1 2017-11-22 Marek Polacek <polacek@redhat.com>
2 H.J. Lu <hongjiu.lu@intel.com>
3 Jason Merrill <jason@redhat.com>
8 * calls.c (initialize_argument_information): Call
9 warn_parameter_passing_abi target hook.
10 (store_one_arg): Use 0 for empty record size. Don't push 0 size
12 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
13 * common.opt: Update -fabi-version description.
14 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
15 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
17 (ix86_is_empty_record): New function.
18 (ix86_warn_parameter_passing_abi): New function.
19 (TARGET_EMPTY_RECORD_P): Redefine.
20 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
21 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
22 * doc/tm.texi: Regenerated.
23 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
24 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
25 * dwarf2out.c (get_ultimate_context): Move to tree.c.
26 * explow.c (hard_function_value): Call arg_int_size_in_bytes
27 instead of int_size_in_bytes.
28 * expr.c (copy_blkmode_to_reg): Likewise.
29 * function.c (aggregate_value_p): Return 0 for empty types.
30 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
31 (locate_and_pad_parm): Call arg size_in_bytes instead
33 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
34 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
35 * target.def (empty_record_p, warn_parameter_passing_abi): New target
37 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
38 (std_gimplify_va_arg_expr): Skip empty records. Call
39 arg_size_in_bytes instead size_in_bytes.
40 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
41 * tree-core.h (tree_type_common): Add empty_flag.
42 (tree_decl_common): Update comments.
43 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
45 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
46 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
48 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
49 * tree.c (default_is_empty_type): New function.
50 (default_is_empty_record): New function.
51 (arg_int_size_in_bytes): New function.
52 (arg_size_in_bytes): New function.
53 (get_ultimate_context): New function.
54 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
55 TRANSLATION_UNIT_WARN_EMPTY_P.
56 (default_is_empty_record, arg_int_size_in_bytes,
57 arg_size_in_bytes, get_ultimate_context): Declare.
59 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
61 * config/arm/arm.c (cmse_clear_registers): New function.
62 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
63 code by call to cmse_clear_registers.
64 (cmse_nonsecure_entry_clear_before_return): Likewise.
66 2017-11-22 Tamar Christina <tamar.christina@arm.com>
68 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
69 (vdot_s32, vdotq_s32): New.
70 (vdot_lane_u32, vdotq_lane_u32): New.
71 (vdot_lane_s32, vdotq_lane_s32): New.
74 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
77 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
78 for unsigned values with fewer HWIs than the precision.
79 (test_overflow): New function.
80 (wide_int_cc_tests): Call it.
82 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
84 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
85 has the same mode class as Pmode.
87 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
89 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
90 padding_bits_to_clear_ptr.
91 (cmse_nonsecure_entry_clear_before_return): Likewise.
93 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
95 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
96 auto_sbitap instead of integer bitfield to control register needing
99 2017-11-22 Jakub Jelinek <jakub@redhat.com>
101 PR tree-optimization/83044
102 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
103 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
104 that need the upper bound. Subtract offset from
105 get_addr_base_and_unit_offset only if positive and subtract it
106 before division by eltsize rather than after it.
109 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
110 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
111 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
115 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
116 expand_binop with *mul_widen_optab, make sure at least one of the
117 operands doesn't have VOIDmode.
120 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
122 PR rtl-optimization/82044
123 PR tree-optimization/82042
124 * dse.c (record_store): Check for overflow.
125 (check_mem_read_rtx): Properly check for overflow if width == -1, call
126 add_wild_read instead of clear_rhs_from_active_local_stores on
127 overflow and log it into dump_file.
129 2017-11-22 Richard Biener <rguenther@suse.de>
131 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
132 fake edges to exit when looking for a place to insert.
133 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
135 (insert_into_preds_of_block): Commit edge insertion immediately,
136 assert that doesn't require new BBs.
137 (fini_pre): Release expressions.
138 (pass_pre::execute): Shuffle things around a bit, if the fn
139 is too large do not compute AVAIL either as this is really the
142 2017-11-22 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/83089
145 * tree-if-conv.c (pass_if_conversion::execute): If anything
146 changed reset SCEV and free the number of iteration estimates.
148 2017-11-21 Martin Sebor <msebor@redhat.com>
150 PR tree-optimization/82945
151 * calls.h (warn_nonstring_bound): Remove unused function.
153 2017-11-21 Martin Sebor <msebor@redhat.com>
155 PR tree-optimization/82945
156 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
157 * calls.h (maybe_warn_nonstring_arg): Declare new function.
158 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
160 (initialize_argument_information): Call maybe_warn_nonstring_arg.
161 * calls.h (get_attr_nonstring_decl): Declare new function.
162 * doc/extend.texi (attribute nonstring): Update.
163 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
164 get_attr_nonstring_decl and handle it.
165 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
166 detection of nul-termination.
167 (strlen_to_stridx): Change to a pointer.
168 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
169 (pass_strlen::execute): Same.
171 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
173 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
174 for the new option -mprefer-vector-width=[none|128|256|512].
175 * config/i386/i386.c (ix86_target_string): remove old style options
176 -mprefer-avx256 and make -mprefer-avx128 as alias.
177 (ix86_option_override_internal): Apply defaults for the
178 -mprefer-vector-width=[128|256] option.
179 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
180 Implement macros to work with -mprefer-vector-width=.
181 * config/i386/i386.opt: Implemented option
182 -mprefer-vector-width=[none|128|256|512].
183 * doc/invoke.texi: Documentation for
184 -mprefer-vector-width=[none|128|256|512].
186 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
188 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
190 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
192 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
194 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
196 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
198 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
199 * expr.c (expand_expr_real_2): Likewise.
200 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
201 fold_binary_loc): Likewise.
202 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
203 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
204 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
205 MINUS_EXPR transformations.
206 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
207 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
208 * tree-inline.c (estimate_operator_cost): Likewise.
209 * tree-pretty-print.c (dump_generic_node, op_code_prio,
210 op_symbol_code): Likewise.
211 * tree-vect-stmts.c (vectorizable_operation): Likewise.
212 * vr-values.c (extract_range_from_binary_expr): Likewise.
213 * varasm.c (initializer_constant_valid_p_1): Likewise.
214 * tree.def: New tree code POINTER_DIFF_EXPR.
216 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
218 * config/i386/i386.md (*bswap<mode>2_movbe): Add
219 integer suffix to movbe mnemonic.
220 (*bswaphi2_movbe): Ditto.
221 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
223 2017-11-21 Jakub Jelinek <jakub@redhat.com>
226 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
227 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
228 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
230 2017-11-21 Martin Liska <mliska@suse.cz>
232 * tree-inline.c (expand_call_inline): Remove not needed
235 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
236 Jakub Jelinek <jakub@redhat.com>
239 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
240 Remove static keyword from f variable.
242 2017-11-21 Jakub Jelinek <jakub@redhat.com>
244 PR tree-optimization/83086
245 * gimple-ssa-store-merging.c
246 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
247 rather than n.base_addr.
249 2017-11-21 Martin Liska <mliska@suse.cz>
251 PR rtl-optimization/82044
252 PR tree-optimization/82042
253 * dse.c (check_mem_read_rtx): Check for overflow.
255 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
257 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
260 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
262 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
263 and its corresponding call arg location note.
265 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
266 Andrew Burgess <andrew.burgess@embecosm.com>
268 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
270 (arc_return_slot_offset): Likewise.
271 (arc_eh_return_address_location): New declaration.
272 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
273 (MUST_SAVE_REGISTER): Add exception handler case.
274 (MUST_SAVE_RETURN_ADDR): Likewise.
275 (arc_frame_pointer_required): Likewise.
276 (arc_frame_pointer_needed): New function.
277 (arc_compute_frame_size): Changed.
278 (arc_expand_prologue): Likewise.
279 (arc_expand_epilogue): Likewise.
280 (arc_initial_elimination_offset): Likewise.
281 (arc_return_slot_offset): Delete.
282 (arc_eh_return_address_location): New function.
283 (arc_builtin_setjmp_frame_value): Likewise.
284 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
285 (EH_RETURN_STACKADJ_RTX): Define.
286 (EH_RETURN_HANDLER_RTX): Likewise.
287 * config/arc/arc.md (eh_return): Delete.
289 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
291 * print-tree.h (debug_vec_tree): Remove prototype.
292 * gdbinit.in (pvt): Remove macro.
294 2017-11-21 Jakub Jelinek <jakub@redhat.com>
296 PR tree-optimization/83047
297 * gimple-ssa-store-merging.c
298 (imm_store_chain_info::output_merged_store): If the loads with the
299 same vuse are in different basic blocks, for load_gsi pick a load
300 location that is dominated by the other loads.
303 * config/i386/i386.c (ix86_memmodel_check): Start
304 -Winvalid-memory-model diagnostics with lowercase letter.
307 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
308 set in_cold_section_p to the partition containing loc_list->first.
309 When seeing loc_list->last_before_switch node, update secname and
310 perform range_across_switch second partition handling only after that.
313 * run-rtl-passes.c: Include debug.h.
314 (run_rtl_passes): Call debug_hooks->assembly_start.
315 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
319 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
320 OPTAB_DIRECT in calls to expand_simple_binop.
322 2017-11-20 David Malcolm <dmalcolm@redhat.com>
325 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
327 2017-11-20 David Malcolm <dmalcolm@redhat.com>
330 * spellcheck.h (best_match::blithely_get_best_candidate): New
333 2017-11-20 Jakub Jelinek <jakub@redhat.com>
335 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
336 with lower case letter.
338 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
340 * config/i386/i386.md (bswaphi2): New expander.
341 (*bswaphi2_movbe): New insn pattern.
342 (bswaphi -> rorhi pepehole2): New peephole pattern.
344 2017-11-20 Jeff Law <law@redhat.com>
346 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
347 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
348 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
349 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
350 gimple-ssa-evrp-analyze.h.
352 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
354 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
356 2017-11-20 Bin Cheng <bin.cheng@arm.com>
358 * tree-predcom.c: Add general comment on Store-Store chains.
359 (split_data_refs_to_components): Postpone clearing eliminate_store_p
361 (get_chain_last_ref_at): Rename into...
362 (get_chain_last_write_at): ...this.
363 (get_chain_last_write_before_load): New function.
364 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
365 CT_STORE_STORE when write reference is added.
366 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
367 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
368 (initialize_root_vars_store_elim_1): Ditto.
369 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
370 definition is created.
371 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
372 chain by replacing it with dominant stored value.
374 2017-11-20 Bin Cheng <bin.cheng@arm.com>
376 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
378 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
380 * vr-values.c (extract_range_from_binary_expr): Use a full range
383 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
385 * config/arm/arm.md (R4_REGNUM): Define constant.
386 (nonsecure_call_internal): Remove r4 clobber.
387 (nonsecure_call_value_internal): Likewise.
388 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
389 clobber and resequence match_operands.
390 (nonsecure_call_value_reg_thumb1_v5): Likewise.
391 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
392 (nonsecure_call_value_reg_thumb2): Likewise.
394 2017-11-20 Jakub Jelinek <jakub@redhat.com>
396 PR tree-optimization/78821
397 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
398 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
399 to the address of the base rather than the base itself.
400 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
401 (find_bswap_or_nop_finalize): New function.
402 (find_bswap_or_nop): Use it.
403 (bswap_replace): Return a tree rather than bool, change first
404 argument from gimple * to gimple_stmt_iterator, allow inserting
405 into an empty sequence, allow ins_stmt to be NULL - then emit
406 all stmts into gsi. Fix up MEM_REF address gimplification.
407 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
408 (struct store_immediate_info): Add N and INS_STMT non-static
410 (store_immediate_info::store_immediate_info): Initialize them
411 from newly added ctor args.
412 (merged_store_group::apply_stores): Formatting fixes. Sort by
414 (stmts_may_clobber_ref_p): For stores call also
415 refs_anti_dependent_p.
416 (gather_bswap_load_refs): New function.
417 (imm_store_chain_info::try_coalesce_bswap): New method.
418 (imm_store_chain_info::coalesce_immediate_stores): Use it.
419 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
420 (imm_store_chain_info::output_merged_store): Fail if number of
421 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
422 and NOP_EXPR rhs_code.
