float<uns>_<mode>di2_hw): Likewise.
33343 (float_<mode>di2_hw): Likewise.
33344 (float_<mode>si2_hw): Likewise.
33345 (floatuns_<mode>di2_hw): Likewise.
33346 (floatuns_<mode>si2_hw): Likewise.
33347 (xscvqp<su>wz_<mode>): Delete, no longer used.
33348 (xscvqp<su>dz_<mode>): Likewise.
33349 (xscv<su>dqp_<mode>): Likewise.
33350 (ieee128_mfvsrd_64bit): Likewise.
33351 (ieee128_mfvsrd_32bit): Likewise.
33352 (ieee128_mfvsrwz): Likewise.
33353 (ieee128_mtvsrw): Likewise.
33354 (ieee128_mtvsrd_64bit): Likewise.
33355 (ieee128_mtvsrd_32bit): Likewise.
33357 2017-01-31 Martin Liska <mliska@suse.cz>
33360 * ipa-prop.c (ipa_free_all_node_params): Call release method
33361 instead of ~sumbol_summary to not to trigger double times
33364 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33366 PR tree-optimization/71691
33367 * bitmap.h (class auto_bitmap): New.
33368 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33369 is_maybe_undefined instead of ssa_undefined_value_p.
33371 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33373 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33374 __S390_ARCH_LEVEL__ to __ARCH__.
33376 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33378 PR tree-optimization/79267
33379 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33380 if should_remove_lhs_p is true.
33382 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33385 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33386 (add_alignment_attribute): New.
33387 (base_type_die): Add alignment attribute.
33388 (subrange_type_die): Likewise.
33389 (modified_type_die): Likewise.
33390 (gen_array_type_die): Likewise.
33391 (gen_descr_array_type_die: Likewise.
33392 (gen_enumeration_type_die): Likewise.
33393 (gen_subprogram_die): Likewise.
33394 (gen_variable_die): Likewise.
33395 (gen_field_die): Likewise.
33396 (gen_ptr_to_mbr_type_die): Likewise.
33397 (gen_struct_or_union_type_die): Likewise.
33398 (gen_subroutine_type_die): Likewise.
33399 (gen_typedef_die): Likewise.
33400 (base_type_cmp): Compare alignment attribute.
33402 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33405 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33406 (setb_unsigned) New pattern for setb with CCUNS.
33407 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33408 subfc./subfe sequence to avoid overflow problems. Generate a
33409 shorter sequence with cmpld/setb for power9.
33410 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33411 for generating subfc. instruction.
33412 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33413 now uses this instruction.
33415 2017-01-30 Ian Lance Taylor <iant@google.com>
33418 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33419 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33421 2017-01-30 Martin Sebor <msebor@redhat.com>
33423 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33424 Move constant to the right of a relational operator.
33425 (get_mpfr_format_length, format_character, format_string): Ditto.
33426 (should_warn_p, maybe_warn): Same.
33428 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33430 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33433 * asan.c (get_translation_unit_decl): Remove function.
33434 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33436 2017-01-30 Martin Liska <mliska@suse.cz>
33438 PR gcov-profile/79259
33439 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33440 -fprofile-generate.
33442 2017-01-30 Martin Liska <mliska@suse.cz>
33445 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33446 Initialize variables with NULL value.
33448 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33451 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33453 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33455 2017-01-30 Richard Biener <rguenther@suse.de>
33457 PR tree-optimization/79276
33458 * tree-vrp.c (process_assert_insertions): Properly adjust common
33459 when removing a duplicate.
33461 2017-01-30 Richard Biener <rguenther@suse.de>
33463 PR tree-optimization/79256
33464 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33465 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33467 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
33469 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
33472 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
33473 ("*r<noxa>sbg_<mode>_sll_bitmask")
33474 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
33475 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
33476 Use contiguous_bitmask_nowrap_operand.
33478 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33481 * config/rs6000/altivec.h (vec_xl): Revise #define.
33482 (vec_xst): Likewise.
33484 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
33486 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
33488 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
33490 PR rtl-optimization/79194
33491 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33492 traps before call to bypass_conditional_jumps.
33494 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33496 PR tree-optimization/71374
33497 * lra-constraints.c (check_conflict_input_operands): New.
33498 (match_reload): Use it.
33500 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33503 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33504 account to calculate conflict_set.
33506 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33508 PR rtl-optimization/78559
33509 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33510 other_insn in combine.
33512 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33514 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33515 uint16_type_node for BT_UINT16.
33517 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33519 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33520 "RTL Tests" to menu.
33521 (GIMPLE Tests): New node.
33522 (RTL Tests): New node.
33524 2017-01-27 Richard Biener <rguenther@suse.de>
33526 PR tree-optimization/79245
33527 * tree-loop-distribution.c (distribute_loop): Apply cost
33528 modeling also to detected patterns.
33530 2017-01-27 Richard Biener <rguenther@suse.de>
33532 PR tree-optimization/71433
33533 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33534 (compare_assert_loc): New function.
33535 (process_assert_insertions): Sort and optimize assert locations
33536 to remove duplicates and push down identical assertions on
33537 edges to their destination block.
33539 2017-01-27 Richard Biener <rguenther@suse.de>
33541 PR tree-optimization/79244
33542 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33543 out SSA names even if abnormal.
33545 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33547 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33548 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33549 instead of MPFR_RNDN.
33551 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33554 * arm.c (arm_option_override): Don't call build_target_option_node
33555 until after doing all option overrides.
33556 (arm_valid_target_attribute_tree): Likewise.
33558 2017-01-27 Martin Liska <mliska@suse.cz>
33560 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33561 for {cd}tors and C++ {cd}tors.
33563 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33565 * config/s390/s390.md ("*setmem_long_and")
33566 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33568 2017-01-26 Martin Sebor <msebor@redhat.com>
33570 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33573 2017-01-26 Martin Sebor <msebor@redhat.com>
33575 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33576 HAVE_DFmode before using XFmode or DFmode.
33577 (parse_directive): Avoid using the z length modifier to avoid
33578 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33580 PR middle-end/78703
33581 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33582 to accept adjustment as an array.
33583 (get_int_range): New function.
33584 (struct directive): Make width and prec arrays.
33585 (directive::set_width, directive::set_precision): Call get_int_range.
33586 (format_integer, format_floating): Handle width and precision ranges.
33587 (format_string, parse_directive): Same.
33589 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33592 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33593 comdat_type_p set, just clone them, but keep the children in the
33597 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33598 which have direct callers with -fvar-tracking-assignments enabled
33600 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33601 inside of type units.
33603 2017-01-26 Martin Sebor <msebor@redhat.com>
33605 PR middle-end/78703
33606 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33608 (struct format_result): Replace number_chars, number_chars_min,
33609 and number_chars_max with a single member of struct result_range.
33611 (format_result::operator+=): Adjust.
33612 (struct fmtresult): Remove bounded. Handle likely and unlikely
33614 (fmtresult::adjust_for_width_or_precision): New function.
33615 (fmtresult:type_max_digits): New function.
33616 (bytes_remaining): Handle likely and unlikely counters.
33617 (min_bytes_remaining): Remove.
33618 (format_percent): Simplify.
33619 (format_integer, format_floating): Set likely and unlikely counters.
33620 (get_string_length, format_character, format_string): Same.
33621 (format_plain, should_warn_p): New function.
33622 (maybe_warn): Call should_warn_p. Update diagnostic messages
33623 and handle those for all directives, including plain strings.
33624 (format_directive): Handle likely and unlikely counters.
33625 Remove unnecessary quoting from diagnostics. Add an informational
33627 (add_bytes): Remove.
33628 (pass_sprintf_length::compute_format_length): Simplify.
33629 (try_substitute_return_value): Handle likely and unlikely counters.
33631 2017-01-26 Carl Love <cel@us.ibm.com>
33633 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33634 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33636 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33639 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33640 endianess for subregs into account.
33641 * lra-constraints.c (lra_constraints): Do risky transformations
33642 always on the first iteration.
33643 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33644 last_call_used_reg_set.
33645 (process_bb_lives): Define and use last_call_used_reg_set.
33646 * lra.c (lra): Always continue after lra_constraints on the first
33649 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33651 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33653 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33655 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33657 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33658 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33659 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33660 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33661 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33662 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33663 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33664 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33665 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33667 2017-01-26 Marek Polacek <polacek@redhat.com>
33670 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33671 for the third operand.
33673 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33675 PR middle-end/79236
33676 * omp-low.c (struct omp_context): Add simt_stmt field.
33677 (scan_omp_for): Return omp_context *.
33678 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33679 context to the _simt_ SIMD stmt.
33680 (lower_omp_for): For combined SIMD with sibling _simt_
33681 SIMD, make sure to use the same decls in _looptemp_
33682 clauses as in the sibling.
33684 2017-01-26 David Sherwood <david.sherwood@arm.com>
33686 PR middle-end/79212
33687 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33690 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33693 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33694 emit fld b; fld a; if possible.
33696 * brig-builtins.def: Update copyright years.
33697 * config/arm/arm_acle_builtins.def: Update copyright years.
33699 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33702 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33703 constraint instead of o for the stxsd instruction.
33705 2017-01-25 Carl Love <cel@us.ibm.com>
33707 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33708 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33710 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33712 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33714 2017-01-25 Richard Biener <rguenther@suse.de>
33716 PR tree-optimization/69264
33717 * target.def (vector_alignment_reachable): Improve documentation.
33718 * doc/tm.texi: Regenerate.
33719 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33721 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33722 earlier changes with respect to TYPE_USER_ALIGN.
33723 (vector_alignment_reachable_p): Likewise. Improve dumping.
33725 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33728 * config/arm/arm.md (xordi3): Force constant operand into a register
33731 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33733 * doc/invoke.texi (-fstore-merging): Correct default optimization
33734 levels at which it is enabled.
33735 (-O): Move -fstore-merging from list to...
33738 2017-01-25 Richard Biener <rguenther@suse.de>
33741 * omp-expand.c: Include debug.h.
33742 (expand_omp_taskreg): Make sure to generate early debug before
33743 outlining anything from a function.
33744 (expand_omp_target): Likewise.
33745 (grid_expand_target_grid_body): Likewise.
33747 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33750 * asan.c (get_translation_unit_decl): New function.
33751 (asan_add_global): Extract modules file name from globals
33752 TRANSLATION_UNIT_DECL name.
33754 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33757 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33758 for long calls with APCS frame and VFP.
33760 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33762 * cfg.c (original_copy_tables_initialized_p): New function.
33763 * cfg.h (original_copy_tables_initialized_p): New decl.
33764 * cfgrtl.c (relink_block_chain): Guard the call to
33765 free_original_copy_tables with a call to
33766 original_copy_tables_initialized_p.