423 (pass_store_merging::process_store): Compute n and ins_stmt, if
424 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
425 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
427 (pass_store_merging::execute): Calculate dominators.
429 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
430 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
431 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
432 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
433 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
434 class pass_optimize_bswap, bswap_replace,
435 pass_optimize_bswap::execute): Moved to ...
436 * gimple-ssa-store-merging.c: ... this file.
437 Include optabs-tree.h.
438 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
439 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
440 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
441 anonymous namespace, remove static keywords.
442 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
443 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
445 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
448 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
450 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
452 * vec.h (debug_helper): New function.
453 (DEFINE_DEBUG_VEC): New macro.
454 * hash-set.h (debug_helper): New function.
455 (DEFINE_DEBUG_HASH_SET): New macro.
456 * cfg.c (debug_slim (edge)): New function.
457 Call DEFINE_DEBUG_VEC for edges.
458 Call DEFINE_DEBUG_HASH_SET for edges.
459 * cfghooks.c (debug_slim (basic_block)): New function.
460 Call DEFINE_DEBUG_VEC for basic blocks.
461 Call DEFINE_DEBUG_HASH_SET for basic blocks.
462 * print-tree.c (debug_slim): New function to handle trees.
463 Call DEFINE_DEBUG_VEC for trees.
464 Call DEFINE_DEBUG_HASH_SET for trees.
465 (debug (vec<tree, va_gc>) &): Remove.
466 (debug (<vec<tree, va_gc>) *): Remove.
467 * print-rtl.c (debug_slim): New function to handle const_rtx.
468 Call DEFINE_DEBUG_VEC for rtx_def.
469 Call DEFINE_DEBUG_VEC for rtx_insn.
470 Call DEFINE_DEBUG_HASH_SET for rtx_def.
471 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
472 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
473 (debug (vec<rtx_insn *> *ptr): Remove.
474 (debug_insn_vector): Remove.
475 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
477 2017-11-20 Tom de Vries <tom@codesourcery.com>
479 PR rtl-optimization/82020
480 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
481 IF_THEN_ELSE condition.
483 2017-11-19 Jeff Law <law@redhat.com>
485 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
486 of degenerates resulting from ignoring an edge.
488 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
491 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
493 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
496 * profile-count.c (profile_count::to_sreal_scale): Fix return value
497 for uninitialied counts.
499 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
502 * tree-inline.c (estimate_num_insns): Set to 1 at least.
504 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
507 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
509 2017-11-19 Tom de Vries <tom@codesourcery.com>
511 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
514 2017-11-19 Tom de Vries <tom@codesourcery.com>
516 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
519 2017-11-19 Tom de Vries <tom@codesourcery.com>
521 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
524 2017-11-19 Tom de Vries <tom@codesourcery.com>
526 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
528 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
530 2017-11-19 Tom de Vries <tom@codesourcery.com>
532 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
534 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
535 semicolon after MCORE_EXPORT_NAME call.
537 2017-11-19 Tom de Vries <tom@codesourcery.com>
540 * vmsdbgout.c (vmsdbgout_early_finish): New function.
541 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
543 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
545 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
546 (cgraph_edge::clone): Cleanup updating of profile.
547 * ipa-cp.c (update_profiling_info): Likewise.
548 * ipa-inline-transform.c (inline_transform): Likewise.
549 * ipa-inline.c (inline_small_functions): Add missing space to dump.
550 * ipa-split.c (execute_split_functions): Do not split when function
552 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
553 * profile-count.c (profile_count::dump): Add global0.
554 (profile_count::to_cgraph_frequency): Do not ICE when entry is
556 (profile_count::to_sreal_scale): Likewise.
557 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
558 (profile_count::combine_with_ipa_count): New function.
559 * profile-count.h (profile_guessed_global0adjusted): New.
560 (profile_count::adjusted_zero): New.
561 (profile_count::global0adjusted): New.
562 (profile_count::combine_with_ipa_count): New.
563 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
564 correct profile of return block of split functions.
565 (copy_cfg_body): Remove unused profile_count.
566 (copy_body): Likewise.
567 (expand_call_inline): Update.
568 (tree_function_versioning): Update.
570 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
572 * hash-set.h (hash_set::empty): New.
573 * tree-ssa-threadbackward.h: Delete.
574 * tree-ssa-threadbackward.c (class thread_jumps): New.
575 Move max_threaded_paths into class.
576 (fsm_find_thread_path): Remove arguments that are now in class.
577 (profitable_jump_thread_path): Rename to...
578 (thread_jumps::profitable_jump_thread_path): ...this.
579 (convert_and_register_jump_thread_path): Rename to...
580 (thread_jumps::convert_and_register_current_path): ...this.
581 (check_subpath_and_update_thread_path): Rename to...
582 (thread_jumps::check_subpath_and_update_thread_path): ...this.
583 (register_jump_thread_path_if_profitable): Rename to...
584 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
585 (handle_phi): Rename to...
586 (thread_jumps::handle_phi): ...this.
587 (handle_assignment): Rename to...
588 (thread_jumps::handle_assignment): ...this.
589 (fsm_find_control_statement_thread_paths): Rename to...
590 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
591 (find_jump_threads_backwards): Rename to...
592 (thread_jumps::find_jump_threads_backwards): ...this.
593 Initialize path local data.
594 (pass_thread_jumps::execute): Call find_jump_threads_backwards
595 from within thread_jumps class.
596 (pass_early_thread_jumps::execute): Same.
598 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
600 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
602 2017-11-17 Steve Ellcey <sellcey@cavium.com>
604 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
605 to canonical location.
607 2017-11-17 Steve Ellcey <sellcey@cavium.com>
610 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
612 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
614 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
616 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
617 rcrt1.o%s/grcrt1.o%s for -static-pie.
619 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
621 * i386.c (ix86_multiplication_cost, ix86_division_cost,
622 ix86_shift_rotate_cost): Break out from ...
623 (ix86_rtx_costs): ... here.
624 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
627 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
629 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
630 when entry block was promoted unlikely.
631 (estimate_bb_frequencies): Increase frequency scale.
632 * profile-count.h (profile_count): Export precision info.
634 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
636 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
637 disturbin profile of entry block.
639 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
641 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
644 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
646 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
647 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
649 2017-11-17 Jeff Law <law@redhat.com>
651 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
652 from evrp_dom_walker class. Various methods moved into new class.
653 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
654 (evrp_range_analyzer::enter): New method.
655 (evrp_range_analyzer::leave): New method.
656 (evrp_dom_walker): Remove delegators no longer needed by this class.
657 Replace vr_values data member with evrp_range_analyzer
659 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
660 method extracted from evrp_dom_walker::before_dom_children.
661 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
662 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
664 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
665 Add private copy constructor and move assignment operators.
666 Privatize methods and class data where trivially possible.
667 (evrp_dom_walker::cleanup): New function, extracted from
668 execute_early_vrp. Simplify access to class data.
670 * vr-values.h (get_output_for_vrp): Prototype.
671 * vr-values.c (get_output_for_vrp): New function extracted from
672 vrp_visit_assignment_or_call and extract_range_from_stmt.
673 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
675 2017-11-17 Luis Machado <luis.machado@linaro.org>
677 * config/aarch64/aarch64.c
678 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
679 (qdf24xx_tunings) <autoprefetcher_model>: Set to
680 tune_params::AUTOPREFETCHER_WEAK.
682 2017-11-17 Tamar Christina <tamar.christina@arm.com>
685 * config/arm/arm.c (arm_valid_target_attribute_rec):
686 Parse "arch=" and "+<ext>".
687 (arm_valid_target_attribute_tree): Re-init global options.
688 (arm_option_override): Make non-static.
689 (arm_options_perform_arch_sanity_checks): Make errors fatal.
690 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
691 (__ARM_FEATURE_CRC32): Support undef.
692 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
693 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
695 2017-11-17 David Malcolm <dmalcolm@redhat.com>
697 * gdbinit.in (break-on-diagnostic): New command.
699 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
701 * config/i386/i386.c (ix86_expand_epilogue): Change simple
702 return to indirect jump for EH return if control-flow
703 protection is enabled. Change explicit 'false' argument in
704 pro_epilogue_adjust_stack with a value of
706 * config/i386/i386.md (simple_return_indirect_internal):
707 Remove SImode restriction to support 64-bit.
709 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
711 * combine.c (added_notes_insn): New.
712 (try_combine): Handle added_notes_insn like added_links_insn.
713 Rewrite return value code.
714 (distribute_notes): Set added_notes_insn to the earliest insn we added
717 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
719 PR rtl-optimization/82621
720 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
721 dest of one of those SETs is unused.
723 2017-11-17 Richard Biener <rguenther@suse.de>
726 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
727 * tree-pretty-print.c (dump_generic_node): Handle
728 annot_expr_parallel_kind.
729 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
730 * gimplify.c (gimple_boolify): Likewise.
732 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
734 * config.gcc (extra_headers): Add cet.h for x86 targets.
735 * config/i386/cet.h: New file.
736 * doc/install.texi: Add --enable-cet/--disable-cet.
738 2017-11-17 Richard Biener <rguenther@suse.de>
740 PR tree-optimization/83017
741 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
742 (gen_parallel_loop): Properly count iterations.
743 (parallelize_loops): Handle loop->can_be_parallel independent
744 of flag_loop_parallelize_all. Make static profitability test match
746 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
747 * invoke.texi (parloops-min-per-thread): Document.
749 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
751 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
752 upstreaming review comments.
754 2017-11-17 Tamar Christina <tamar.christina@arm.com>
756 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
757 with fast unaligned access.
758 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
760 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
762 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
763 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
765 2017-11-17 Richard Biener <rguenther@suse.de>
767 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
768 folding of references.
770 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
773 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
774 of replacing call to strncmp with corresponding call to strcmp when
777 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
779 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
780 option prefer-avx256 for skylake-avx512 configuration.
781 * config/i386/i386.c (ix86_option_override_internal): Ditto.
782 (get_builtin_code_for_version): Ditto.
784 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
785 Monk Chiang <sh.chiang04@gmail.com>
787 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
788 (FIXED_REGISTERS): Reserve more register numbers.
789 (CALL_USED_REGISTERS): Likewise.
790 (REG_ALLOC_ORDER): Likewise.
791 (REG_CLASS_CONTENTS): Likewise.
792 (REGISTER_NAMES): Likewise.
794 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
795 Kito Cheng <kito.cheng@gmail.com>
797 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
799 * config/nds32/iterators.md: Add vector mode iterators and attributes.
801 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
803 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
804 parameter list for vec_splats.
806 2017-11-16 Joseph Myers <joseph@codesourcery.com>
808 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
810 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
812 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
815 * doc/install.texi: Document incompatibility of Perl >=5.6.26
816 with the required version of automake 1.11.6.
818 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
820 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
822 (DU_C2_3_power9): Correct reservation combinations.
823 (FP_DIV_power9, VEC_DIV_power9): New.
824 (power9-alu): Split out rotate/shift...
825 (power9-rot): ...to here, correct dispatch resource.
826 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
828 (power9-fp): Correct latency.
829 (power9-sdiv): Add div/sqrt resource.
830 (power9-ddiv): Correct latency, add div/sqrt resource.
831 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
832 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
834 (power9-qpdiv, power9-qpmul): Adjust resource usage.
836 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
838 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
839 switch statement mapping KF built-ins to TF built-ins if we don't
840 have the proper ISA 3.0 assembler support.
842 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
844 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
845 (lower_emutls_function_body): Do not compute it.
847 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
849 * ipa-split.c (split_bb_info): Turn time to sreal.
850 (split_point): Likewise.
851 (dump_split_point): Likewise.
852 (fine_split_points): Likewise.
853 (execute_split_functions): Only zero split_bbs; turn time to sreals.
855 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
857 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
860 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
862 * predict.c (combine_predictions_for_bb): Preserve zero predicted
864 (expensive_function_p): Remove useless assert.
865 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
867 2017-11-16 Martin Sebor <msebor@redhat.com>
869 PR tree-optimization/82588
870 PR tree-optimization/82583
871 * tree-vrp.c (check_array_ref): Handle flexible array members,
872 string literals, and inner indices.