33767 * cgraph.h (symtab_node::native_rtl_p): New decl.
33768 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33769 (symtab_node::needed_p): Don't assert for early assembly output
33770 for __RTL functions.
33771 (cgraph_node::finalize_function): Set "force_output" for __RTL
33773 (cgraph_node::analyze): Bail out early for __RTL functions.
33774 (analyze_functions): Update assertion to support __RTL functions.
33775 (cgraph_node::expand): Bail out early for __RTL functions.
33776 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33778 * function.h (struct function): Update comment for field
33780 * gimple-expr.c: Include "tree-pass.h".
33781 (gimple_has_body_p): Return false for __RTL functions.
33782 * Makefile.in (OBJS): Add run-rtl-passes.o.
33783 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33785 (gcc::pass_manager::get_clean_slate): New accessor.
33786 * passes.c: Include "insn-addr.h".
33787 (should_skip_pass_p): Add logging. Update logic for running
33788 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33789 property-provider override so it is only done for gimple passes.
33791 (skip_pass): New function.
33792 (execute_one_pass): Call skip_pass when skipping passes.
33793 * read-md.c (md_reader::read_char): Support filtering
33794 the input to a subset of line numbers.
33795 (md_reader::md_reader): Initialize fields
33796 m_first_line and m_last_line.
33797 (md_reader::read_file_fragment): New function.
33798 * read-md.h (md_reader::read_file_fragment): New decl.
33799 (md_reader::m_first_line): New field.
33800 (md_reader::m_last_line): New field.
33801 * read-rtl-function.c (function_reader::create_function): Only
33802 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33803 curr_properties. Set DECL_INITIAL to a dummy block.
33804 (read_rtl_function_body_from_file_range): New function.
33805 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33807 * run-rtl-passes.c: New file.
33808 * run-rtl-passes.h: New file.
33810 2017-01-24 Jeff Law <law@redhat.com>
33812 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33815 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33817 PR tree-optimization/79159
33818 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33819 (record_nonwrapping_iv): Improve boundary using above function if no
33820 value range information.
33822 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33823 Martin Jambor <mjambor@suse.cz>
33825 * brig-builtins.def: New file.
33826 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33827 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33828 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33829 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33830 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33831 * builtin-types.def (BT_INT8): New.
33832 (BT_INT16): Likewise.
33833 (BT_UINT8): Likewise.
33834 (BT_UINT16): Likewise.
33835 (BT_FN_ULONG): Likewise.
33836 (BT_FN_UINT_INT): Likewise.
33837 (BT_FN_UINT_ULONG): Likewise.
33838 (BT_FN_UINT_LONG): Likewise.
33839 (BT_FN_UINT_PTR): Likewise.
33840 (BT_FN_ULONG_PTR): Likewise.
33841 (BT_FN_INT8_FLOAT): Likewise.
33842 (BT_FN_INT16_FLOAT): Likewise.
33843 (BT_FN_UINT32_FLOAT): Likewise.
33844 (BT_FN_UINT16_FLOAT): Likewise.
33845 (BT_FN_UINT8_FLOAT): Likewise.
33846 (BT_FN_UINT64_FLOAT): Likewise.
33847 (BT_FN_UINT16_UINT32): Likewise.
33848 (BT_FN_UINT32_UINT16): Likewise.
33849 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33850 (BT_FN_INT_PTR_INT): Likewise.
33851 (BT_FN_UINT_PTR_UINT): Likewise.
33852 (BT_FN_LONG_PTR_LONG): Likewise.
33853 (BT_FN_ULONG_PTR_ULONG): Likewise.
33854 (BT_FN_VOID_UINT64_UINT64): Likewise.
33855 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33856 (BT_FN_INT8_INT8_INT8): Likewise.
33857 (BT_FN_INT16_INT16_INT16): Likewise.
33858 (BT_FN_INT_INT_INT): Likewise.
33859 (BT_FN_UINT_FLOAT_UINT): Likewise.
33860 (BT_FN_FLOAT_UINT_UINT): Likewise.
33861 (BT_FN_ULONG_UINT_UINT): Likewise.
33862 (BT_FN_ULONG_UINT_PTR): Likewise.
33863 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33864 (BT_FN_UINT_UINT_UINT): Likewise.
33865 (BT_FN_VOID_UINT_PTR): Likewise.
33866 (BT_FN_UINT_UINT_PTR: Likewise.
33867 (BT_FN_UINT32_UINT64_PTR): Likewise.
33868 (BT_FN_INT_INT_UINT_UINT): Likewise.
33869 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33870 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33871 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33872 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33873 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33874 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33875 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33876 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33877 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33878 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33879 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33880 * doc/frontends.texi: List BRIG FE.
33881 * doc/install.texi (Testing): Add BRIG tesring requirements.
33882 * doc/invoke.texi (Overall Options): Mention BRIG.
33883 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33885 2017-01-24 Richard Biener <rguenther@suse.de>
33887 PR translation/79208
33888 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33890 2017-01-24 Martin Jambor <mjambor@suse.cz>
33893 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33894 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33895 and known_contexts.
33897 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33899 PR middle-end/79123
33900 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33901 casts from signed to unsigned really don't have a range.
33903 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33905 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33906 GMP_RNDx for compatiblity.
33908 2017-01-24 Martin Liska <mliska@suse.cz>
33911 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33912 that would prevent us to call alloca with -1 as argument.
33914 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33916 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33917 Avoid -Wformat-security warning.
33919 2017-01-23 Andrew Pinski <apinski@cavium.com>
33921 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33924 2017-01-23 Martin Sebor <msebor@redhat.com>
33926 PR middle-end/78703
33927 * gimple-ssa-sprintf.c (warn_level): New global.
33928 (format_integer): Use it here and throughout the rest of the file.
33929 Use the same switch to compute sign as base.
33930 (maybe_warn): New function.
33931 (format_directive): Factor out warnings into maybe_warn.
33932 Add debugging output. Use warn_level.
33933 (add_bytes): Use warn_level.
33934 (pass_sprintf_length::compute_format_length): Add debugging output.
33935 (try_substitute_return_value): Same.
33936 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33938 PR middle-end/78703
33939 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33940 (struct fmtresult, format_integer, format_floating): Adjust.
33941 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33942 (get_string_length, format_string,format_directive): Same.
33943 (pass_sprintf_length::compute_format_length): Same.
33944 (try_substitute_return_value): Simplify slightly.
33946 PR middle-end/78703
33947 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33948 (fmtresult::operator+=): Outlined.
33949 (struct fmtresult): Add ctors.
33950 (struct conversion_spec): Rename...
33951 (struct directive): ...to this. Add and remove data members.
33952 (directive::set_width, directive::set_precision): New functions.
33953 (format_percent): Use fmtresult ctor.
33954 (get_width_and_precision): Remove.
33955 (format_integer): Make naming changes. Avoid computing width and
33957 (format_floating): Same. Adjust indentation.
33958 (format_character, format_none): New functions.
33959 (format_string): Moved character handling to format_character.
33960 (format_directive): Remove arguments, change return type.
33961 (parse_directive): New function.
33962 (pass_sprintf_length::compute_format_length): Move directive
33963 parsing to parse_directive.
33965 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33967 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33968 (assign_assembler_name_if_needed): ... this.
33969 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33970 (assign_assembler_name_if_needed): ... this.
33971 (free_lang_data_in_cgraph): Adjust callers.
33972 * cgraphunit.c (cgraph_node::analyze): Likewise.
33973 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33976 2017-01-23 Richard Biener <rguenther@suse.de>
33978 PR tree-optimization/79088
33979 PR tree-optimization/79188
33980 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33981 resetting loop bounds after last path deletion. Reset loop
33982 bounds of the target loop, make code match the comments.
33983 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33984 Make sure loops need no fixups.
33986 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33988 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33989 exponent support with double type for first argument.
33990 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33991 type returned by __builtin_vec_extract_sig,
33992 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33993 functions from "vector int" to "vector unsigned int" or from
33994 "vector long long int" to "vector unsigned long long int".
33995 Changed type returned by __builtin_vec_extract_exp,
33996 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33997 functions from "vector int" to "vector unsigned int" or from
33998 "vector long long int" to "vector unsigned long long int".
33999 Changed return type of __builtin_vec_test_data_class,
34000 __builtin_vec_test_data_class_sp, and
34001 __builtin_vec_test_data_class_dp from "vector int" to
34002 "vector bool int" or from "vector long long int" to "vector bool
34003 long long int" and changed second argument type from "unsigned
34004 int" to "int". Added new overloaded function forms "vector float
34005 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34006 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34007 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34008 double, vector unsigned long long int)" and "vector double
34009 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34010 long int)". Changed return type of
34011 __builtin_scalar_test_data_class and
34012 __builtin_scalar_test_data_class_sp and
34013 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34014 int" and changed second argument from "unsigned int" to "int".
34015 Changed type returned by __builtin_scalar_test_neg,
34016 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34017 from "int" to "bool int". Added new overloaded function form
34018 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34019 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34020 exponent double-precision with floating point first argument.
34021 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34022 documentation of scalar_test_data_class, scalar_test_neg,
34023 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34024 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34025 vec_test_data_class built-in functions to reflect refinements in
34026 their type signatures.
34028 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34030 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34032 (aarch64_elf_asm_destructor): Likewise.
34034 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34036 PR rtl-optimization/78634
34037 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34038 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34039 * ifcvt.c (noce_try_cmove): Add missing cost check.
34041 PR rtl-optimization/71724
34042 * combine.c (if_then_else_cond): Look for situations where it is
34043 beneficial to undo the work of one of the recursive calls.
34045 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34047 PR tree-optimization/70754
34048 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34049 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34050 combined stmt before it if not NULL.
34051 (combine_chains): Process refs reversely and compute dominance point
34054 2017-01-23 Martin Liska <mliska@suse.cz>
34056 PR tree-optimization/79196
34057 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34058 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34060 (strlen_optimize_stmt): Call the renamed function.
34062 2017-01-23 Michael Matz <matz@suse.de>
34064 PR tree-optimization/78384
34065 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34067 2017-01-23 Richard Biener <rguenther@suse.de>
34069 PR tree-optimization/79186
34070 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34071 both incoming edges before moving an assert.
34073 2017-01-23 Martin Jambor <mjambor@suse.cz>
34075 * ipa-prop.c (load_from_param_1): Removed.
34076 (load_from_unmodified_param): Bits from load_from_param_1 put back
34078 (load_from_param): Removed.
34079 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34080 with stmt. Reverted back to use of load_from_unmodified_param.
34082 2017-01-23 Martin Jambor <mjambor@suse.cz>
34085 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34086 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34087 field a pointer to garbage collected vector, mark lattices and
34088 ipcp_orig_node with GTY((skip)).
34089 (ipa_get_param_count): Adjust to descriptors being a pointer.
34090 (ipa_get_param): Likewise.
34091 (ipa_get_type): Likewise.
34092 (ipa_get_param_move_cost): Likewise.
34093 (ipa_set_param_used): Likewise.
34094 (ipa_get_controlled_uses): Likewise.
34095 (ipa_set_controlled_uses): Likewise.
34096 (ipa_is_param_used): Likewise.
34097 (ipa_node_params_t): Move into garbage collector. New methods insert
34099 (ipa_node_params_sum): Annotate wth GTY(()).
34100 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34102 (ipa_load_from_parm_agg): Adjust declaration.
34103 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34104 * ipa-profile.c (ipa_profile): Likewise.
34105 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34106 (ipa_populate_param_decls): Make descriptors parameter garbage
34108 (ipa_dump_param): Adjust to descriptors being a pointer.
34109 (ipa_alloc_node_params): Likewise.
34110 (ipa_initialize_node_params): Likewise.
34111 (load_from_param_1): Make descriptors parameter garbage collected.
34112 (load_from_unmodified_param): Likewise.
34113 (load_from_param): Likewise.
34114 (ipa_load_from_parm_agg): Likewise.
34115 (ipa_node_params::~ipa_node_params): Removed.
34116 (ipa_free_all_node_params): Remove call to delete operator.
34117 (ipa_node_params_t::insert): New.
34118 (ipa_node_params_t::remove): Likewise.
34119 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34120 copy known_csts and known_contexts vectors.
34121 (ipa_read_node_info): Adjust to descriptors being a pointer.
34122 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34124 (ipcp_transform_function): Make descriptors variable garbage
34127 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34129 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34130 * config/i386/avx512dqintrin.h: Ditto.
34131 * config/i386/avx512fintrin.h: Ditto.
34132 * config/i386/i386.c: Handle new builtins.
34133 * config/i386/i386-builtin.def: Add new builtins.
34134 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34135 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34137 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34138 Martin Liska <mliska@suse.cz>
34140 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34141 * asan.c (asan_expand_poison_ifn): Support stores and use
34142 appropriate ASAN report function.
34143 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34144 * internal-fn.def (ASAN_POISON_USE): Declare.
34145 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34146 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34147 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34148 ASAN_POISON calls w/o LHS.
34149 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34150 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34151 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34152 * gimplify.c (asan_poison_variables): Add attribute
34153 use_after_scope_memory to variables that really needs to live
34155 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34156 having the attribute.
34158 2017-01-23 Martin Liska <mliska@suse.cz>
34160 * asan.c (create_asan_shadow_var): New function.
34161 (asan_expand_poison_ifn): Likewise.
34162 * asan.h (asan_expand_poison_ifn): New declaration.
34163 * internal-fn.c (expand_ASAN_POISON): Likewise.
34164 * internal-fn.def (ASAN_POISON): New builtin.
34165 * sanopt.c (pass_sanopt::execute): Expand
34166 asan_expand_poison_ifn.
34167 * tree-inline.c (copy_decl_for_dup_finish): Make function
34169 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34170 * tree-ssa.c (is_asan_mark_p): New function.
34171 (execute_update_addresses_taken): Rewrite local variables
34172 (identified just by use-after-scope as addressable) into SSA.
34174 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34176 * doc/install.texi (Specific): opensource.apple.com uses https
34177 now. Remove trailing slash.
34179 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34181 * README.Portability: Remove note on an Irix compatibility issue.
34183 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34185 * gcov.c (INCLUDE_ALGORITHM): Define.
34186 (INCLUDE_VECTOR): Define.
34187 No longer include <vector> and <algorithm> directly.
34189 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34191 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34193 * doc/invoke.texi (Code Gen Options): Ditto.
34195 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34198 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34200 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34202 rtl-optimization/79125
34203 * cprop.c (local_cprop_pass): Handle cases where we make an
34204 unconditional trap.
34206 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34210 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34211 read from, for big endian.
34213 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34215 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34216 register pauth builtins for LP64 only.
34218 2017-01-20 Marek Polacek <polacek@redhat.com>
34221 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34224 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34226 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34228 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34229 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34230 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34232 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34235 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34236 in big-endian mode when they are not a single duplicated value.
34238 2017-01-20 Richard Biener <rguenther@suse.de>
34240 * BASE-VER: Bump to 7.0.1.
34242 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34244 * omp-low.c (omplow_simd_context): New struct. Use it...
34245 (lower_rec_simd_input_clauses): ...here and...
34246 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34247 references to idx, lane, max_vf, is_simt.
34249 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34251 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34254 2017-01-20 Martin Jambor <mjambor@suse.cz>
34256 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34257 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34259 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34260 (GTFILES): Rename hsa.c to hsa-common.c.
34261 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34262 * hsa-dump.c: Likewise.
34263 * hsa-gen.c: Likewise.
34264 * hsa-regalloc.c: Likewise.
34265 * ipa-hsa.c: Likewise.
34266 * omp-expand.c: Likewise.
34267 * omp-low.c: Likewise.
34268 * toplev.c: Likewise.
34270 2017-01-20 Marek Polacek <polacek@redhat.com>
34273 * doc/invoke.texi: Document -Wduplicated-branches.
34274 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34275 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34276 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34277 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34278 return 0 only when not OEP_LEXICOGRAPHIC.
34279 (fold_build_cleanup_point_expr): Use the expression
34280 location when building CLEANUP_POINT_EXPR.
34281 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34282 * tree.c (add_expr): Handle error_mark_node.
34284 2017-01-20 Martin Liska <mliska@suse.cz>
34287 * tree-profile.c (init_ic_make_global_vars): Do not call
34289 (gimple_init_gcov_profiler): Likewise.
34291 2017-01-20 Martin Liska <mliska@suse.cz>
34294 * cgraph.h (maybe_create_reference): Remove argument and
34296 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34298 * ipa-cp.c (create_specialized_node): Likewise.
34299 * symtab.c (symtab_node::maybe_create_reference): Handle
34300 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34302 2017-01-20 Martin Liska <mliska@suse.cz>
34304 * read-rtl-function.c (function_reader::create_function): Use
34305 build_decl instread of build_decl_stat.
34307 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34309 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34310 * config/i386/avx512dqintrin.h: Ditto.
34311 * config/i386/avx512fintrin.h: Ditto.
34312 * config/i386/i386-builtin-types.def: Add new types.
34313 * config/i386/i386.c: Handle new types.
34314 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34315 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34316 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34317 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34318 (__builtin_ia32_kshiftridi): New.
34319 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34321 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34325 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34326 define to rs6000_init_stack_protect_guard.
34327 (rs6000_init_stack_protect_guard): New function.
34329 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34330 Yunqiang Su <yunqiang.su@imgtec.com>
34332 * config.gcc (supported_defaults): Add madd4.
34333 (with_madd4): Add validation.
34334 (all_defaults): Add madd4.
34335 * config/mips/mips.opt (mmadd4): New option.
34336 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34338 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34340 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34341 (ISA_HAS_FUSED_MADD4): Likewise.
34342 * doc/invoke.texi (-mmadd4): Document the new option.
34343 * doc/install.texi (--with-madd4): Document the new option.
34345 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34347 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34348 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34349 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34350 (aarch64_init_pauth_hint_builtins): New.
34351 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34352 (aarch64_expand_builtin): Expand new builtins.
34354 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34356 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34357 * combine-stack-adj.c (no_unhandled_cfa): Handle
34358 REG_CFA_TOGGLE_RA_MANGLE.
34359 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34360 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34361 info for return address signing.
34362 (aarch64_expand_epilogue): Likewise.
34364 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34366 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34367 * config/aarch64/aarch64-protos.h
34368 (aarch64_return_address_signing_enabled): New declaration.
34369 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34371 (aarch64_expand_prologue): Sign return address before it's pushed onto
34373 (aarch64_expand_epilogue): Authenticate return address fetched from
34375 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34376 (aarch64_attributes): New function attributes for "sign-return-address".
34377 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34378 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34379 ("*do_return"): Generate combined instructions according to key index.
34380 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34381 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34383 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34384 * config/aarch64/aarch64.opt (msign-return-address=): New.
34385 * doc/extend.texi (AArch64 Function Attributes): Documents
34386 "sign-return-address=".
34387 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34389 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34391 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34392 overall option summary.
34394 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34396 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34397 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34398 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34399 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34401 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34403 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34404 -mpower9-minmax by default for -mcpu=power9.
34405 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34406 128-bit floating point.
34408 2017-01-20 Alan Modra <amodra@gmail.com>
34410 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34411 optimizing for size.
34413 2017-01-20 Alan Modra <amodra@gmail.com>
34416 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34417 for strcmp and strncmp from corresponding builtin decl.
34419 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34421 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34422 instead of i386/rtems-64.h.
34423 * config/i386/rtems-64.h: Remove.
34425 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34429 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34431 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34433 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34435 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34436 Change int to HOST_WIDE_INT.
34437 * config/aarch64/aarch64-protos.h
34438 (aarch64_simd_gen_const_vector_dup): Likewise.
34439 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34441 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34443 * langhooks-def.h (lhd_type_for_size): New decl.
34444 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34445 * langhooks.c (lhd_type_for_size): New function, taken from
34448 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34450 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34451 define_bypass for CR latency.
34452 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34453 (power9-alu2): Add define_bypass for CR latency.
34455 (power9-mul): Update insn latency.
34456 (power9-mul-compare): Update insn latency, bypass latency and remove
34459 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34461 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
34463 * config/aarch64/aarch64.md
34464 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
34465 aarch64_nopcrelative_literal_loads.
34466 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
34468 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
34470 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
34471 TARGET_LOONGSON_3A.
34472 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
34474 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
34477 * config.gcc (supported_defaults): Add lxc1-sxc1.
34478 (with_lxc1_sxc1): Add validation.
34479 (all_defaults): Add lxc1-sxc1.
34480 * config/mips/mips.opt (mlxc1-sxc1): New option.
34481 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34483 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34484 __mips_no_lxc1_sxc1.
34485 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
34486 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
34487 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
34489 2017-01-19 Richard Biener <rguenther@suse.de>
34491 PR tree-optimization/72488
34492 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34493 sure to restore SSA info.
34494 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34496 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34498 PR rtl-optimization/79121
34499 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34500 of the inner type when shifting an extended value.
34502 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34505 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34506 interposable aliases.
34508 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34511 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34512 Use the evmergelohi instruction.
34513 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34514 (mov_sitf_e500_subreg8_2_be): Likewise.
34515 (mov_sitf_e500_subreg12_2_le): Likewise.