873 (search_for_addr_array): Add detail to diagnostics.
875 2017-11-16 Nathan Sidwell <nathan@acm.org>
879 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
880 (SET_DECL_ASSEMBLER_NAME): Forward to
881 overwrite_decl_assembler_name.
882 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
883 (overwrite_decl_assembler_name): Declare.
884 * tree.c (overwrite_decl_assembler_name): New.
885 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
886 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
887 (LANG_HOOKS_INITIALIZER): Add it.
888 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
889 * langhooks.c (lhd_set_decl_assembler_name): Use
890 SET_DECL_ASSEMBLER_NAME.
891 (lhd_overwrite_decl_assembler_name): Default implementation.
893 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
894 Jackson Woodruff <jackson.woodruff@arm.com>
896 PR tree-optimization/71026
897 * match.pd: Canonicalize constant multiplies in division.
899 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
901 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
902 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
903 Remove OPT_fomit_frame_pointer entry.
904 * common/config/arc/arc-common.c: Likewise.
905 * common/config/arm/arm-common.c: Likewise.
906 * common/config/avr/avr-common.c: Likewise.
907 * common/config/c6x/c6x-common.c: Likewise.
908 * common/config/cr16/cr16-common.c: Likewise.
909 * common/config/cris/cris-common.c: Likewise.
910 * common/config/epiphany/epiphany-common.c: Likewise.
911 * common/config/fr30/fr30-common.c: Likewise.
912 * common/config/frv/frv-common.c: Likewise.
913 * common/config/ia64/ia64-common.c: Likewise.
914 * common/config/iq2000/iq2000-common.c: Likewise.
915 * common/config/lm32/lm32-common.c: Likewise.
916 * common/config/m32r/m32r-common.c: Likewise.
917 * common/config/mcore/mcore-common.c: Likewise.
918 * common/config/microblaze/microblaze-common.c: Likewise.
919 * common/config/mips/mips-common.c: Likewise.
920 * common/config/mmix/mmix-common.c: Likewise.
921 * common/config/mn10300/mn10300-common.c: Likewise.
922 * common/config/nios2/nios2-common.c: Likewise.
923 * common/config/pa/pa-common.c: Likewise.
924 * common/config/pdp11/pdp11-common.c: Likewise.
925 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
926 * common/config/riscv/riscv-common.c: Likewise.
927 * common/config/rs6000/rs6000-common.c: Likewise.
928 * common/config/rx/rx-common.c: Likewise.
929 * common/config/s390/s390-common.c: Likewise.
930 * common/config/sh/sh-common.c: Likewise.
931 * common/config/sparc/sparc-common.c: Likewise.
932 * common/config/tilegx/tilegx-common.c: Likewise.
933 * common/config/tilepro/tilepro-common.c: Likewise.
934 * common/config/v850/v850-common.c: Likewise.
935 * common/config/visium/visium-common.c: Likewise.
936 * common/config/xstormy16/xstormy16-common.c: Likewise.
937 * common/config/xtensa/xtensa-common.c: Likewise.
938 * invoke.texi (-fomit-frame-pointer): Update documentation.
940 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
942 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
944 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
946 * cfg.c (scale_bbs_frequencies_int,
947 cale_bbs_frequencies_gcov_type): Remove.
948 * cfg.h (scale_bbs_frequencies_int,
949 cale_bbs_frequencies_gcov_type): Remove.
951 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
953 * tree-ssa-loop-manip.c
954 (scale_dominated_blocks_in_loop): Update to profile counts.
955 (tree_transform_and_unroll_loop): Likewise.
957 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
959 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
960 scale_bbs_frequencies_int.
962 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
964 * final.c (compute_alignments): Use counts rather than frequencies.
966 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
968 * cfgloopanal.c: Include sreal.h
969 (average_num_loop_insns): Use counts and sreal for accounting.
971 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
973 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
976 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
977 Kito Cheng <kito.cheng@gmail.com>
979 * config/nds32/constraints.md: Provide more constraints.
980 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
981 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
982 support constraints usage.
984 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
986 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
988 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
989 Kito Cheng <kito.cheng@gmail.com>
991 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
992 * config/nds32/nds32.opt: Refine the layout.
993 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
994 TARGET_EXT_STRING): Support new options.
995 * config/nds32/nds32.h: Likewise.
996 * config/nds32/nds32.md: Likewise.
997 * config/nds32/nds32-predicates.c: Likewise.
998 * config/nds32/constraints.md: Likewise.
999 * common/config/nds32/nds32-common.c: Likewise.
1001 2017-11-16 Julia Koval <julia.koval@intel.com>
1004 * config/i386/gfniintrin.h: Add sse check.
1005 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1007 2017-11-16 Julia Koval <julia.koval@intel.com>
1009 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1010 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1011 (ix86_handle_option): Handle -mavx512vbmi2.
1012 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1013 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1014 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1015 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1016 (ix86_valid_target_attribute_inner_p): Ditto.
1017 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1018 * config/i386/i386.opt (mavx512vbmi2): New option.
1019 * doc/invoke.texi: Add new option.
1021 [2017-11-16 Julia Koval <julia.koval@intel.com>
1023 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1024 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1025 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1026 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1027 _mm512_gf2p8mul_epi8): New intrinsics.
1028 * config/i386/i386-builtin-types.def
1029 (V64QI_FTYPE_V64QI_V64QI): New type.
1030 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1031 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1032 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1033 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1034 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1035 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1037 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1039 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1040 explicitly as a stream of bytes.
1042 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1044 * config/rs6000/altivec.h (vec_xst_be): New #define.
1045 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1046 and externalize from *altivec_vperm_<mode>_internal.
1047 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1049 (XL_BE_V8HI): Likewise.
1050 (XL_BE_V4SI): Likewise.
1051 (XL_BE_V4SI): Likewise.
1052 (XL_BE_V2DI): Likewise.
1053 (XL_BE_V4SF): Likewise.
1054 (XL_BE_V2DF): Likewise.
1055 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1056 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1057 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1058 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1059 VSX_BUILTIN_VEC_XST_BE.
1060 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1061 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1063 (altivec_init_builtins): Replace conditional calls to def_builtin
1064 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1065 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1066 with unconditional calls. Remove calls to def_builtin for
1067 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1068 __builtin_vec_xst_be.
1069 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1070 to define_expand, and add alternate RTL generation for P8.
1071 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1072 vsx_ld_elemrev_v8hi.
1073 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1074 add alternate RTL generation for P8.
1075 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1076 vsx_ld_elemrev_v16qi.
1077 (vsx_st_elemrev_v8hi): Convert define_insn
1078 to define_expand, and add alternate RTL generation for P8.
1079 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1080 vsx_st_elemrev_v8hi.
1081 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1082 add alternate RTL generation for P8.
1083 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1084 vsx_st_elemrev_v16qi.
1086 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1089 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1090 TARGET_AVX512ER check.
1091 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1092 neither -mzeroupper nor -mno-zeroupper is used and
1093 TARGET_EMIT_VZEROUPPER is set.
1094 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1095 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1097 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1099 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1100 folding of vector compares.
1101 (fold_build_vec_cmp): New helper function.
1102 (fold_compare_helper): New helper function.
1103 (builtin_function_type): Add compare builtins to the list of functions
1104 having unsigned arguments. Cosmetic updates to comment indentation.
1105 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1106 the not+eq combination.
1108 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1110 PR tree-optimization/82726
1111 PR tree-optimization/70754
1112 * tree-predcom.c (order_drefs_by_pos): New function.
1113 (combine_chains): Move code setting has_max_use_after to...
1114 (try_combine_chains): ...here. New parameter. Sort combined chains
1115 according to position information.
1116 (tree_predictive_commoning_loop): Update call to above function.
1117 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1119 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1121 PR tree-optimization/82726
1123 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1125 PR tree-optimization/70754
1126 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1127 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1128 combined stmt before it if not NULL.
1129 (combine_chains): Process refs reversely and compute dominance point
1133 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1135 PR tree-optimization/79663
1136 * tree-predcom.c (combine_chains): Process refs in reverse order
1137 only for ZERO length chains, and add explaining comment.
1139 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1141 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1142 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1143 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1145 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1147 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1149 2017-11-15 Martin Liska <mliska@suse.cz>
1151 * tree-cfg.c (pass_warn_function_return::execute):
1152 Compare warn_return_type for greater than zero.
1154 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1158 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1159 to return true on Xeon and not on Xeon Phi.
1160 (ix86_check_avx256_register): Changed to ...
1161 (ix86_check_avx_upper_register): ... this. Add extra check for
1162 VALID_AVX512F_REG_OR_XI_MODE.
1163 (ix86_avx_u128_mode_needed): Changed
1164 ix86_check_avx256_register to ix86_check_avx_upper_register.
1165 (ix86_check_avx256_stores): Changed to ...
1166 (ix86_check_avx_upper_stores): ... this. Changed
1167 ix86_check_avx256_register to ix86_check_avx_upper_register.
1168 (ix86_avx_u128_mode_after): Changed
1169 avx_reg256_found to avx_upper_reg_found. Changed
1170 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1171 (ix86_avx_u128_mode_entry): Changed
1172 ix86_check_avx256_register to ix86_check_avx_upper_register.
1173 (ix86_avx_u128_mode_exit): Ditto.
1174 * config/i386/i386.h: (host_detect_local_cpu): New define.
1176 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1178 * config/arm/xgene1.md (xgene1): Split into automatons
1179 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1180 (xgene1_f_load): Adjust reservations and/or types.
1181 (xgene1_f_store): Likewise.
1182 (xgene1_load_pair): Likewise.
1183 (xgene1_store_pair): Likewise.
1184 (xgene1_fp_load1): Likewise.
1185 (xgene1_load1): Likewise.
1186 (xgene1_store1): Likewise.
1187 (xgene1_move): Likewise.
1188 (xgene1_alu): Likewise.
1189 (xgene1_simd): Likewise.
1190 (xgene1_bfm): Likewise.
1191 (xgene1_neon_load1): Likewise.
1192 (xgene1_neon_store1): Likewise.
1193 (xgene1_neon_logic): Likewise.
1194 (xgene1_neon_st1): Likewise.
1195 (xgene1_neon_ld1r): Likewise.
1196 (xgene1_alu_cond): Added.
1197 (xgene1_shift_reg): Likwise.
1198 (xgene1_bfx): Likewise.
1199 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1201 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1204 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1206 (can_widen_mult_without_libcall): New function.
1207 (expand_mul_overflow): If only checking unsigned mul overflow,
1208 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1209 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1210 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1211 (expand_DIVMOD): Formatting fix.
1212 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1213 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1214 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1216 PR tree-optimization/82977
1217 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1218 constructed temporary to strlen_to_stridx.put.
1220 2017-11-15 Martin Liska <mliska@suse.cz>
1222 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1223 * configure: Regenerate.
1225 2017-11-15 Martin Liska <mliska@suse.cz>
1228 * config/sh/sh-mem.cc: Use proper probability for
1231 2017-11-14 Jeff Law <law@redhat.com>
1233 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1234 the red zone for stack_clash_protection_final_dynamic_probe targets
1235 when the total dynamic stack size is zero bytes.
1237 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1238 blocks is post order.
1240 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1242 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1243 * final.c (rest_of_clean_state): Set it for the
1244 -fcompare-debug dump.
1245 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1246 class when TDF_COMPARE_DEBUG is set.
1248 * dwarf2out.c (gen_producer_string): Discard
1251 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1254 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1255 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1256 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1257 (__TGMATH_REAL_2_3): Remove macros.
1258 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1259 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1261 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1263 (remquo): Define using __TGMATH_REAL_3.
1265 2017-11-14 Jeff Law <law@redhat.com>
1267 * vr-values.c: New file with contents extracted from tree-vrp.c.
1268 * Makefile.in (OBJS): Add vr-values.o
1269 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
1270 (set_value_range, set_and_canonicalize_value_range): Likewise.
1271 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
1272 (value_range_constant_singleton, symbolic_range_p): Likewise.
1273 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
1274 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
1275 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
1276 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
1277 (find_case_label_range, find_case_label_index): Likewise.
1278 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
1279 (range_int_cst_singleton_p, value_inside_range): Likewise.
1280 (get_single_symbol): Likewise.
1281 (switch_update): Move structure definition here.
1282 (to_remove_edges, to_update_switch_stmts): Provide externs.
1283 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
1284 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
1285 (vrp_val_is_min, set_value_range): Likewise.
1286 (set_and_canonicalize_value_range, copy_value_range): Likewise.
1287 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
1288 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
1289 (range_is_nonnull, range_int_cst_p): Likewwise.
1290 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
1291 (get_single_symbol, operand_less_p): Likewise
1292 (compare_values_warnv, compare_values): Likewise.
1293 (value_inside_range, value_range_constant_singleton): Likewise.
1294 (zero_nonzero_bitgs_from_vr): Likewise.
1295 (extract_range_from_binary_expr_1): Likewise.
1296 (overflow_comparison_p): Likewise.
1297 (to_remove_edges, to_update_switch_stmts): Likewise.
1298 (find_case_label-index, find_case_label_range): Likewise.
1299 (switch_update, set_value_range_to_nonnegative): Remove.
1300 (set_value_range_to_truthvalue): Likewise.
1301 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
1302 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1303 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1304 (find_case_label_ranges, test_for_singularity): Likewise.
1305 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1306 (x_vr_values): Move to its remaining use site.
1308 2017-11-10 Jeff Law <law@redhat.com>
1310 * vr-values.h (VR_INITIALIZER): Move #define here.
1311 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1312 * Makefile.in (OBJS): Add tree-evrp.o
1313 * tree-vrp.h (assert_info): Move structure definition here.
1314 (set_value_range_to_varying): Prototype.
1315 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1316 (infer_value_range, register_edge_assert_for): Likewise.
1317 (stmt_interesting_for_vrp): Likewise.
1318 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1319 (set_value_range_to_varying): No longer static.
1320 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1321 (infer_value_range, register_edge_assert_for): Likewise.
1323 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1325 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1326 generate the XXBRD instruction.
1328 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1329 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1330 as being compatible if -mabi=ieeelongdouble.
1331 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1332 to setup float128 built-ins with hardware support.
1333 (BU_FLOAT128_HW_2): Likewise.
1334 (BU_FLOAT128_HW_3): Likewise.
1335 (BU_FLOAT128_HW_VSX_1): Likewise.
1336 (BU_FLOAT128_HW_VSX_2): Likewise.
1337 (scalar_extract_expq): Change float128 built-in functions to
1338 accommodate having both KFmode and TFmode functions. Use the
1339 KFmode variant as the default.
1340 (scalar_extract_sigq): Likewise.
1341 (scalar_test_neg_qp): Likewise.
1342 (scalar_insert_exp_q): Likewise.
1343 (scalar_insert_exp_qp): Likewise.
1344 (scalar_test_data_class_qp): Likewise.
1345 (sqrtf128_round_to_odd): Delete processing the round to odd
1346 built-in functions as special built-in functions, and define them
1347 as float128 built-ins. Use the KFmode variant as the default.
1348 (truncf128_round_to_odd): Likewise.
1349 (addf128_round_to_odd): Likewise.
1350 (subf128_round_to_odd): Likewise.
1351 (mulf128_round_to_odd): Likewise.
1352 (divf128_round_to_odd): Likewise.
1353 (fmaf128_round_to_odd): Likewise.
1354 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1355 support for KFmode and TFmode xststdcqp calls.
1356 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1357 point, switch the built-in handlers for the get/set float128
1358 exponent, get float128 mantissa, float128 test built-ins, and the
1359 float128 round to odd built-in functions. Eliminate creating the
1360 float128 round to odd built-in functions as special built-ins.
1361 (rs6000_init_builtins): Eliminate special creation of the float128
1362 round to odd built-in functions.
1363 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1364 function insns to support both TFmode and KFmode variants.
1365 (xsxsigqp_<mode>): Likewise.
1366 (xsiexpqpf_<mode>): Likewise.
1367 (xsiexpqp_<mode>): Likewise.
1368 (xststdcqp_<mode>): Likewise.
1369 (xststdcnegqp_<mode>): Likewise.
1370 (xststdcqp_<mode>): Likewise.
1372 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1374 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1375 unused path_in_freq_ptr parameter.
1376 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1378 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1380 * ipa-inline.c (edge_badness): Dump sreal frequency.
1381 (compute_inlined_call_time): Match natural implementaiton ...
1382 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1383 forgotten division by CGRAPH_FREQ_BASE.
1385 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1387 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1388 Solaris 11. Update comment.
1389 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1391 * config/sol2.h (STARTFILE_SPEC): Likewise.
1392 * configure: Regenerate.
1394 2017-11-14 Carl Love <cel@us.ibm.com>
1396 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1397 le_ and be_ prefixes to swap* variables. Remove
1398 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1401 2017-11-14 Jason Merrill <jason@redhat.com>
1403 Support GTY((cache)) on hash_map.
1404 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1405 (ggc_cache_remove): Override it instead of ggc_mx.
1406 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1407 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1408 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1409 (simple_cache_map_traits): Override maybe_mx.
1410 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1411 (hash_map): Friend gt_cleare_cache.
1412 (gt_cleare_cache): New.
1413 * tree.h (tree_cache_traits): New hash_map traits class.
1414 (tree_cache_map): New typedef.
1416 2017-11-14 Richard Biener <rguenther@suse.de>
1418 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1419 paramter and handling.
1420 (cleanup_control_flow_bb): Likewise.
1421 (cleanup_control_flow_pre): New helper performing a DFS walk
1422 to call cleanup_control_flow_bb in PRE order.
1423 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1424 via cleanup_control_flow_pre.
1426 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1428 * config/aarch64/aarch64-simd.md
1429 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1430 (*aarch64_simd_bsl<mode>_alt): Likewise.
1431 (aarch64_simd_bsldi_internal): New.
1432 (aarch64_simd_bsldi_alt): Likewise.
1434 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1436 * tracer.c (better_p): Do not compare frequencies.
1437 * reg-stack.c (better_edge): Likewise.
1438 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1441 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1443 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1444 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1445 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1446 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1447 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1448 * predict.c (maybe_hot_frequency_p): Inline to ...
1449 (maybe_hot_count_p): ... here; rewrite to counts.
1450 (counts_to_freqs): Rename to ...
1451 (update_max_bb_count): ... this one.
1452 (expensive_function_p): Use counts.
1453 (estimate_bb_frequencies): Update.
1454 (rebuild_frequencies): Update.
1455 * predict.h (counts_to_freqs): Rename to ...
1456 (update_max_bb_count): ... this one.
1457 * profile.c (compute_branch_probabilities): Add debug info
1458 * tree-inline.c (expand_call_inline): Update debug info.
1459 (optimize_inline_calls): Use update_max_bb_count..
1460 (tree_function_versioning): Use update_max_bb_count..
1461 * value-prof.c (gimple_value_profile_transformations):
1462 Do not use update_max_bb_count.
1464 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1466 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1467 always use frequencies.
1469 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1471 * bb-reorder.c: Remove frequencies from comments.
1472 (better_edge_p): Use profile counts.
1473 (find_traces): Dump profile counts.
1474 (rotate_loop): Use profile counts.
1475 (find_traces_1_round): Likewise.
1476 (connect_better_edge_p): Use counts instead of probabilities for
1478 (copy_bb_p): Drop early check for non-0 frequency.
1479 (sanitize_hot_paths): Update comments.
1481 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1483 * ipa-split.c (struct split_point): Add count.
1484 (consider_split): Do not compute incoming frequency; compute incoming
1485 count and store it to split_point.
1486 (split_function): Set count of the call to split part correctly.
1488 2017-11-13 Carl Love <cel@us.ibm.com>
1490 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1492 2017-11-13 Tom Tromey <tom@tromey.com>
1494 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1496 2017-11-13 Carl Love <cel@us.ibm.com>
1498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1499 Add support for builtins:
1500 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1501 vector {un,}signed {char,int,short},
1502 vector {un,}signed {char,int,short}) arguments.
1503 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1504 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1505 Add BU_P9V_AV_2 expansions for the builtins.
1506 * config/rs6000/altivec.h (vec_first_match_index,
1507 vec_first_mismatch_index, vec_first_match_or_eos_index,
1508 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1509 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1510 new extern declaration.
1511 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1512 * config/rs6000/vsx.md (first_match_index_<mode>,
1513 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1514 first_mismatch_or_eos_index_<mode>): Add define expand.
1515 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1516 * doc/extend.texi: Update the built-in documenation file for the new
1519 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1521 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1522 into the min/max operations for _Float<N> and _Float<N>X types.
1524 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1527 * dwarf2out.c (do_eh_frame): New static variable.
1528 (dwarf2out_begin_prologue): Set it.
1529 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1531 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1533 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1535 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1537 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1540 * debug.h (dwarf2out_do_eh_frame): Declare.
1541 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1542 (dwarf2out_do_frame): Use it.
1543 (dwarf2out_do_cfi_asm): Likewise.
1544 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1545 (dwarf2out_assembly_start): Likewise.
1546 (dwarf2out_begin_prologue): Fix comment.
1547 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1548 if the target needs either debug or unwind DWARF2 info.
1549 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1550 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1551 -ftrapping-math, -ftrapv and -fwrapv.
1553 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1555 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1556 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1557 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1558 (estimate_edge_size_and_time): Likewise.
1559 (ipa_merge_fn_summary_after_inlining): Likewise.
1560 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1561 (compute_uninlined_call_time): Use sreal_frequency.
1562 (compute_inlined_call_time): Likewise.
1563 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1564 * profile-count.c: Include sreal.h.
1565 (profile_count::to_sreal_scale): New.
1566 * profile-count.h: Forward declare sreal.
1567 (profile_count::to_sreal_scale): Declare.
1569 2017-11-13 Nathan Sidwell <nathan@acm.org>
1571 * diagnostic.c (maybe_line_and_column): New.
1572 (diagnostic_get_location_text): Use it.
1573 (diagnostic_report_current_module): Likewise.
1574 (test_diagnostic_get_location_text): Add tests.
1576 2017-11-13 Luis Machado <luis.machado@linaro.org>
1578 * doc/md.texi (Specifying processor pipeline description): Fix
1579 incorrect latency for the div instruction example.
1581 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1583 PR tree-optimization/78821
1584 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1585 that bit_not_p is the same.
1586 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1587 (split_group): Count precisely bit_not_p bits in each statement.
1588 (invert_op): New function.
1589 (imm_store_chain_info::output_merged_store): Use invert_op to
1590 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1591 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1593 2017-11-13 Martin Liska <mliska@suse.cz>
1595 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1596 (struct source_info): Likewise.
1597 (add_branch_counts): Likewise.
1598 (add_line_counts): Likewise.
1599 (function_summary): Likewise.
1600 (output_intermediate_line): Likewise.
1601 (generate_results): Likewise.
1603 2017-11-13 Martin Liska <mliska@suse.cz>
1605 * gcov.c (struct block_info): Remove typedef for block_t.
1606 (struct line_info): Likewise.
1607 (line_info::has_block): Likewise.
1608 (EXIT_BLOCK): Likewise.
1609 (unblock): Likewise.
1610 (circuit): Likewise.
1611 (get_cycles_count): Likewise.
1612 (process_file): Likewise.
1613 (read_graph_file): Likewise.
1614 (solve_flow_graph): Likewise.
1615 (find_exception_blocks): Likewise.
1616 (add_line_counts): Likewise.
1617 (accumulate_line_info): Likewise.
1618 (output_line_details): Likewise.
1620 2017-11-13 Martin Liska <mliska@suse.cz>
1622 * gcov.c (struct arc_info): Remove typedef for arc_t.
1623 (struct line_info): Likewise.
1624 (add_branch_counts): Likewise.
1625 (output_branch_count): Likewise.
1626 (function_info::~function_info): Likewise.
1627 (circuit): Likewise.
1628 (output_intermediate_line): Likewise.
1629 (read_graph_file): Likewise.
1630 (solve_flow_graph): Likewise.
1631 (find_exception_blocks): Likewise.
1632 (add_line_counts): Likewise.
1633 (accumulate_line_info): Likewise.