34516 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34517 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34518 (mov_sitf_e500_subreg8_2_le): Likewise.
34519 (mov_sitf_e500_subreg12_2_be): Likewise.
34521 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34523 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34524 attribute from vecsimple to vecperm.
34525 (altivec_vbpermq2): Likewise.
34527 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34530 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34532 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34533 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34534 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34535 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34536 case where N arg is SIZE_MAX.
34537 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34538 (cmpstrsi): Add pattern.
34540 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34542 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34543 __builtin_vec_revb builtins.
34544 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34545 built-in functions to support generation of the ISA 3.0 XXBR<x>
34546 vector byte reverse instructions.
34547 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34548 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34549 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34550 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34551 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34552 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34553 (P9V_BUILTIN_VEC_REVB): Likewise.
34554 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34555 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34556 (p9_xxbrq_v16qi): Likewise.
34557 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34558 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34559 (p9_xxbrh_v8hi): Likewise.
34560 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34561 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34562 vec_revb built-in functions.
34564 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34566 PR rtl-optimization/78952
34567 * config/i386/i386.md (any_extract): New code iterator.
34568 (*insvqi_2): Use any_extract for source operand.
34569 (*insvqi_3): Use any_shiftrt for source operand.
34571 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34573 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34575 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34577 2017-01-18 Matthias Klose <doko@ubuntu.com>
34579 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34581 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34583 * config/rs6000/altivec.h (vec_bperm): Change #define.
34584 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34585 (altivec_vbpermq2): New define_insn.
34586 (altivec_vbpermd): Likewise.
34587 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34588 function interface.
34589 (VBPERMD): Likewise.
34590 (VBPERM): New polymorphic function interface.
34591 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34592 Add entries for P9V_BUILTIN_VEC_VBPERM.
34593 * doc/extend.texi: Add interfaces for vec_bperm.
34595 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34597 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34598 first letter of error messages.
34599 (s390_resolve_overloaded_builtin): Likewise.
34600 * config/s390/s390.c (s390_expand_builtin): Likewise.
34601 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34602 (s390_valid_target_attribute_inner_p): Likewise.
34603 * config/s390/s390.md ("tabort"): Likewise.
34605 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34607 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34608 (ISA_AVOID_DIV_HILO): New macro.
34609 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34610 (ISA_HAS_DDIV): Likewise.
34612 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34614 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34616 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34618 * doc/invoke.texi (fabi-version): Spelling fix.
34620 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34623 * doc/invoke.texi (fabi-version): Mention mangling fix for
34626 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34629 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34631 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34634 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34635 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34637 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34639 (stack_protect_test): Ditto.
34640 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34641 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34643 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34644 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34645 -mstack-protector-guard-offset=.
34646 (RS/6000 and PowerPC Options): Ditto.
34648 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34650 * config/i386/i386.h (MASK_CLASS_P): New define.
34651 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34652 there are no registers from different register sets also when
34653 mask registers are used. Update function comment.
34654 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34655 to (*k/*r) and (*k/*km) alternatives.
34657 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34659 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34660 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34661 (EH_RETURN_HANDLER_RTX): New define.
34662 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34663 Force frame pointer in EH return functions.
34664 (aarch64_expand_epilogue): Add barrier for eh_return.
34665 (aarch64_final_eh_return_addr): Remove.
34666 (aarch64_eh_return_handler_rtx): New function.
34667 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34669 (aarch64_eh_return_handler_rtx): New prototype.
34671 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34673 * config/rs6000/altivec.h (vec_rlmi): New #define.
34674 (vec_vrlnm): Likewise.
34675 (vec_rlnm): Likewise.
34676 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34677 (UNSPEC_VRLNM): Likewise.
34678 (VIlong): New mode iterator.
34679 (altivec_vrl<VI_char>mi): New define_insn.
34680 (altivec_vrl<VI_char>nm): Likewise.
34681 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34683 (VRLDNM): Likewise.
34684 (RLNM): New polymorphic function entry.
34685 (VRLWMI): New monomorphic function entry.
34686 (VRLDMI): Likewise.
34687 (RLMI): New polymorphic function entry.
34688 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34689 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34690 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34693 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34696 * dwarf2out.c (field_byte_offset): Restore the
34697 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34698 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34699 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34702 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34705 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34707 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34710 * dwarf2out.c (add_data_member_location_attribute): For constant
34711 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34712 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34713 DW_AT_byte_size attributes.
34715 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34717 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34718 after forcing to constant memory when the code model is medium.
34720 2017-01-17 Julia Koval <julia.koval@intel.com>
34723 * config/i386/avx512fintrin.h
34724 (_mm512_i32gather_ps): Change __addr type to void const*.
34725 (_mm512_mask_i32gather_ps): Ditto.
34726 (_mm512_i32gather_pd): Ditto.
34727 (_mm512_mask_i32gather_pd): Ditto.
34728 (_mm512_i64gather_ps): Ditto.
34729 (_mm512_mask_i64gather_ps): Ditto.
34730 (_mm512_i64gather_pd): Ditto.
34731 (_mm512_mask_i64gather_pd): Ditto.
34732 (_mm512_i32gather_epi32): Ditto.
34733 (_mm512_mask_i32gather_epi32): Ditto.
34734 (_mm512_i32gather_epi64): Ditto.
34735 (_mm512_mask_i32gather_epi64): Ditto.
34736 (_mm512_i64gather_epi32): Ditto.
34737 (_mm512_mask_i64gather_epi32): Ditto.
34738 (_mm512_i64gather_epi64): Ditto.
34739 (_mm512_mask_i64gather_epi64): Ditto.
34740 (_mm512_i32scatter_ps): Change __addr type to void*.
34741 (_mm512_mask_i32scatter_ps): Ditto.
34742 (_mm512_i32scatter_pd): Ditto.
34743 (_mm512_mask_i32scatter_pd): Ditto.
34744 (_mm512_i64scatter_ps): Ditto.
34745 (_mm512_mask_i64scatter_ps): Ditto.
34746 (_mm512_i64scatter_pd): Ditto.
34747 (_mm512_mask_i64scatter_pd): Ditto.
34748 (_mm512_i32scatter_epi32): Ditto.
34749 (_mm512_mask_i32scatter_epi32): Ditto.
34750 (_mm512_i32scatter_epi64): Ditto.
34751 (_mm512_mask_i32scatter_epi64): Ditto.
34752 (_mm512_i64scatter_epi32): Ditto.
34753 (_mm512_mask_i64scatter_epi32): Ditto.
34754 (_mm512_i64scatter_epi64): Ditto.
34755 (_mm512_mask_i64scatter_epi64): Ditto.
34756 * config/i386/avx512pfintrin.h
34757 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34758 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34759 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34760 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34761 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34762 (_mm512_prefetch_i32scatter_ps): Ditto.
34763 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34764 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34765 (_mm512_prefetch_i64scatter_pd): Ditto.
34766 (_mm512_prefetch_i64scatter_ps): Ditto.
34767 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34768 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34769 * config/i386/avx512vlintrin.h
34770 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34771 (_mm_mmask_i32gather_ps): Ditto.
34772 (_mm256_mmask_i32gather_pd): Ditto.
34773 (_mm_mmask_i32gather_pd): Ditto.
34774 (_mm256_mmask_i64gather_ps): Ditto.
34775 (_mm_mmask_i64gather_ps): Ditto.
34776 (_mm256_mmask_i64gather_pd): Ditto.
34777 (_mm_mmask_i64gather_pd): Ditto.
34778 (_mm256_mmask_i32gather_epi32): Ditto.
34779 (_mm_mmask_i32gather_epi32): Ditto.
34780 (_mm256_mmask_i32gather_epi64): Ditto.
34781 (_mm_mmask_i32gather_epi64): Ditto.
34782 (_mm256_mmask_i64gather_epi32): Ditto.
34783 (_mm_mmask_i64gather_epi32): Ditto.
34784 (_mm256_mmask_i64gather_epi64): Ditto.
34785 (_mm_mmask_i64gather_epi64): Ditto.
34786 (_mm256_i32scatter_ps): Change __addr type to void*.
34787 (_mm256_mask_i32scatter_ps): Ditto.
34788 (_mm_i32scatter_ps): Ditto.
34789 (_mm_mask_i32scatter_ps): Ditto.
34790 (_mm256_i32scatter_pd): Ditto.
34791 (_mm256_mask_i32scatter_pd): Ditto.
34792 (_mm_i32scatter_pd): Ditto.
34793 (_mm_mask_i32scatter_pd): Ditto.
34794 (_mm256_i64scatter_ps): Ditto.
34795 (_mm256_mask_i64scatter_ps): Ditto.
34796 (_mm_i64scatter_ps): Ditto.
34797 (_mm_mask_i64scatter_ps): Ditto.
34798 (_mm256_i64scatter_pd): Ditto.
34799 (_mm256_mask_i64scatter_pd): Ditto.
34800 (_mm_i64scatter_pd): Ditto.
34801 (_mm_mask_i64scatter_pd): Ditto.
34802 (_mm256_i32scatter_epi32): Ditto.
34803 (_mm256_mask_i32scatter_epi32): Ditto.
34804 (_mm_i32scatter_epi32): Ditto.
34805 (_mm_mask_i32scatter_epi32): Ditto.
34806 (_mm256_i32scatter_epi64): Ditto.
34807 (_mm256_mask_i32scatter_epi64): Ditto.
34808 (_mm_i32scatter_epi64): Ditto.
34809 (_mm_mask_i32scatter_epi64): Ditto.
34810 (_mm256_i64scatter_epi32): Ditto.
34811 (_mm256_mask_i64scatter_epi32): Ditto.
34812 (_mm_i64scatter_epi32): Ditto.
34813 (_mm_mask_i64scatter_epi32): Ditto.
34814 (_mm256_i64scatter_epi64): Ditto.
34815 (_mm256_mask_i64scatter_epi64): Ditto.
34816 (_mm_i64scatter_epi64): Ditto.
34817 (_mm_mask_i64scatter_epi64): Ditto.
34818 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34819 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34820 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34821 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34822 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34823 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34824 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34825 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34826 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34827 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34828 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34829 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34830 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34831 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34832 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34833 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34834 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34835 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34836 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34837 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34838 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34839 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34840 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34841 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34842 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34843 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34844 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34845 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34846 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34847 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34848 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34849 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34850 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34851 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34852 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34853 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34854 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34855 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34856 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34857 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34858 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34859 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34860 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34861 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34862 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34863 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34864 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34865 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34866 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34867 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34868 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34869 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34870 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34871 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34872 definitions accordingly.
34874 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34875 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34878 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34881 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34884 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34885 hard regs for inner regno.
34887 2017-01-17 Martin Liska <mliska@suse.cz>
34890 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34891 assumption and add comment.