1634 (output_line_details): Likewise.
1635 (output_function_details): Likewise.
1637 2017-11-13 Martin Liska <mliska@suse.cz>
1639 * gcov.c (struct function_info): Remove typedef for function_t.
1640 (struct source_info): Likewise.
1641 (source_info::get_functions_at_location): Likewise.
1642 (solve_flow_graph): Likewise.
1643 (find_exception_blocks): Likewise.
1644 (add_line_counts): Likewise.
1645 (output_intermediate_file): Likewise.
1646 (process_file): Likewise.
1647 (generate_results): Likewise.
1648 (release_structures): Likewise.
1649 (read_graph_file): Likewise.
1650 (read_count_file): Likewise.
1651 (accumulate_line_counts): Likewise.
1652 (output_lines): Likewise.
1654 2017-11-13 Martin Liska <mliska@suse.cz>
1656 * gcov.c (function_info::function_info): Remove num_counts
1657 and add vector<gcov_type>.
1658 (function_info::~function_info): Use the vector.
1659 (process_file): Likewise.
1660 (read_graph_file): Likewise.
1661 (read_count_file): Likewise.
1662 (solve_flow_graph): Likewise.
1664 2017-11-13 Martin Liska <mliska@suse.cz>
1666 * gcov.c (function_info::is_artificial): New function.
1667 (process_file): Erase all artificial early.
1668 (generate_results): Skip as all artificial are already
1671 2017-11-13 Martin Liska <mliska@suse.cz>
1673 * gcov.c (read_graph_file): Store to global vector of functions.
1674 (read_count_file): Iterate the vector.
1675 (process_file): Likewise.
1676 (generate_results): Likewise.
1677 (release_structures): Likewise.
1679 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1681 PR tree-optimization/82954
1682 * gimple-ssa-store-merging.c
1683 (imm_store_chain_info::coalesce_immediate_stores): If
1684 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1686 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1688 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1689 Upddate call to ENDIAN_LANE_N.
1690 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1691 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1692 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1693 and use aarch64_endian_lane_rtx.
1694 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1696 2017-11-12 Tom de Vries <tom@codesourcery.com>
1698 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1700 2017-11-12 Tom de Vries <tom@codesourcery.com>
1702 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1704 2017-11-12 Tom de Vries <tom@codesourcery.com>
1706 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1708 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1709 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1710 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1712 2017-11-11 Martin Sebor <msebor@redhat.com>
1715 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1718 * prefic.c (translate_name): Replace strncpy with memcpy to
1719 avoid -Wstringop-truncation.
1721 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1723 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1725 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1727 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1729 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1731 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1734 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1736 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1737 enable generating XXBRH if the value is in a vector register.
1738 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1739 value is in a vector register.
1740 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1741 register to register bswap64's instead of doing the GPR sequence
1742 used on previous machines.
1743 (bswapdi2_xxbrd): New insn.
1744 (bswapdi2_reg): Disallow on ISA 3.0.
1745 (register to register bswap64 splitter): Do not split the insn on
1746 ISA 3.0 systems that use XXBRD.
1748 2017-11-10 Martin Sebor <msebor@redhat.com>
1751 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1753 (find_subframework_file): Same.
1755 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1757 * auto-profile.c (afdo_indirect_call): Drop frequency.
1758 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1759 (cgraph_node::create_edge): Drop frequency argument.
1760 (cgraph_node::create_indirect_edge): Drop frequency argument.
1761 (cgraph_edge::make_speculative): Drop frequency arguments.
1762 (cgraph_edge::resolve_speculation): Do not update frequencies
1763 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1764 (cgraph_node::dump): Check consistency in IPA mode.
1765 (cgraph_edge::maybe_hot_p): Use IPA counter.
1766 (cgraph_edge::verify_count_and_frequency): Rename to ...
1767 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1768 (cgraph_node::verify_node): Update.
1769 * cgraph.h (struct cgraph_edge): Drop frequency.
1770 (cgraph_edge::frequency): New function.
1771 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1773 (cgraph_edge::rebuild_edges): Likewise.
1774 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1775 (duplicate_thunk_for_node): Do not pass frequency.
1776 (cgraph_node::create_clone): Scale only counts.
1777 (cgraph_node::create_virtual_clone): Do not pass frequency.
1778 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1779 (cgraph_node::create_version_clone): Do not pass frequency.
1780 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1781 (cgraph_node::expand_thunk): Do not pass frequency.
1782 (cgraph_node::create_wrapper): Do not pass frequency.
1783 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1785 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1786 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1787 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1788 (gather_caller_stats): Use frequency function.
1789 (ipcp_cloning_candidate_p): Use frequency function.
1790 (ipcp_propagate_stage): Use frequency function.
1791 (get_info_about_necessary_edges): Use frequency function.
1792 (update_profiling_info): Update only IPA profile.
1793 (update_specialized_profile): Use frequency functoin.
1794 (perhaps_add_new_callers): Update only IPA profile.
1795 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1796 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1797 (dump_ipa_call_summary): Use frequency function.
1798 (estimate_edge_size_and_time): Use frequency function.
1799 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1800 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1801 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1802 (update_noncloned_counts): ... ths one; scale counts only.
1803 (clone_inlined_nodes): Do not scale frequency.
1804 (inline_call): Do not pass frequency.
1805 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1806 (compute_inlined_call_time): Use IPA profile.
1807 (want_inline_small_function_p): Use IPA profile.
1808 (want_inline_self_recursive_call_p): Use IPA profile.
1809 (edge_badness): Use IPA profile.
1810 (lookup_recursive_calls): Use IPA profile.
1811 (recursive_inlining): Do not pass frequency.
1812 (resolve_noninline_speculation): Do not update frequency.
1813 (inline_small_functions): Collect max of IPA profile.
1814 (dump_overall_stats): Dump IPA porfile.
1815 (dump_inline_stats): Dump IPA porfile.
1816 (ipa_inline): Collect IPA stats.
1817 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1818 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1819 (ipa_propagate_frequency): Use frequency function.
1820 (ipa_profile): Cleanup.
1821 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1822 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1823 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1824 (input_node): Do not stream frequency.
1825 (input_edge): Do not stream frequency.
1826 (merge_profile_summaries): Scale only IPA profiles.
1827 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1828 * predict.c (drop_profile): Do not recompute frequency.
1829 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1830 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1831 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1832 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1833 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1834 * tree-inline.c (copy_bb): Do not scale frequency.
1835 (expand_call_inline): Do not scale frequency.
1836 (tree_function_versioning): Do not scale frequency.
1837 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1839 2017-11-10 Julia Koval <julia.koval@intel.com>
1841 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1842 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1843 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1844 (_mm256_maskz_gf2p8affine_epi64_epi8)
1845 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1846 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1847 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1848 (__builtin_ia32_vgf2p8affineqb_v32qi)
1849 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1850 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1852 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1856 (arm_option_override): Refactor.
1857 (arm_option_reconfigure_globals): New.
1858 (arm_options_perform_arch_sanity_checks): New.
1859 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1861 (arm_options_perform_arch_sanity_checks): Likewise
1863 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1865 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1866 (power9-qpmul): New.
1867 * rs6000/rs6000.md ("type" attr): Add qmul.
1868 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1869 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1870 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1872 2017-11-10 Martin Sebor <msebor@redhat.com>
1875 * builtins.c (compute_objsize): Handle arrays that
1876 compute_builtin_object_size likes to fail for. Make extern.
1877 * builtins.h (compute_objsize): Declare.
1878 (check_strncpy_sizes): New function.
1879 (expand_builtin_strncpy): Call check_strncpy_sizes.
1880 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1881 -Wstringop-truncation.
1882 (gimple_fold_builtin_strncat): Same.
1883 * gimple.c (gimple_build_call_from_tree): Set call location.
1884 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1885 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1886 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1887 (handle_builtin_strlen): Use strlen_to_stridx.
1888 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1890 Use strlen_to_stridx.
1891 (pass_strlen::execute): Release strlen_to_stridx.
1892 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1893 (-Wstringop-truncation): Document new option.
1895 2017-11-10 Martin Liska <mliska@suse.cz>
1897 PR gcov-profile/82702
1898 * gcov.c (main): Handle intermediate files in a different
1900 (get_gcov_intermediate_filename): New function.
1901 (output_gcov_file): Remove support of intermediate files.
1902 (generate_results): Allocate intermediate file.
1903 (release_structures): Clean-up properly fn_end.
1904 (output_intermediate_file): Start iterating with line 1.
1906 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1908 PR tree-optimization/82929
1909 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1910 ops_swapped_p non-static data member.
1911 (store_immediate_info::store_immediate_info): Clear it.
1912 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1913 ops set ops_swapped_p.
1914 (count_multiple_uses): Handle ops_swapped_p.
1916 2017-11-10 Martin Liska <mliska@suse.cz>
1918 * coverage.c (coverage_init): Stream information about
1919 support of has_unexecuted_blocks.
1920 * doc/gcov.texi: Document that.
1921 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1922 * gcov.c (read_graph_file): Likewise.
1923 (output_line_beginning): Fix a small issue with
1926 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1928 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1929 reference of trivial component.
1931 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1934 * gimple-ssa-store-merging.c
1935 (pass_store_merging::terminate_all_aliasing_chains): For
1936 gimple_store_p stmts also call refs_output_dependent_p.
1938 PR rtl-optimization/82913
1939 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1942 2017-11-09 Jeff Law <law@redhat.com>
1944 * vr-values.h: New file with vr_values class.
1945 * tree-vrp.c: Include vr-values.h
1946 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1947 data objects into the vr_values class.
1948 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1949 (get_value_range): Make it a member function within vr_values class.
1950 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1951 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1952 (op_with_constant_singleton_value_range): Likewise.
1953 (extract_range_for_var_from_comparison_expr): Likewise.
1954 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1955 (extract_range_from_binary_expr): Likewise.
1956 (extract_range_from_unary_expr): Likewise.
1957 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1958 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1959 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1960 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1961 (compare_name_with_value, compare_names): Likewise.
1962 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1963 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1964 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1965 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1966 (extract_range_from_phi_node): Likewise.
1967 (simplify_truth_ops_using_ranges): Likewise.
1968 (simplify_div_or_mod_using_ranges): Likewise.
1969 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1970 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1971 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1972 (simplify_float_conversion_using_ranges): Likewise.
1973 (simplify_internal_call_using_ranges): Likewise.
1974 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1975 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1976 poitner via x_vr_values for calls into gimple folder.
1977 (vrp_initialize_lattice): Make this the vr_values ctor.
1978 (vrp_free_lattice): Make this the vr_values dtor.
1979 (set_vr_value): New function.
1980 (class vrp_prop): Add vr_values data member. Add various member
1981 functions as well as member functions that delegate to vr_values.
1982 (check_array_ref): Make a member function within vrp_prop class.
1983 (search_for_addr_array, vrp_initialize): Likewise.
1984 (vrp_finalize): Likewise. Revamp to avoid direct access to
1985 vr_value, values_propagated, etc.
1986 (check_array_bounds): Extract vrp_prop class instance pointer from
1987 walk info structure. Use it to call member functions.
1988 (check_all_array_refs): Make a member function within vrp_prop class.
1989 Smuggle class instance pointer via walk info structure.
1990 (x_vr_values): New local static.
1991 (vrp_valueize): Use x_vr_values to get class instance.
1992 (vr_valueize_1): Likewise.
1993 (class vrp_folder): Add vr_values data member. Add various member
1994 functions as well as member functions that delegate to vr_values.
1995 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1996 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1997 class instance from vr_values. Use it to call member functions.
1998 (vrp_dom_walker): Add vr_values data member.
1999 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2000 instance via x_vr_values.
2001 (identify_jump_threads): Accept vr_values as argument. Store
2002 it into the walker structure.
2003 (evrp_dom_walker): Add vr_values class data member. Add various
2005 (evrp_dom_walker::try_find_new_range): Use vr_values data
2006 member to access the memory allocator.
2007 (evrp_dom_walker::before_dom_children): Store vr_values class
2008 instance into the vrp_folder class.
2009 (evrp_dom_walker::push_value_range): Rework to avoid direct
2010 access to num_vr_values and vr_value.