34893 2017-01-17 Nathan Sidwell <nathan@acm.org>
34895 * ipa-visibility.c (localize_node): New function, broken out of ...
34896 (function_and_variable_visibility): ... here. Call it.
34898 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34900 PR middle-end/77445
34901 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34902 correctly set frequency of oudgoing edge.
34903 (duplicate_thread_path): Fix profile updating.
34905 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34908 * configure.ac: Add GCC_BASE_VER.
34909 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34910 version from BASE-VER file.
34911 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34912 (gcc.o): Depend on $(BASEVER).
34913 * common.opt (dumpfullversion): New option.
34914 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34915 * doc/invoke.texi: Document -dumpfullversion.
34916 * doc/install.texi: Document --with-gcc-major-version-only.
34917 * configure: Regenerated.
34919 2017-01-17 Richard Biener <rguenther@suse.de>
34921 PR tree-optimization/71433
34922 * tree-vrp.c (register_new_assert_for): Merge same asserts
34923 on all incoming edges.
34924 (process_assert_insertions_for): Handle insertions at the
34927 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34929 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34930 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34932 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34935 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34938 2017-01-17 Alan Modra <amodra@gmail.com>
34941 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34942 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34943 symbolic stack limit when pic.
34945 2017-01-16 Martin Sebor <msebor@redhat.com>
34947 PR tree-optimization/78608
34948 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34950 2017-01-16 Jeff Law <law@redhat.com>
34953 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34954 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34955 for several include directories that may be relative to sysroot.
34956 * config/i386/x-mingw32 (gplus_includedir): Define.
34957 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34958 (native_system_includedir): Likewise.
34959 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34960 override if TARGET_SYSTEM_ROOT is defined.
34961 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34963 PR tree-optimization/79090
34964 PR tree-optimization/33562
34965 PR tree-optimization/61912
34966 PR tree-optimization/77485
34967 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34968 and computed trims into the dump file.
34970 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34972 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34974 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34977 * gimplify.c (gimplify_init_constructor): If want_value and
34978 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34982 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34983 sequence. Formatting fixes.
34984 (doloop_optimize): Formatting fixes.
34987 * gcc.c (debug_level_greater_than_spec_func): New function.
34988 (static_spec_functions): Add debug-level-gt spec function.
34989 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34991 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34992 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34993 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34994 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34995 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34996 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34998 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35000 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35001 QImode fixups to general and mask registers only.
35003 2017-01-16 Carl Love <cel@us.ibm.com>
35005 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35006 for built-in functions
35007 vector signed char vec_nabs (vector signed char)
35008 vector signed short vec_nabs (vector signed short)
35009 vector signed int vec_nabs (vector signed int)
35010 vector signed long long vec_nabs (vector signed long long)
35011 vector float vec_nabs (vector float)
35012 vector double vec_nabs (vector double)
35013 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35015 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35016 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35017 * doc/extend.texi: Update the documentation file for the new built-in
35020 2017-01-16 Martin Sebor <msebor@redhat.com>
35022 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35025 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35027 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35028 UNSPEC_VSX__XXSPLTD to require special splat handling.
35030 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35033 * system.h: Poison strndup.
35035 2017-01-16 Alan Modra <amodra@gmail.com>
35038 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35041 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35043 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35045 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35047 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35048 call recog here. Assert that INSN_CODE (insn) is non-negative.
35050 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35053 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35055 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35056 in the currently scheduled RTL fragment.
35058 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35060 PR rtl-optimization/78751
35061 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35064 2017-01-14 Jeff Law <law@redhat.com>
35066 PR tree-optimization/79090
35067 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35068 variable length stores.
35069 (compute_trims): Delete dead assignment to *trim_tail.
35070 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35073 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35075 PR rtl-optimization/78626
35076 PR rtl-optimization/78727
35077 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35078 of a block, and split such blocks after everything else is finished.
35080 2017-01-14 Alan Modra <amodra@gmail.com>
35083 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35084 target legitimate_combined_insn.
35085 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35086 (rs6000_legitimate_combined_insn): New function.
35087 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35089 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35090 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35091 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35093 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35095 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35097 2017-01-13 Jeff Law <law@redhat.com>
35099 PR tree-optimization/33562
35100 PR tree-optimization/61912
35101 PR tree-optimization/77485
35102 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35104 (delete_dead_assignment): Likewise.
35105 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35106 statement to delete_dead_call and delete_dead_assignment.
35108 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35111 * substring-locations.c (format_warning_va): Strengthen case 1 so
35112 that both endpoints of the substring must be within the format
35113 range for just the substring to be printed.
35115 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35117 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35118 * config/i386/i386.c (ix86_target_string): Add missing options
35119 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35120 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35121 flags_other and ix86_target_other to flags2_other. Display unknown
35123 (ix86_valid_target_attribute_inner_p): Add missing options and
35124 reorder options by implied ISAs, as in ix86_target_string.
35126 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35128 * hash-table.h (hash_table::too_empty_p): New function.
35129 (hash_table::expand): Use it.
35130 (hash_table::traverse): Likewise.
35131 (hash_table::empty_slot): Use sizeof (value_type) instead of
35132 sizeof (PTR) to convert bytes to elements. Shrink the table
35133 if the current size is excessive for the current number of
35136 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35138 * ira-costs.c (record_reg_classes): Break from the inner loop
35139 early once alt_fail is known to be true. Update outer loop
35140 handling accordingly.
35142 2017-01-13 Jeff Law <law@redhat.com>
35144 * tree-ssa-dse.c (decrement_count): New function.
35145 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35146 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35147 when we know the partially dead statement is a mem* function.
35149 PR tree-optimization/61912
35150 PR tree-optimization/77485
35151 * tree-ssa-dse.c: Include expr.h.
35152 (maybe_trim_constructor_store): New function.
35153 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35155 PR tree-optimization/33562
35156 PR tree-optimization/61912
35157 PR tree-optimization/77485
35158 * doc/invoke.texi: Document new dse-max-object-size param.
35159 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35160 * tree-ssa-dse.c: Include params.h.
35161 (dse_store_status): New enum.
35162 (initialize_ao_ref_for_dse): New, partially extracted from
35164 (valid_ao_ref_for_dse, normalize_ref): New.
35165 (setup_live_bytes_from_ref, compute_trims): Likewise.
35166 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35167 (maybe_trim_partially_dead_store): Likewise.
35168 (maybe_trim_complex_store): Likewise.
35169 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35170 Track what bytes live from the original store. Return tri-state
35171 for dead, partially dead or live.
35172 (dse_dom_walker): Add constructor, destructor and new private members.
35173 (delete_dead_call, delete_dead_assignment): New extracted from
35175 (dse_optimize_stmt): Make a member of dse_dom_walker.
35176 Use initialize_ao_ref_for_dse.
35178 PR tree-optimization/33562
35179 PR tree-optimization/61912
35180 PR tree-optimization/77485
35181 * sbitmap.h (bitmap_count_bits): Prototype.
35182 (bitmap_clear_range, bitmap_set_range): Likewise.
35183 * sbitmap.c (bitmap_clear_range): New function.
35184 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35186 2017-01-13 Martin Liska <mliska@suse.cz>
35189 * function.c (set_cfun): Add new argument force.
35190 * function.h (set_cfun): Likewise.
35191 * ipa-inline-transform.c (inline_call): Use the function when
35192 strict alising from is dropped for function we inline to.
35194 2017-01-13 Richard Biener <rguenther@suse.de>
35196 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35197 for dumping GIMPLE INTEGER_CSTs.
35199 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35201 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35202 to 201112L since C++17.
35204 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35207 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35208 if -fsanitize=kernel-address is present.
35210 2017-01-13 Richard Biener <rguenther@suse.de>
35212 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35213 as _Literal ( type ) number in case usual suffixes do not
35214 preserve all information.
35216 2017-01-13 Richard Biener <rguenther@suse.de>
35218 PR tree-optimization/77283
35219 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35220 and ssa-iterators.h.
35221 (is_feasible_trace): Implement a cost model based on joiner
35224 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35227 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35228 char or short to __float128/_Float128 directly.
35230 2017-01-12 Martin Sebor <msebor@redhat.com>
35232 to -Wformat-overflow.
35233 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35234 (min_bytes_remaining): Same.
35235 (get_string_length): Same.
35236 (format_string): Same.
35237 (format_directive): Same.
35239 (pass_sprintf_length::handle_gimple_call): Same.
35241 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35243 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35244 info.nowrite calls with no lhs that can't throw. Return bool
35245 whether gsi_remove has been called or not.
35246 (pass_sprintf_length::handle_gimple_call): Return bool whether
35247 try_substitute_return_value called gsi_remove. Formatting fix.
35248 (pass_sprintf_length::execute): Don't use gsi_remove if
35249 handle_gimple_call returned true.
35252 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35253 be removed due to side-effects, don't remove following barrier nor
35254 turn the successor edge into fallthru edge.
35256 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35259 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35260 element-reversing loads and stores as not swappable.
35262 2017-01-12 Nathan Sidwell <nathan@acm.org>
35263 Nicolai Stange <nicstange@gmail.com>
35265 * combine.c (try_combine): Don't ignore result of overlap checking
35266 loop. Combine overlap & asm check into single loop.
35268 2017-01-12 Richard Biener <rguenther@suse.de>
35270 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35271 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35273 2017-01-12 Richard Biener <rguenther@suse.de>
35275 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35276 and TS_TARGET_OPTION directly derive from TS_BASE.
35277 * tree-core.h (tree_optimization_option): Derive from tree_base.
35278 (tree_target_option): Likewise.
35280 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35282 * config/i386/i386.c (memory_address_length): Increase len
35283 only when rip_relative_addr_p returns false.
35285 2017-01-11 Julia Koval <julia.koval@intel.com>
35287 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35288 (OPTION_MASK_ISA_SGX_SET): New.
35289 (ix86_handle_option): Handle OPT_msgx.
35290 * config.gcc: Added sgxintrin.h.
35291 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35292 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35293 * config/i386/i386.c (ix86_target_string): Add -msgx.
35295 (ix86_option_override_internal): Handle new options.
35296 (ix86_valid_target_attribute_inner_p): Add sgx.
35297 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35298 * config/i386/i386.opt: Add msgx.
35299 * config/i386/sgxintrin.h: New file.
35300 * config/i386/x86intrin.h: Add sgxintrin.h.
35302 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35305 * fold-const.c (maybe_nonzero_address): Return 1 for function
35307 (tree_single_nonzero_warnv_p): Don't handle function local objects
35311 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35314 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35317 * opts.c: Include "spellcheck.h"
35318 (struct string_fragment): New struct.
35319 (struct edit_distance_traits<const string_fragment &>): New
35321 (get_closest_sanitizer_option): New function.
35322 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35324 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35326 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35328 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35329 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35330 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35331 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35332 for initial die_offset if dwarf_split_debug_info.
35333 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35334 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35336 (output_skeleton_debug_sections): Formatting fix. Use
35337 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35338 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35340 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35342 * config/arm/cortex-a53.md: Add bypasses for
35343 cortex_a53_r2f_cvt.
35344 (cortex_a53_r2f): Only use for transfers.
35345 (cortex_a53_f2r): Likewise.
35346 (cortex_a53_r2f_cvt): Add reservation for conversions.
35347 (cortex_a53_f2r_cvt): Likewise.
35349 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35351 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35352 to all inlined functions, change static to extern.
35354 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35357 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35359 (arm_assemble_integer): Likewise.
35361 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35363 * config.gcc: Use new awk script to check CPU, FPU and architecture
35364 parameters for --with-... options.
35365 * config/arm/parsecpu.awk: New file
35366 * config/arm/arm-cpus.in: New file.
35367 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35369 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35371 * config/arm/t-arm: Update dependency rules.
35372 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35373 of processing .def files.
35374 * config/arm/genopt.sh: Deleted.
35375 * config/arm/gentune.sh: Deleted.
35376 * config/arm/arm-cores.def: Deleted.
35377 * config/arm/arm-arches.def: Deleted.
35378 * config/arm/arm-fpus.def: Deleted.
35379 * config/arm/arm-tune.md: Regenerated.
35380 * config/arm/arm-tables.opt: Regenerated.
35381 * config/arm/arm-cpu.h: New generated file.
35382 * config/arm/arm-cpu-data.h: New generated file.
35383 * config/arm/arm-cpu-cdata.h: New generated file.
35385 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35388 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35390 (input_varpool_node): Unpack dynamically_initialized bit.
35392 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35394 PR rtl-optimization/79032
35395 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35396 the alignment of the adjusted memory reference against that of MODE,
35397 instead of the alignment of the original memory reference.
35399 2017-01-11 Martin Jambor <mjambor@suse.cz>
35401 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35403 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35404 decorated functions.
35406 2017-01-11 Richard Biener <rguenther@suse.de>
35408 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35409 set range/nonnull info for PHI results. Do not set it on
35410 stmts marked for removal.
35412 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35414 * expr.c (store_field): In the bitfield case, fetch the return value
35415 from the registers before applying a single big-endian adjustment.
35416 Always do a final load for a BLKmode value not larger than a word.
35418 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35421 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35422 that we correctly handle column numbers greater than
35423 LINE_MAP_MAX_COLUMN_NUMBER.
35425 2017-01-10 Martin Sebor <msebor@redhat.com>
35427 PR middle-end/78245
35428 * gimple-ssa-sprintf.c (get_destination_size): Call
35429 {init,fini}object_sizes.
35430 * tree-object-size.c (addr_object_size): Adjust.
35431 (pass_through_call): Adjust.
35432 (pass_object_sizes::execute): Adjust.
35433 * tree-object-size.h (fini_object_sizes): Declare.
35435 2017-01-10 Martin Sebor <msebor@redhat.com>
35437 PR tree-optimization/78775
35438 * builtins.c (get_size_range): Move...
35439 * calls.c: ...to here.
35440 (alloc_max_size): Accept zero argument.
35441 (operand_signed_p): Remove.
35442 (maybe_warn_alloc_args_overflow): Call get_size_range.
35443 * calls.h (get_size_range): Declare.
35445 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35447 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35448 from TI's devices.csv file as of September 2016.
35449 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35451 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35453 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35454 * doc/invoke.texi: Likewise.
35455 * doc/md.texi: Likewise.
35456 * doc/objc.texi: Likewise.
35458 2017-01-10 Joshua Conner <joshconner@google.com>
35460 * config/arm/fuchsia-elf.h: New file.
35461 * config/fuchsia.h: New file.
35462 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
35463 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
35465 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
35467 2016-01-10 Richard Biener <rguenther@suse.de>
35469 PR tree-optimization/79034
35470 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35471 Propagate out degenerate PHIs in the joiner.
35473 2017-01-10 Martin Liska <mliska@suse.cz>
35475 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
35476 (sort_congruence_classes_by_decl_uid): Likewise.
35477 (sort_congruence_class_groups_by_decl_uid): Likewise.
35478 (sem_item_optimizer::merge_classes): Sort class, groups in these
35479 classes and members in the groups by DECL_UID of declarations.
35480 This would make merge operations stable.
35482 2017-01-10 Martin Liska <mliska@suse.cz>
35484 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
35485 usage of m_classes_vec.
35486 (sem_item_optimizer::~sem_item_optimizer): Likewise.
35487 (sem_item_optimizer::get_group_by_hash): Likewise.
35488 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
35489 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
35490 (sem_item_optimizer::verify_classes): Likewise.
35491 (sem_item_optimizer::process_cong_reduction): Likewise.
35492 (sem_item_optimizer::dump_cong_classes): Likewise.
35493 (sem_item_optimizer::merge_classes): Likewise.
35494 * ipa-icf.h (congruence_class_hash): Rename from
35495 congruence_class_group_hash. Remove declaration of m_classes_vec.
35497 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35499 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35500 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35501 * config.gcc: Add avx512vpopcntdqintrin.h.
35502 * config/i386/avx512vpopcntdqintrin.h: New.
35503 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35504 * config/i386/i386-builtin-types.def: Add new types.
35505 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35506 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35507 __builtin_ia32_vpopcountq_v8di_mask): New.
35508 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35509 __AVX512VPOPCNTDQ__.
35510 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35511 (PTA_AVX512VPOPCNTDQ): Define.
35512 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35513 TARGET_AVX512VPOPCNTDQ_P): Define.
35514 * config/i386/i386.opt: Add mavx512vpopcntdq.
35515 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35516 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35518 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35520 PR middle-end/77484
35521 * predict.def (PRED_CALL): Set to 67.
35523 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35525 * expr.c (store_field): In the bitfield case, if the value comes from
35526 a function call and is of an aggregate type returned in registers, do
35527 not modify the field mode; extract the value in all cases if the mode
35528 is BLKmode and the size is not larger than a word.
35530 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35533 * config/i386/cpuid.h: Fix undefined behavior.
35535 2017-01-04 Jeff Law <law@redhat.com>
35537 PR tree-optimization/79007
35538 PR tree-optimization/67955
35539 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35540 conservative for pt.null when flag_non_call_exceptions is on.
35542 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35544 PR translation/79019
35545 PR translation/79020
35546 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35547 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35548 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35550 * config/avr/avr.opt (maccumulate-args): Likewise.
35551 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35552 * common.opt (freport-bug): Likewise.
35553 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35554 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35555 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35556 translatable string.
35557 * config/i386/i386.c (function_value_32): Likewise.
35558 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35559 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35561 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35562 * common/config/msp430/msp430-common.c (msp430_handle_option):
35564 * symtab.c (symtab_node::verify_base): Likewise.
35565 * opts.c (set_debug_level): Likewise.
35566 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35567 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35568 missing whitespace to translatable strings.
35569 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35570 * config/sh/superh.h: Likewise.
35571 * config/i386/xopintrin.h: Likewise.
35572 * config/i386/znver1.md: Likewise.
35573 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35574 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35575 * double-int.h (struct double_int): Likewise.
35576 * double-int.c (div_and_round_double): Likewise.
35577 * wide-int.cc: Likewise.
35578 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35579 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35580 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35581 (crossjumps_occurred): ... this.
35582 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35585 PR tree-optimization/78899
35586 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35587 returning bool return struct loop *, NULL for failure and the new
35589 (versionable_outer_loop_p): Don't version outer loop if it has
35590 dont_vectorized bit set.
35591 (tree_if_conversion): When versioning outer loop, ensure
35592 tree_if_conversion is performed also on the inner loop of the
35593 non-vectorizable outer loop copy.
35594 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35595 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35596 prevent vectorization of it.
35597 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35598 the outer loop vectorization of the non-scalar version is attempted
35599 before vectorization of the inner loop in scalar version. If
35600 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35601 vectorization of its inner loop.
35602 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35603 has 2 inner loops, rename also on edges from bb whose single pred
35604 is outer_loop->header. Fix typo in function comment.
35606 2017-01-09 Martin Sebor <msebor@redhat.com>
35609 * asan.c (asan_emit_stack_protection): Increase local buffer size
35610 to avoid snprintf truncation warning.
35612 2017-01-09 Andrew Pinski <apinski@cavium.com>
35614 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35615 to reference thunderx2t99 for the tuning structure
35616 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35618 (thunderx2t99_extra_costs): This.
35619 * config/aarch64/aarch64-tune.md: Regenerate.
35620 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35621 (vulcan_addrcost_table): This.
35622 (vulcan_regmove_cost): Rename to ...
35623 (thunderx2t99_regmove_cost): This.
35624 (vulcan_vector_cost): Rename to ...
35625 (thunderx2t99_vector_cost): this.
35626 (vulcan_branch_cost): Rename to ...
35627 (thunderx2t99_branch_cost): This.
35628 (vulcan_tunings): Rename to ...
35629 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35630 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35632 2017-01-09 Martin Jambor <mjambor@suse.cz>
35636 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35637 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35638 (propagate_vr_accross_jump_function): Use the above function for all
35639 value range computations for pass-through jump functions and type
35640 converasion from explicit value range values.
35641 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35642 parameters from TYPE_ARG_TYPES.
35643 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35644 (ipa_write_node_info): Stream type of the actual argument.
35645 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35647 2017-01-09 Martin Liska <mliska@suse.cz>
35650 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35651 (lookup_compiler): Do not show error message with have_E.
35653 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35655 PR tree-optimization/78938
35656 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35657 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35658 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35659 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35662 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35664 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35667 2017-01-09 Richard Biener <rguenther@suse.de>
35669 PR tree-optimization/78997
35670 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35671 name condition properly.
35673 2017-01-09 Richard Biener <rguenther@suse.de>
35676 * dwarf2out.c (is_cxx): New overload with context.
35677 (is_naming_typedef_decl): Use it.
35679 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35681 * invoke.texi (Option Summary): Correct spacing in option lists
35682 and add line breaks to fix over-long lines.
35684 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35686 PR middle-end/17660
35688 * extend.texi (Common Variable Attributes): Add xref to GCC
35689 Internals manual to explain mode attribute keywords.
35691 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35694 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35695 and Preprocessor Options.
35696 (Options for Linking): Document -pthread here....
35697 (RS/6000 and PowerPC Options): ...not here.
35698 (Solaris 2 Options): ...or here.
35699 * doc/cppopts.texi: Document -pthread.