2011 (evrp_dom_walker::pop_value_range): Likewise.
2012 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2013 Use vr_values to get to dump_all_value_ranges member function.
2014 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2015 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2016 Pass vr_values class instance down to identify_jump_threads.
2017 Remove call to vrp_free_lattice.
2018 (debug_all_value_ranges): Remove.
2020 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2021 (vrp_folder): Likewise.
2023 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2024 Get it from the existing bitmap instead.
2025 (vrp_intersect_ranges_1): Likewise.
2027 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2029 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2031 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2032 set bit_not_p to it.
2033 (imm_store_chain_info::coalesce_immediate_stores): Break group
2034 if bit_not_p is different.
2035 (count_multiple_uses, split_group,
2036 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2037 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2038 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2039 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2040 to store_immediate_info ctor.
2042 2017-11-09 Jim Wilson <jimw@sifive.com>
2044 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2045 (scan_prog_file): Likewise.
2047 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2049 * bb-reorder.c (max_entry_frequency): Remove.
2050 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2051 connect_traces, push_to_next_round_p): Remove prototypes.
2052 (find_traces_1_round): Use counts only.
2053 (push_to_next_round_p): Likewise.
2054 (find_traces): Likewise.
2055 (rotate_loop): Likewise.
2056 (find_traces_1_round): Likewise.
2057 (connect_traces): Likewise.
2058 (edge_order): Likewise.
2060 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2062 * config/arm/arm.c (output_return_instruction): Add comments to
2063 indicate requirement for cmse_nonsecure_entry return to account
2064 for the size of clearing instruction output here.
2065 (thumb_exit): Likewise.
2066 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2067 return in hardfloat mode.
2069 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2071 * config/rs6000/rs6000.c (machine_function): Add a bool,
2072 "toc_is_wrapped_separately".
2073 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2074 if it wasn't explicitly set or unset, we are optimizing for speed, and
2075 doing separate shrink-wrapping.
2076 (rs6000_get_separate_components): Enable the TOC component if
2077 saving the TOC register in the prologue.
2078 (rs6000_components_for_bb): Handle the TOC component.
2079 (rs6000_emit_prologue_components): Store the TOC register where needed.
2080 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2081 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2083 2017-11-09 Martin Jambor <mjambor@suse.cz>
2085 * ipa-param-manipulation.c: New file.
2086 * ipa-param-manipulation.h: Likewise.
2087 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2088 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2089 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2090 (ipa_parm_adjustment): Likewise.
2091 (ipa_parm_adjustment_vec): Likewise.
2092 (ipa_get_vector_of_formal_parms): Moved declaration to
2093 ipa-param-manipulation.h.
2094 (ipa_get_vector_of_formal_parm_types): Likewise.
2095 (ipa_modify_formal_parameters): Likewise.
2096 (ipa_modify_call_arguments): Likewise.
2097 (ipa_combine_adjustments): Likewise.
2098 (ipa_dump_param_adjustments): Likewise.
2099 (ipa_modify_expr): Likewise.
2100 (ipa_get_adjustment_candidate): Likewise.
2101 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2102 ipa-param-manipulation.c.
2103 (ipa_get_vector_of_formal_parm_types): Likewise.
2104 (ipa_modify_formal_parameters): Likewise.
2105 (ipa_modify_call_arguments): Likewise.
2106 (ipa_modify_expr): Likewise.
2107 (get_ssa_base_param): Likewise.
2108 (ipa_get_adjustment_candidate): Likewise.
2109 (index_in_adjustments_multiple_times_p): Likewise.
2110 (ipa_combine_adjustments): Likewise.
2111 (ipa_dump_param_adjustments): Likewise.
2112 * tree-sra.c: Also include ipa-param-manipulation.h
2113 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2116 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2117 Alan Hayward <alan.hayward@arm.com>
2118 David Sherwood <david.sherwood@arm.com>
2120 * doc/sourcebuild.texi (vect_masked_store): Document.
2122 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2123 Alan Hayward <alan.hayward@arm.com>
2124 David Sherwood <david.sherwood@arm.com>
2126 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2128 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2129 Alan Hayward <alan.hayward@arm.com>
2130 David Sherwood <david.sherwood@arm.com>
2132 * doc/sourcebuild.texi (vect_variable_length): Document.
2134 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2140 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2141 Alan Hayward <alan.hayward@arm.com>
2142 David Sherwood <david.sherwood@arm.com>
2144 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2146 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2147 Alan Hayward <alan.hayward@arm.com>
2148 David Sherwood <david.sherwood@arm.com>
2150 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2151 previously undocumented selectors.
2152 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2154 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2156 * doc/rtl.texi (const_vector): Say that elements can be
2157 const_wide_ints too.
2158 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2159 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2160 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2161 * optabs.c (expand_vector_broadcast): Likewise.
2163 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2164 Alan Hayward <alan.hayward@arm.com>
2165 David Sherwood <david.sherwood@arm.com>
2167 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2168 scaled index even if the unscaled address was invalid.
2169 Don't increase the complexity of using a scale in that case.
2171 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2172 Alan Hayward <alan.hayward@arm.com>
2173 David Sherwood <david.sherwood@arm.com>
2175 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2176 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2177 UNITS_PER_WORD bytes.
2178 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2179 based on REGMODE_NATURAL_SIZE of the inner mode.
2180 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2181 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2182 * expmed.c (lowpart_bit_field_p): Divide the value up into
2183 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2184 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2185 whether something is likely to occupy more than one register.
2187 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2190 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2191 profile_count::adjust_for_ipa_scaling.
2192 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2193 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2195 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2197 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2199 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2200 (split_group): Add total_orig and total_new arguments, estimate the
2201 number of statements related to the store group without store merging
2202 and with store merging.
2203 (imm_store_chain_info::output_merged_store): Adjust split_group
2204 callers, punt if estimated number of statements with store merging
2205 is not smaller than estimated number of statements without it.
2207 (handled_load): Remove has_single_use checks.
2208 (pass_store_merging::process_store): Likewise.
2210 2017-11-09 Richard Biener <rguenther@suse.de>
2212 PR tree-optimization/82902
2213 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2215 2017-11-09 Martin Liska <mliska@suse.cz>
2218 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2221 2017-11-09 Martin Liska <mliska@suse.cz>
2223 PR tree-optimization/82669
2224 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2226 2017-11-09 Martin Liska <mliska@suse.cz>
2228 PR gcov-profile/48463
2229 * coverage.c (coverage_begin_function): Output also end locus
2230 of a function and information whether the function is
2232 * gcov-dump.c (tag_function): Parse and print the information.
2233 * gcov.c (INCLUDE_MAP): Add include.
2234 (INCLUDE_SET): Likewise.
2235 (struct line_info): Move earlier in the source file because
2236 of vector<line_info> in function_info structure.
2237 (line_info::line_info): Likewise.
2238 (line_info::has_block): Likewise.
2239 (struct source_info): Add new member index.
2240 (source_info::get_functions_at_location): New function.
2241 (function_info::group_line_p): New function.
2242 (output_intermediate_line): New function.
2243 (output_intermediate_file): Use the mentioned function.
2244 (struct function_start): New.
2245 (struct function_start_pair_hash): Likewise.
2246 (process_file): Add code that identifies group functions.
2247 Assign lines either to global or function scope.
2248 (generate_results): Skip artificial functions.
2249 (find_source): Assign index for each source file.
2250 (read_graph_file): Read new flag artificial and end_line.
2251 (add_line_counts): Assign it either to global of function scope.
2252 (accumulate_line_counts): Isolate core of the function to
2253 accumulate_line_info and call it for both function and global
2255 (accumulate_line_info): New function.
2256 (output_line_beginning): Fix GNU coding style.
2257 (print_source_line): New function.
2258 (output_line_details): Likewise.
2259 (output_function_details): Likewise.
2260 (output_lines): Iterate both source (global) scope and function
2262 (struct function_line_start_cmp): New class.
2263 * doc/gcov.texi: Reflect changes in documentation.
2265 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2268 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
2269 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
2270 and similarly for not instead of neg.
2272 2017-11-08 Andi Kleen <ak@linux.intel.com>
2274 * config/i386/i386.opt: Add -mforce-indirect-call.
2275 * config/i386/predicates.md: Check for flag_force_indirect_call.
2276 * doc/invoke.texi: Document -mforce-indirect-call
2278 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
2280 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
2282 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
2283 config/riscv/riscv.c (predict.h): New include.
2284 (riscv_slow_unaligned_access_p): No longer static.
2285 (riscv_block_move_straight): Add require.
2286 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2288 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2291 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
2292 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
2293 nonimmediate_operand predicate for operand 1 instead of
2296 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2298 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
2300 * config/aarch64/constraints.md (Uml): New constraint.
2301 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2304 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2306 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2307 of two vec_duplicates into a vec_concat.
2309 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2311 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2312 Simplify vec_merge of vec_duplicate and vec_concat.
2313 * config/aarch64/constraints.md (Utq): New constraint.
2314 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2317 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2319 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2320 Simplify vec_merge of vec_duplicate and const_vector.
2321 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2323 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2324 mode iterator. Update predicate on operand 1 to
2325 handle non-const_vec constants. Delete constraints.
2326 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2328 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2330 PR tree-optimization/78821
2331 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2333 (store_operand_info::store_operand_info): Initialize it to false.
2334 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2335 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2336 store in the group, and if chain_info is non-NULL, to ignore altogether
2338 (compatible_load_p): Fail if bit_not_p does not match.
2339 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2340 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2341 (pass_store_merging::process_store): Adjust
2342 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2343 call terminate_all_aliasing_chains newly when adding a store into
2344 a chain with non-NULL chain_info.
2346 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2348 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2350 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2352 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2354 (aarch64_layout_frame): Initialise emit_frame_chain.
2355 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2356 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2358 2017-11-08 Martin Liska <mliska@suse.cz>
2360 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2361 of gimple_call_internal_p.
2363 2017-11-07 Tom de Vries <tom@codesourcery.com>
2365 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2367 2017-11-07 Tom de Vries <tom@codesourcery.com>
2369 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2372 2017-11-08 Martin Liska <mliska@suse.cz>
2375 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2377 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2379 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2381 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2384 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2385 (*cmp<mode>_ccz_1): New insn with $k alternative.
2388 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2389 RTX_COMM_COMPARE as commutative as well.
2390 (ix86_binary_operator_ok): Formatting fix.
2391 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2392 *<code><mode>3<mask_name><round_saeonly_name>,
2393 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2394 *<s>mul<mode>3_highpart<mask_name>,
2395 *vec_widen_umult_even_v16si<mask_name>,
2396 *vec_widen_umult_even_v8si<mask_name>,
2397 *vec_widen_umult_even_v4si<mask_name>,
2398 *vec_widen_smult_even_v16si<mask_name>,
2399 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2400 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2401 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2402 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2403 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2404 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2405 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2406 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2407 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2408 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2409 ix86_binary_operator_ok. Formatting fixes.
2410 (*<plusminus_insn><mode>3<mask_name><round_name>,
2411 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2414 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2416 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2417 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2418 eq and ne if TARGET_ISEL.
2419 (cmp): New code_iterator.
2420 (UNS, UNSU_, UNSIK): New code_attrs.
2421 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2422 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2424 ("eq<mode>3"): ... this.
2425 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2427 ("ne<mode>3"): ... this.
2429 2017-11-07 Julia Koval <julia.koval@intel.com>
2432 * common/config/i386/i386-common.c
2433 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2434 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2435 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2436 * config/i386/i386.opt: Ditto.
2437 * config/i386/i386.c (ix86_target_string): Ditto.
2438 (ix86_option_override_internal): Ditto.
2439 (ix86_init_mpx_builtins): Move MPX to args2.
2440 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2441 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2442 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2443 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2444 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2445 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2446 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2448 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2451 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2452 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2453 (zero-extendsidi peephole2): Remove peephole.
2455 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2458 * stor-layout.c: Include attribs.h.
2459 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2460 explicit lookup of "aligned" attribute.
2462 2017-11-07 Andrew Waterman <andrew@sifive.com>
2464 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2465 (riscv_expand_block_move): Likewise.
2466 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2468 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2469 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2470 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2471 (riscv_adjust_block_mem): Likewise.