35701 2017-01-08 Martin Sebor <msebor@redhat.com>
35703 PR middle-end/77708
35704 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35705 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35706 New member functions.
35707 (format_directive): Used them.
35709 (pass_sprintf_length::handle_gimple_call): Same.
35710 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35711 to avoid truncation for any argument.
35712 (extract_affine_mul): Same.
35713 * tree.c (get_file_function_name): Same.
35715 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35717 PR middle-end/77484
35718 * predict.def (PRED_INDIR_CALL): Set to 86.
35720 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35722 PR preprocessor/54124
35723 * doc/cppopts.texi: Reformat -d subtable to list the full name
35724 of the options. Add cross-reference to the docs for the general
35725 compiler -d options.
35726 * doc/invoke.texi (Developer Options): Add cross-reference to the
35727 preprocessor-specific -d option documentation.
35729 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35731 PR preprocessor/13498
35732 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35733 redudant material, and reflect new command-line options.
35734 (System Headers): Likewise.
35736 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35738 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35739 -isystem, and -idirafter. Copy-edit.
35740 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35741 default for -ftrack-macro-expansion. Delete obsolete and
35742 badly-formatted implementation details about -fdebug-cpp output.
35743 * doc/cppwarnopts.texi: Copy-edit.
35745 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35748 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35749 that the transition from a max line width >= 1<<10 to narrower
35750 lines works correctly.
35752 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35754 * doc/options.texi (PerFunction): New.
35755 * opt-functions.awk (switch_flags): Map both Optimization and
35756 PerFunction to CL_OPTIMIZATION.
35757 * opth-gen.awk: Test for PerFunction flag along with
35759 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35760 it only when the latter is present. Skip those that don't in
35761 the hash function generator.
35762 * common.opt (fvar-tracking): Mark as PerFunction instead of
35764 (fvar-tracking-assignments): Likewise.
35765 (fvar-tracking-assignments-toggle): Likewise.
35766 (fvar-tracking-uninit): Likewise.
35768 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35770 PR translation/79018
35771 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35774 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35777 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35778 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35779 TARGET_LONG_JUMP_TABLE_OFFSETS.
35780 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35781 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35782 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35783 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35784 * config/m68k/m68k.md (tablejump expander): Likewise.
35785 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35786 TARGET_LONG_JUMP_TABLE_OFFSETS.
35787 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35788 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35790 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35791 David Holsgrove <david.holsgrove@xilinx.com>
35793 * common/config/microblaze/microblaze-common.c
35794 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35795 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35797 * config/microblaze/microblaze.c (microblaze_must_save_register)
35798 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35800 (microblaze_eh_return): New function.
35801 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35802 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35803 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35804 * config/microblaze/microblaze.md (eh_return): New pattern.
35806 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35808 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35809 GCC_DIAGNOSTIC_STRINGIFY): Define.
35811 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35813 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35815 * config/arm/arm.md (<mcrr>): New.
35817 * config/arm/arm.c (arm_arch5te): New.
35818 (arm_option_override): Set arm_arch5te.
35819 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35821 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35822 (arm_mcrr_qualifiers): ... this. New.
35823 (MRRC_QUALIFIERS): Define to...
35824 (arm_mrrc_qualifiers): ... this. New.
35825 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35827 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35828 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35829 (MRRCI, mrrc, MRRC): New.
35830 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35831 VUNSPEC_MRRC2): New.
35833 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35835 * config/arm/arm.md (<mcr>): New.
35837 * config/arm/arm.c (arm_coproc_builtin_available): Add
35838 support for mcr, mrc, mcr2 and mrc2.
35839 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35840 (arm_mcr_qualifiers): ... this. New.
35841 (MRC_QUALIFIERS): Define to ...
35842 (arm_mrc_qualifiers): ... this. New.
35843 (MCR_QUALIFIERS): Define to ...
35844 (arm_mcr_qualifiers): ... this. New.
35845 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35847 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35848 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35849 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35850 VUNSPEC_MRC2): New.
35852 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35854 * config/arm/arm.md (*ldc): New.
35858 * config/arm/arm.c (arm_coproc_builtin_available): Add
35859 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35860 (arm_coproc_ldc_stc_legitimate_address): New.
35861 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35862 'qualifier_const_pointer'.
35863 (LDC_QUALIFIERS): Define to...
35864 (arm_ldc_qualifiers): ... this. New.
35865 (STC_QUALIFIERS): Define to...
35866 (arm_stc_qualifiers): ... this. New.
35867 * config/arm/arm-protos.h
35868 (arm_coproc_ldc_stc_legitimate_address): New.
35869 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35870 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35871 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35872 stc2, stcl, stc2l): New.
35873 * config/arm/constraints.md (Uz): New.
35874 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35875 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35876 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35877 VUNSPEC_STC2L): New.
35879 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35881 * config/arm/arm.md (<cdp>): New.
35882 * config/arm/arm.c (neon_const_bounds): Rename this ...
35883 (arm_const_bounds): ... this.
35884 (arm_coproc_builtin_available): New.
35885 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35886 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35887 (CDP_QUALIFIERS): Define to...
35888 (arm_cdp_qualifiers): ... this. New.
35890 (arm_expand_builtin_args): Add case for 6 arguments.
35891 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35892 (arm_const_bounds): ... this.
35893 (arm_coproc_builtin_available): New.
35894 * config/arm/arm_acle.h (__arm_cdp): New.
35896 * config/arm/arm_acle_builtins.def (cdp): New.
35898 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35899 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35900 'arm_const_bounds'.
35901 * config/arm/types.md (coproc): New.
35902 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35903 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35904 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35905 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35907 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35909 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35910 (UBINOP_QUALIFIERS): New.
35912 (acle_builtin_data): New. Change comment.
35913 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35914 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35915 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35916 arm_acle_builtins.def.
35917 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35918 (arm_init_acle_builtins): New.
35919 (CRC32_BUILTIN): Remove.
35920 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35921 crc32cb, crc32ch and crc32cw.
35922 (arm_init_crc32_builtins): Remove.
35923 (arm_init_builtins): Use arm_init_acle_builtins rather
35924 than arm_init_crc32_builtins.
35925 (arm_expand_acle_builtin): New.
35926 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35927 * config/arm/arm_acle_builtins.def: New.
35929 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35931 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35932 (arm_builtin_datum): ... this.
35933 (arm_init_neon_builtin): Rename to ...
35934 (arm_init_builtin): ... this. Add a new parameters PREFIX
35935 and USE_SIG_IN_NAME.
35936 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35937 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35938 'arm_builtin_datum'.
35939 (arm_init_vfp_builtins): Likewise.
35940 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35941 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35942 (arm_expand_neon_args): Rename to ...
35943 (arm_expand_builtin_args): ... this. Rename builtin_arg
35944 enum values and differentiate between ARG_BUILTIN_MEMORY
35945 and ARG_BUILTIN_NEON_MEMORY.
35946 (arm_expand_neon_builtin_1): Rename to ...
35947 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35948 values, arm_expand_builtin_args and add bool parameter NEON.
35949 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35950 (arm_expand_vfp_builtin): Likewise.
35951 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35953 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35955 PR middle-end/77484
35956 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35957 * predict.c (tree_estimate_probability_bb): Reverse direction of
35958 polymorphic call predictor.
35960 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35962 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35963 (determine_pass_name_match): ...this new function and...
35964 (should_skip_pass_p): ...this new function.
35966 2017-01-06 Nathan Sidwell <nathan@acm.org>
35968 * ipa-visibility.c (function_and_variable_visibility): Reformat
35969 comments and long lines. Remove extrneous if.
35970 * symtab.c (symtab_node::make_decl_local): Fix code format.
35971 (symtab_node::set_section_for_node): Fix comment typo.
35973 2017-01-06 Martin Liska <mliska@suse.cz>
35976 * lra-constraints.c: Rename invariant to lra_invariant.
35977 * predict.c (set_even_probabilities): Initialize e to NULL.
35979 2017-01-05 Martin Sebor <msebor@redhat.com>
35981 PR tree-optimization/78910
35982 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35983 (format_integer): Correct off-by-one error in the handling
35984 of precision with negative numbers in signed conversions..
35986 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35988 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35990 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35992 PR tree-optimization/71016
35993 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35994 factor_out_conditional_conversion. Formatting fix.
35995 (factor_out_conditional_conversion): Add cond_stmt argument.
35996 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35997 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36000 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36002 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36003 read-rtl-function.o, and selftest-rtl.o.
36004 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36005 (selftest::aarch64_test_loading_full_dump): New function.
36006 (selftest::aarch64_run_selftests): New function.
36007 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36008 selftest::aarch64_run_selftests.
36009 * config/i386/i386.c
36010 (selftest::ix86_test_loading_dump_fragment_1): New function.
36011 (selftest::ix86_test_loading_call_insn): New function.
36012 (selftest::ix86_test_loading_full_dump): New function.
36013 (selftest::ix86_test_loading_unspec): New function.
36014 (selftest::ix86_run_selftests): Call the new functions.
36015 * emit-rtl.c (maybe_set_max_label_num): New function.
36016 * emit-rtl.h (maybe_set_max_label_num): New decl.
36017 * function.c (instantiate_decls): Guard call to
36018 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36019 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36021 * gensupport.c (gen_reader::gen_reader): Pass "false"
36022 for new "compact" param of rtx_reader.
36023 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36024 rather than an empty string for NULL strings.
36025 * read-md.c: Potentially include config.h rather than bconfig.h.
36026 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36027 (have_error): New global, copied from errors.c.
36028 (md_reader::read_name): Rename to...
36029 (md_reader::read_name_1): ...this, adding "out_loc" param,
36030 and converting "missing name or number" to returning false, rather
36032 (md_reader::read_name): Reimplement in terms of read_name_1.
36033 (md_reader::read_name_or_nil): New function.
36034 (md_reader::read_string): Handle "(nil)" by returning NULL.
36035 (md_reader::md_reader): Add new param "compact".
36036 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36037 (md_reader::read_file): New method.
36038 * read-md.h (md_reader::md_reader): Add new param "compact".
36039 (md_reader::read_file): New method.
36040 (md_reader::is_compact): New accessor.
36041 (md_reader::read_name): Convert return type from void to file_location.
36042 (md_reader::read_name_or_nil): New decl.
36043 (md_reader::read_name_1): New decl.
36044 (md_reader::m_compact): New field.
36045 (noop_reader::noop_reader): Pass "false" for new "compact" param
36047 (rtx_reader::rtx_reader): Add new "compact" param.
36048 (rtx_reader::read_rtx_operand): Make virtual and convert return
36049 type from void to rtx.
36050 (rtx_reader::read_until): New decl.
36051 (rtx_reader::handle_any_trailing_information): New virtual function.