2472 (riscv_block_move_loop): Likewise.
2473 (riscv_expand_block_move): Likewise.
2474 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2476 2017-11-07 Michael Clark <michaeljclark@mac.com>
2478 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2479 (MUSL_DYNAMIC_LINKER): Likewise.
2481 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2483 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2484 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2487 2017-11-07 Richard Biener <rguenther@suse.de>
2489 * match.pd: Fix build.
2491 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2492 Jackson Woodruff <jackson.woodruff@arm.com>
2494 PR tree-optimization/71026
2495 * match.pd: Canonicalize negate in division.
2497 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2500 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2502 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2504 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2505 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2507 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2509 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2510 non-scalar integral types.
2511 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2512 (-(A-B), -(~A)): New transformations.
2514 2017-11-07 Tom de Vries <tom@codesourcery.com>
2516 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2517 semicolon after "do {} while (0)".
2518 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2519 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2520 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2521 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2522 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2524 2017-11-07 Tom de Vries <tom@codesourcery.com>
2526 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2527 after "do {} while (0)".
2528 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2529 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2530 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2531 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2532 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2534 2017-11-07 Tom de Vries <tom@codesourcery.com>
2537 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2539 (arm_rtx_costs_internal): Add missing semicolon after
2540 HANDLE_NARROW_SHIFT_ARITH call.
2542 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2544 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2545 disable isel if it was not set explicitly.
2547 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2549 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2550 (add_type_attribute) likewise.
2552 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2554 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2555 of struct ix86_frame.
2556 (ix86_initial_elimination_offset): Likewise.
2557 (ix86_expand_split_stack_prologue): Likewise.
2559 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2561 * tree-vrp.h (enum value_range_type): Update stale comment.
2563 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2564 Alan Hayward <alan.hayward@arm.com>
2565 David Sherwood <david.sherwood@arm.com>
2567 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2568 (aarch64_expand_vec_perm_const): Take the number of units too.
2569 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2570 (aarch64_expand_vec_perm_const): Likewise.
2571 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2572 (vec_perm<mode>): Update accordingly.
2574 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2575 Alan Hayward <alan.hayward@arm.com>
2576 David Sherwood <david.sherwood@arm.com>
2578 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2579 Take the number of units too.
2580 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2581 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2582 but check for a vector mode before rather than after the call.
2583 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2584 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2585 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2586 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2587 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2588 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2589 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2590 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2591 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2592 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2593 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2594 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2595 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2596 (aarch64_sqdmull2_n<mode>): Update accordingly.
2598 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2599 Alan Hayward <alan.hayward@arm.com>
2600 David Sherwood <david.sherwood@arm.com>
2602 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2603 the number of units too.
2604 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2605 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2606 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2607 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2608 (vec_store_lanesxi<mode>): Update accordingly.
2610 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2611 Alan Hayward <alan.hayward@arm.com>
2612 David Sherwood <david.sherwood@arm.com>
2614 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2615 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2616 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2617 of units rather than the mode.
2618 * config/aarch64/iterators.md (nunits): New mode attribute.
2619 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2620 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2621 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2622 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2623 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2624 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2625 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2626 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2627 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2628 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2629 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2630 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2631 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2632 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2633 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2634 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2635 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2636 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2637 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2638 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2639 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2640 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2641 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2642 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2643 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2644 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2645 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2646 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2647 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2648 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2649 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2650 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2651 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2652 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2653 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2654 (aarch64_simd_vec_setv2di): Likewise.
2656 2017-11-06 Carl Love <cel@us.ibm.com>
2658 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2660 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2661 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2663 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2665 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2666 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2667 Add power 8 macro expansions.
2668 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2669 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2670 power 8 instructions. (VSX_XXBR): Add iterator.
2672 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2674 * config/arm/arm.md (predicable_short_it): Change default to "no",
2675 improve documentation, remove uses that are identical to the default.
2676 (enabled_for_depr_it): Rename to enabled_for_short_it.
2677 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2678 Remove default uses.
2679 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2680 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2681 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2682 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2684 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2687 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2688 float128 helper macros, which are no longer used after deleting
2689 the old 'q' built-in functions, and moving the round to odd
2690 built-in functions to being special built-in functions.
2691 (BU_FLOAT128_2): Likewise.
2692 (BU_FLOAT128_1_HW): Likewise.
2693 (BU_FLOAT128_2_HW): Likewise.
2694 (BU_FLOAT128_3_HW): Likewise.
2695 (FABSQ): Delete old 'q' built-in functions.
2696 (COPYSIGNQ): Likewise.
2697 (SQRTF128_ODD): Move round to odd built-in functions to be
2698 special built-in functions, so that we can handle
2699 -mabi=ieeelongdouble.
2700 (TRUNCF128_ODD): Likewise.
2701 (ADDF128_ODD): Likewise.
2702 (SUBF128_ODD): Likewise.
2703 (MULF128_ODD): Likewise.
2704 (DIVF128_ODD): Likewise.
2705 (FMAF128_ODD): Likewise.
2706 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2707 built-in names to 'f128'.
2708 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2709 old 'q' built-in functions, as the machine independent code for
2710 'f128' built-in functions handles this.
2711 (rs6000_expand_builtin): Add expansion for float128 round to odd
2712 functions, keying off on -mabi=ieeelongdouble of whether to use
2713 the KFmode or TFmode variant.
2714 (rs6000_init_builtins): Initialize the _Float128 round to odd
2716 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2717 _Float128 'q' built-in functions are now mapped into the new
2718 'f128' built-in functions.
2720 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2722 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2723 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2725 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2728 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2729 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2730 * toplev.c: Include "ipa-fnsummary.h".
2731 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2733 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2735 PR tree-optimization/82838
2736 * gimple-ssa-store-merging.c
2737 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2738 on a separate gimple_seq which is then appended to seq.
2740 2017-11-06 Jeff Law <law@redhat.com>
2743 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2744 (get_probe_interval): New functions.
2745 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2746 (ix86_adjust_stack_and_probe): Likewise.
2747 (output_adjust_stack_and_probe): Likewise.
2748 (ix86_emit_probe_stack_range): Likewise.
2749 (ix86_expand_prologue): Likewise.
2751 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2753 PR tree-optimization/82816
2754 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2755 if the modes of the two types are the same.
2756 (convert_plusminus_to_widen): Likewise.
2758 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2760 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2762 (usadv16qi): New define_expand.
2763 (usadv8hi): New define_expand.
2765 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2768 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2770 (inline_transform): Likewise.
2771 * predict.c (counts_to_freqs): Remove useless conditional.
2772 * profile-count.h (profile_count::apply_scale): Move sanity check.
2773 * tree-inline.c (copy_bb): Always scale.
2774 (copy_cfg_body): Likewise.
2776 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2779 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2781 (*cmp_ite0): Add enabled_for_depr_it attribute.
2782 (*cmp_ite1): Likewise.
2784 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2786 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2789 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2791 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2792 return the value by pointer.
2793 (extract_range_from_multiplicative_op_1): Update accordingly.
2794 Return as soon as an operation fails.
2796 2017-11-05 Tom de Vries <tom@codesourcery.com>
2799 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2800 (DEF_SANITIZER_BUILTIN): ... here.
2801 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2802 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2804 2017-11-05 Tom de Vries <tom@codesourcery.com>
2807 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2809 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2810 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2811 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2813 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2814 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2815 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2817 2017-11-05 Tom de Vries <tom@codesourcery.com>
2820 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2823 2017-11-04 Michael Clark <michaeljclark@mac.com>
2825 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2826 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2828 (*addsi3_extended): Likewise.
2829 (*addsi3_extended2): Likewise.
2830 (<optab>si3): Likewise.
2831 (<optab>di3): Likewise.
2832 (<optab><mode>3): Likewise.
2833 (<*optabe>si3_internal): Likewise.
2834 (zero_extendqi<SUPERQI:mode>2): Likewise.
2835 (*add<mode>hi3): Likewise.
2836 (*xor<mode>hi3): Likewise.
2837 (<optab>di3): Likewise.
2838 (*<optab>si3_extend): Likewise.
2839 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2840 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2841 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2843 2017-11-04 Andrew Waterman <andrew@sifive.com>
2845 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2846 TARGET_STRICT_ALIGN based upon -mtune argument.
2848 2017-11-04 Andrew Waterman <andrew@sifive.com>
2850 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2852 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2854 config/i386/i386.c (choose_basereg): Use optional scratch
2855 register and add assertion.
2856 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2857 needed, and don't allocate stack.
2858 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2859 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2860 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2862 2017-11-03 Jeff Law <law@redhat.com>
2864 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2865 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2866 to probe at the start of a noreturn function.
2868 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2870 PR tree-optimization/78821
2871 * gimple-ssa-store-merging.c: Update the file comment.
2872 (MAX_STORE_ALIAS_CHECKS): Define.
2873 (struct store_operand_info): New type.
2874 (store_operand_info::store_operand_info): New constructor.
2875 (struct store_immediate_info): Add rhs_code and ops data members.
2876 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2877 and op1r arguments to the ctor, initialize corresponding data members.
2878 (struct merged_store_group): Add load_align_base and load_align
2880 (merged_store_group::merged_store_group): Initialize them.
2881 (merged_store_group::do_merge): Update them.
2882 (merged_store_group::apply_stores): Pick the constant for
2883 encode_tree_to_bitpos from one of the two operands, or skip
2884 encode_tree_to_bitpos if neither operand is a constant.
2885 (class pass_store_merging): Add process_store method decl. Remove
2886 bool argument from terminate_all_aliasing_chains method decl.
2887 (pass_store_merging::terminate_all_aliasing_chains): Remove
2888 var_offset_p argument and corresponding handling.
2889 (stmts_may_clobber_ref_p): New function.
2890 (compatible_load_p): New function.
2891 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2892 if there is overlap and rhs_code is not INTEGER_CST. For
2893 non-overlapping stores terminate group if rhs is not mergeable.
2894 (get_alias_type_for_stmts): Change first argument from
2895 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2896 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2897 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2899 (get_location_for_stmts): Change first argument from
2900 auto_vec<gimple *> & to vec<gimple *> &.
2901 (struct split_store): Remove orig_stmts data member, add orig_stores.
2902 (split_store::split_store): Create orig_stores rather than orig_stmts.
2903 (find_constituent_stmts): Renamed to ...
2904 (find_constituent_stores): ... this. Change second argument from
2905 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2906 to info structures rather than the statements.
2907 (split_group): Rename ALLOW_UNALIGNED argument to
2908 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2909 it. Adjust find_constituent_stores caller.
2910 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2911 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2912 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2913 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2914 (mem_valid_for_store_merging): New function.
2915 (handled_load): New function.
2916 (pass_store_merging::process_store): New method.
2917 (pass_store_merging::execute): Use process_store method. Adjust
2918 terminate_all_aliasing_chains caller.
2920 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2922 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2923 Return true for more constants, symbols and label references.
2924 (aarch64_valid_floating_const): Remove unused function.
2926 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2929 * config/aarch64/aarch64.c (aarch64_layout_frame):
2930 Undo forcing of LR at bottom of frame.
2932 2017-11-03 Jeff Law <law@redhat.com>
2935 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2936 for int_registers_saved.
2938 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2939 extracted from tree-ssa-dom.c.
2940 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2941 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2942 (record_equivalences_from_incoming_edge): Add additional argument
2943 to single_pred_edge_ignoring_loop_edges call.
2944 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2945 (uncprop_dom_walker::before_dom_children): Add additional argument
2946 to single_pred_edge_ignoring_loop_edges call.
2947 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2948 single_pred_edge_ignoring_loop_edges rather than open coding.
2949 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2951 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2953 * match.pd (-(-A)): Rewrite.
2955 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2957 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2958 (rs6000_emit_int_cmove): New declaration.
2959 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2960 (rs6000_emit_sISEL): Delete.
2961 (rs6000_emit_int_cmove): Make non-static.
2962 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2963 instead of rs6000_emit_sISEL.
2965 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2967 * asan.c (create_cond_insert_point): Maintain profile.
2968 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2970 * basic-block.h (struct basic_block_def): Remove frequency.