36052 (rtx_reader::postprocess): New virtual function.
36053 (rtx_reader::finalize_string): New virtual function.
36054 (rtx_reader::m_in_call_function_usage): New field.
36055 (rtx_reader::m_reuse_rtx_by_id): New field.
36056 * read-rtl-function.c: New file.
36057 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36058 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36059 (selftest::verify_three_block_rtl_cfg): New decl.
36060 * read-rtl-function.h: New file.
36061 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36062 For host, include function.h, memmodel.h, and emit-rtl.h.
36063 (one_time_initialization): New function.
36064 (struct compact_insn_name): New struct.
36065 (compact_insn_names): New array.
36066 (find_code): Handle insn codes in compact dumps.
36067 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36068 (bind_subst_iter_and_attr): Likewise.
36069 (add_condition_to_string): Likewise.
36070 (add_condition_to_rtx): Likewise.
36071 (apply_attribute_uses): Likewise.
36072 (add_current_iterators): Likewise.
36073 (apply_iterators): Likewise.
36074 (initialize_iterators): Guard usage of apply_subst_iterator with
36075 #ifdef GENERATOR_FILE.
36076 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36077 (md_reader::read_mapping): Likewise.
36078 (add_define_attr_for_define_subst): Likewise.
36079 (add_define_subst_attr): Likewise.
36080 (read_subst_mapping): Likewise.
36081 (check_code_iterator): Likewise.
36082 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36084 (one_time_initialization): New function.
36085 (rtx_reader::read_until): New method.
36086 (read_flags): New function.
36087 (parse_reg_note_name): New function.
36088 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36089 Handle reuse_rtx ids.
36090 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36091 Add parsing support for RTL dumps, mirroring the special-cases in
36092 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36093 values, and calling handle_any_trailing_information.
36094 (rtx_reader::read_rtx_operand): Convert return type from void
36095 to rtx, returning return_rtx. Handle case 'e'. Call
36096 finalize_string on XSTR and XTMPL fields.
36097 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36098 "(nil)" values were omitted. Call the postprocess vfunc on the
36100 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36101 class ctor. Initialize m_in_call_function_usage. Call
36102 one_time_initialization.
36103 * rtl-tests.c (selftest::test_uncond_jump): Call
36104 set_new_first_and_last_insn.
36105 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36106 * selftest-rtl.c: New file.
36107 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36108 (selftest::get_insn_by_uid): New decl.
36109 * selftest-run-tests.c (selftest::run_tests): Call
36110 read_rtl_function_c_tests.
36111 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36112 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36115 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36117 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36118 operands in a special way. Assert that pos+len <= mode precision.
36120 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36122 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36123 3 argument Alias with unlimited for the negative form.
36124 (fno-vect-cost-model): Removed.
36126 2017-01-05 Martin Liska <mliska@suse.cz>
36128 * hsa-gen.c (gen_hsa_divmod): New function.
36129 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36131 2017-01-05 Martin Liska <mliska@suse.cz>
36134 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36137 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36139 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36140 small constant length operands.
36142 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36144 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36145 between loop iterations.
36147 2017-01-05 Martin Liska <mliska@suse.cz>
36150 * gimplify.c (gimplify_decl_expr): Compare to
36151 asan_poisoned_variables instread of checking flags.
36152 (gimplify_target_expr): Likewise.
36153 (gimplify_expr): Likewise.
36154 (gimplify_function_tree): Conditionally initialize
36155 asan_poisoned_variables.
36157 2017-01-04 Jeff Law <law@redhat.com>
36159 PR tree-optimizatin/78812
36160 * rtl.h (contains_mem_rtx_p): Prototype.
36161 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36162 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36163 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36164 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36165 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36167 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36169 * input.c (assert_char_at_range): Default-initialize actual_range.
36171 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36173 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36174 to match the caller.
36176 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36178 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36179 insns after final jump in test to emit dummy move.
36181 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36183 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36184 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36186 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36188 * multiple_target.c (create_dispatcher_calls): Init e_next.
36189 * tree-ssa-loop-split.c (split_loop): Init border.
36190 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36193 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36198 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36199 (altivec_register_operand): Do not return true if the operand
36200 contains a SUBREG mixing SImode and SFmode.
36201 (vsx_register_operand): Likewise.
36202 (vsx_reg_sfsubreg_ok): New predicate.
36203 (vfloat_operand): Do not return true if the operand contains a
36204 SUBREG mixing SImode and SFmode.
36205 (vint_operand): Likewise.
36206 (vlogical_operand): Likewise.
36207 (gpc_reg_operand): Likewise.
36208 (int_reg_operand): Likewise.
36209 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36210 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36211 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36213 (rs6000_emit_move_si_sf_subreg): New helper function.
36214 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36215 fixup SUBREGs involving SImode and SFmode.
36216 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36217 numbers for the new peephole2 optimization.
36218 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36219 the GLIBC math library that do AND/IOR/XOR operations on single
36220 precision floating point.
36221 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36222 target macros to say whether we need to avoid SUBREGs mixing
36224 (TARGET_ALLOW_SF_SUBREG): Likewise.
36225 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36226 (UNSPEC_SI_FROM_SF): Likewise.
36227 (iorxor): Change spacing.
36228 (and_ior_xor): New iterator for AND, IOR, and XOR.
36229 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36230 (movdi_from_sf_zero_ext): Likewise.
36231 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36232 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36233 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36234 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36235 (fms<mode>4): Likewise.
36236 (fnma<mode>4): Likewise.
36237 (fnms<mode>4): Likewise.
36238 (nfma<mode>4): Likewise.
36239 (nfms<mode>4): Likewise.
36241 2017-01-04 Marek Polacek <polacek@redhat.com>
36244 * doc/invoke.texi: Document -Wpointer-compare.
36246 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36248 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36251 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36252 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36255 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36258 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36259 documentation of the powerpc_popcntb_ok attribute.
36260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36261 code to issue warning messages if a requested CPU configuration is
36262 not supported by the binary (assembler and loader) toolchain.
36263 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36264 made to define a built-in function that has been disabled.
36265 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36266 made to define a built-in function that has been disabled.
36267 (altivec_init_builtins): Add comment explaining why definition
36268 of the DST built-in functions is not preceded by an assertion
36269 check. Add assertions to prevent ICE if attempts are made to
36270 define an altivec predicate or an abs* built-in function that has
36272 (htm_init_builtins): Add comment explaining why definition of the
36273 htm built-in functions is not preceded by an assertion check.
36275 2017-01-04 Jeff Law <law@redhat.com>
36277 PR tree-optimizatin/67955
36278 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36279 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36280 the points-to solution does not include pt_null. Use DECL_PT_UID
36283 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36285 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36286 Use gen_int_mode instead of gen_lopwart for const_int operands.
36288 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36290 PR tree-optimization/71563
36291 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36292 out of range value - has low bits known to be zero.
36294 2017-01-04 Alan Modra <amodra@gmail.com>
36296 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36297 * configure: Regenerate.
36298 * config.in: Regenerate.
36300 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36303 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36304 a substring of the message, but strcmp with the whole message. Ifdef
36305 ENABLE_NLS, translate the message first using dgettext.
36307 2017-01-03 Jeff Law <law@redhat.com>
36309 PR tree-optimizatin/78856
36310 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36311 (mark_threaded_blocks): Remove code to truncate thread paths that
36312 cross multiple loop headers. Instead invalidate the cached loop
36313 iteration information and handle case of a thread path walking
36314 into an irreducible region.
36316 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36319 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36320 assertions. Add support for doing the signbit if the IEEE 128-bit
36321 floating point value is in a GPR.
36322 * config/rs6000/rs6000.md (Fsignbit): Delete.
36323 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36324 Update the length attribute if the value is in a GPR.
36325 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36326 the sign or zero extension instruction, since the value is always 0/1.
36327 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36330 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36331 extracting SImode to a GPR register so that we can generate a
36332 store, limit the vector to be in a traditional Altivec register
36333 for the vextuwrx instruction.
36335 2017-01-03 Ian Lance Taylor <iant@google.com>
36337 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36339 2017-01-03 Martin Sebor <msebor@redhat.com>
36341 PR tree-optimization/78696
36342 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36343 precision. Use MPFR for %f for greater fidelity. Correct handling
36345 (pass_sprintf_length::compute_format_length): Set width and precision
36346 specified by asrerisk to void_node for vararg functions.
36347 (try_substitute_return_value): Adjust dump output.
36349 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36351 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36353 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36355 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36356 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36357 -mlra/-mno-lra was passed to the compiler.
36359 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36361 PR rtl-optimization/65618
36362 * emit-rtl.c (try_split): Move initialization of "before" and
36363 "after" to just before the call to emit_insn_after_setloc.
36365 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36367 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36369 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36371 * dwarf2out.c (gen_enumeration_type_die): When
36372 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36374 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36376 PR tree-optimization/78965
36377 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36378 Change first argument from const call_info & to call_info &. For %n
36379 set info.nowrite to false.
36381 PR middle-end/78901
36382 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36383 possibly throwing calls.
36385 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36386 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36387 and fns handling, rather than in a separate case SSA_NAME.
36389 2017-01-02 Jeff Law <law@redhat.com>
36391 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36392 fixes for first_period and second_period variables.
36394 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36397 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36398 (*insvqi_1): New insn pattern.
36399 (*insvqi_1_mem_rex64): Ditto.
36400 (*insvqi_2): Ditto.
36401 (*insvqi_3): Rename from *insvqi.
36403 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36405 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36407 * doc/cfg.texi (Edges): Remove reference to Java.
36408 (Maintaining the CFG): Ditto.
36410 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36412 PR middle-end/77674
36413 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36414 transparent aliases.
36416 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36418 PR middle-end/77484
36419 * predict.def (PRED_CALL): Update hitrate.
36420 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36421 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36422 into direct/indirect/polymorphic variants.
36424 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36426 Update copyright years.
36428 * gcc.c (process_command): Update copyright notice dates.
36429 * gcov-dump.c (print_version): Ditto.
36430 * gcov.c (print_version): Ditto.
36431 * gcov-tool.c (print_version): Ditto.
36432 * gengtype.c (create_file): Ditto.
36433 * doc/cpp.texi: Bump @copying's copyright year.
36434 * doc/cppinternals.texi: Ditto.
36435 * doc/gcc.texi: Ditto.
36436 * doc/gccint.texi: Ditto.
36437 * doc/gcov.texi: Ditto.
36438 * doc/install.texi: Ditto.
36439 * doc/invoke.texi: Ditto.
36441 Copyright (C) 2017 Free Software Foundation, Inc.
36443 Copying and distribution of this file, with or without modification,
36444 are permitted in any medium without royalty provided the copyright
36445 notice and this notice are preserved.