2971 (EDGE_FREQUENCY): Use to_frequency
2972 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2974 (find_traces): Update to use to_frequency.
2975 (find_traces_1_round): Likewise; use only IPA counts.
2976 (bb_to_key): Likewise.
2977 (connect_traces): Use IPA counts only.
2978 (copy_bb_p): Update to use to_frequency.
2979 (fix_up_crossing_landing_pad): Likewise.
2980 (sanitize_hot_paths): Likewise.
2981 * bt-load.c (basic_block_freq): Likewise.
2982 * cfg.c (init_flow): Set count_max to uninitialized.
2983 (check_bb_profile): Remove frequencies; check counts.
2984 (dump_bb_info): Do not dump frequencies.
2985 (update_bb_profile_for_threading): Update counts only.
2986 (scale_bbs_frequencies_int): Likewise.
2987 (MAX_SAFE_MULTIPLIER): Remove.
2988 (scale_bbs_frequencies_gcov_type): Update counts only.
2989 (scale_bbs_frequencies_profile_count): Update counts only.
2990 (scale_bbs_frequencies): Update counts only.
2991 * cfg.h (struct control_flow_graph): Add count-max.
2992 (update_bb_profile_for_threading): Update prototype.
2993 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2994 (find_many_sub_basic_blocks): Likewise.
2995 * cfgcleanup.c (try_forward_edges): Likewise.
2996 (try_crossjump_to_edge): Likewise.
2997 * cfgexpand.c (expand_gimple_cond): Likewise.
2998 (expand_gimple_tailcall): Likewise.
2999 (construct_init_block): Likewise.
3000 (construct_exit_block): Likewise.
3001 * cfghooks.c (verify_flow_info): Check consistency of counts.
3002 (dump_bb_for_graph): Do not dump frequencies.
3003 (split_block_1): Do not update frequencies.
3004 (split_edge): Do not update frequencies.
3005 (make_forwarder_block): Do not update frequencies.
3006 (duplicate_block): Do not update frequencies.
3007 (account_profile_record): Do not update frequencies.
3008 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3009 for global heuristics.
3010 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3011 (expected_loop_iterations_unbounded): Use counts only.
3012 * cfgloopmanip.c (scale_loop_profile): Simplify.
3013 (create_empty_loop_on_edge): Simplify
3015 (duplicate_loop_to_header_edge): Simplify
3016 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3017 (update_br_prob_note): Take care of removing note when profile
3019 (relink_block_chain): Do not dump frequency.
3020 (rtl_account_profile_record): Use to_frequency.
3021 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3022 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3023 (cgraph_update_edges_for_call_stmt_node): Likewise.
3024 (cgraph_edge::verify_count_and_frequency): Update.
3025 (cgraph_node::verify_node): Temporarily disable frequency verification.
3026 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3027 to_cgraph_frequency.
3028 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3029 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3031 (cgraph_node::expand_thunk): Update profile.
3032 * except.c (dw2_build_landing_pads): Do not update frequency.
3033 * final.c (compute_alignments): Use to_frequency.
3034 (dump_basic_block_info): Do not dump frequency.
3035 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3036 (dump_gimple_bb_header): Do not dump frequency.
3037 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3039 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3040 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3041 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3042 (init_before_recovery): Do not update frequency.
3043 (sched_create_recovery_edges): Do not update frequency.
3044 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3045 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3046 (ipa_cp_c_finalize): Set max_count to uninitialized.
3047 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3048 (param_change_prob): Use counts.
3049 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3051 * ipa-split.c (consider_split): Use to_frequency.
3052 (split_function): Use to_frequency.
3053 * ira-build.c (loop_compare_func): Likewise.
3054 (mark_loops_for_removal): Likewise.
3055 (mark_all_loops_for_removal): Likewise.
3056 * loop-doloop.c (doloop_modify): Do not update frequency.
3057 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3059 * lto-streamer-in.c (input_function): Update count_max.
3060 * omp-expand.c (expand_omp_taskreg): Update count_max.
3061 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3062 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3063 (maybe_hot_count_p): Use ipa counts only.
3064 (maybe_hot_bb_p): Simplify.
3065 (maybe_hot_edge_p): Simplify.
3066 (probably_never_executed): Do not take frequency argument.
3067 (probably_never_executed_bb_p): Do not pass frequency.
3068 (probably_never_executed_edge_p): Likewise.
3069 (combine_predictions_for_bb): Check that profile is nonzero.
3070 (propagate_freq): Do not set frequency.
3071 (drop_profile): Simplify.
3072 (counts_to_freqs): Simplify.
3073 (expensive_function_p): Use to_frequency.
3074 (propagate_unlikely_bbs_forward): Simplify.
3075 (determine_unlikely_bbs): Simplify.
3076 (estimate_bb_frequencies): Add hack to silence graphite issues.
3077 (compute_function_frequency): Use ipa counts.
3078 (pass_profile::execute): Update.
3079 (rebuild_frequencies): Use counts only.
3080 (force_edge_cold): Use counts only.
3081 * profile-count.c (profile_count::dump): Dump new count types.
3082 (profile_count::differs_from_p): Check compatiblity.
3083 (profile_count::to_frequency): New function.
3084 (profile_count::to_cgraph_frequency): New function.
3085 * profile-count.h (struct function): Declare.
3086 (enum profile_quality): Add profile_guessed_local and
3087 profile_guessed_global0.
3088 (class profile_proability): Decrease number of bits to 29;
3089 update from_reg_br_prob_note and to_reg_br_prob_note.
3090 (class profile_count: Update comment; decrease number of bits
3091 to 61. Check compatibility.
3092 (profile_count::compatible_p): New private member function.
3093 (profile_count::ipa_p): New member function.
3094 (profile_count::operator<): Handle global zero correctly.
3095 (profile_count::operator>): Handle global zero correctly.
3096 (profile_count::operator<=): Handle global zero correctly.
3097 (profile_count::operator>=): Handle global zero correctly.
3098 (profile_count::nonzero_p): New member function.
3099 (profile_count::force_nonzero): New member function.
3100 (profile_count::max): New member function.
3101 (profile_count::apply_scale): Handle IPA scalling.
3102 (profile_count::guessed_local): New member function.
3103 (profile_count::global0): New member function.
3104 (profile_count::ipa): New member function.
3105 (profile_count::to_frequency): Declare.
3106 (profile_count::to_cgraph_frequency): Declare.
3107 * profile.c (OVERLAP_BASE): Delete.
3108 (compute_frequency_overlap): Delete.
3109 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3110 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3111 * sched-ebb.c (rank): Use counts only.
3112 * shrink-wrap.c (handle_simple_exit): Use counts only.
3113 (try_shrink_wrapping): Use counts only.
3114 (place_prologue_for_one_component): Use counts only.
3115 * tracer.c (find_best_predecessor): Use to_frequency.
3116 (find_trace): Use to_frequency.
3117 (tail_duplicate): Use to_frequency.
3118 * trans-mem.c (expand_transaction): Do not update frequency.
3119 * tree-call-cdce.c: Do not update frequency.
3120 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3121 (gimple_merge_blocks): Likewise.
3122 (gimple_split_edge): Likewise.
3123 (gimple_duplicate_sese_region): Likewise.
3124 (gimple_duplicate_sese_tail): Likewise.
3125 (move_sese_region_to_fn): Likewise.
3126 (gimple_account_profile_record): Likewise.
3127 (insert_cond_bb): Likewise.
3128 * tree-complex.c (expand_complex_div_wide): Likewise.
3129 * tree-eh.c (lower_resx): Update profile.
3130 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3132 (initialize_cfun): Do not initialize frequencies
3133 (freqs_to_counts): Delete.
3134 (copy_cfg_body): Ignore count parameter.
3135 (copy_body): Update.
3136 (expand_call_inline): Update count_max.
3137 (optimize_inline_calls): Update count_max.
3138 (tree_function_versioning): Update count_max.
3139 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3140 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3142 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3143 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3144 (try_peel_loop): Likewise.
3145 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3147 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3148 (tree_transform_and_unroll_loop): Do not use frequencies
3149 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3150 Use reliable prediction only.
3151 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3152 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3153 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3154 probability scaling.
3155 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3156 not update frequency
3157 (any_remaining_duplicated_blocks): Likewise.
3158 (update_profile): Likewise.
3159 (estimated_freqs_path): Delete.
3160 (freqs_to_counts_path): Delete.
3161 (clear_counts_path): Delete.
3162 (ssa_fix_duplicate_block_edges): Likewise.
3163 (duplicate_thread_path): Likewise.
3164 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3165 * tree-tailcall.c (decrease_profile): Do not update frequency.
3166 (eliminate_tail_call): Likewise.
3167 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3168 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3169 (optimize_mask_stores): Likewise.
3170 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3171 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3172 (ubsan_expand_ptr_ifn): Update profile.
3173 * value-prof.c (gimple_ic): Simplify.
3174 * value-prof.h (gimple_ic): Update prototype.
3175 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3176 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3178 (want_inline_self_recursive_call_p): Likewise.
3179 (resolve_noninline_speculation): Only cummulate defined counts.
3180 (inline_small_functions): Use nonzero_p.
3181 (ipa_inline): Do not access freed node.
3183 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3185 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3186 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3188 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3190 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3191 non-legitimate address.
3193 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3195 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3196 (*lt0_<mode>di, *lt0_<mode>si): New.
3198 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3200 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3201 TARGET_PAIRED_FLOAT.
3203 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3204 Jim Wilson <jim.wilson@linaro.org>
3206 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3207 * config/aarch64/aarch64-tune.md: Regenerated.
3208 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3209 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3211 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3213 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3214 (arc_expand_prologue): Restore blink for millicode.
3215 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3217 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3220 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3221 gen_vec_duplicate after forcing the scalar into a register.
3223 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3225 * combine (try_combine): Print the insns input to try_combine to the
3228 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3231 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3232 Remove second argument from aarch64_process_target_attr call.
3233 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3235 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3237 (aarch64_handle_attr_arch): Remove second argument.
3238 (aarch64_handle_attr_cpu): Ditto.
3239 (aarch64_handle_attr_tune): Ditto.
3240 (aarch64_handle_attr_isa_flags): Ditto.
3241 (aarch64_process_one_target_attr): Ditto.
3242 (aarch64_process_target_attr): Ditto.
3243 (aarch64_option_valid_attribute_p): Remove second argument.
3244 on aarch64_process_target_attr call.
3246 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3248 * diagnostic.c: Include "selftest-diagnostic.h".
3249 (selftest::assert_location_text): New function.
3250 (selftest::test_diagnostic_get_location_text): New function.
3251 (selftest::diagnostic_c_tests): Call it.
3253 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3255 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3256 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3257 (class selftest::test_diagnostic_context): Move to...
3258 * selftest-diagnostic.c: New file.
3259 * selftest-diagnostic.h: New file.
3261 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3263 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3264 offset range for FT32B.
3265 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3266 * config/ft32/ft32.md: Add TARGET_NOPM.
3267 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3268 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
3270 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
3272 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
3274 2017-11-02 Jeff Law <law@redhat.com>
3276 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
3277 virtual keyword on FINAL OVERRIDE members.
3279 * tree-ssa-propagate.h (ssa_propagation_engine): Group
3280 virtuals together. Add virtual destructor.
3281 (substitute_and_fold_engine): Similarly.
3283 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3285 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
3287 2017-11-02 Richard Biener <rguenther@suse.de>
3289 PR tree-optimization/82795
3290 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
3292 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3294 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
3296 (gcc_GAS_CHECK_FEATURE): Remove.
3297 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
3298 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
3299 * configure: Regenerate.
3301 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3303 * config/arc/arc.c (hwloop_optimize): Account for empty
3306 2017-11-02 Richard Biener <rguenther@suse.de>
3309 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3310 Truncate ARRAY_REF index and element size.
3312 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3314 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3316 2017-11-01 Jeff Law <law@redhat.com>
3318 * tree-ssa-ccp.c (ccp_folder): New class derived from
3319 substitute_and_fold_engine.
3320 (ccp_folder::get_value): New member function.
3321 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3322 (ccp_fold_stmt): Remove prototype.
3323 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3324 * tree-ssa-copy.c (copy_folder): New class derived from
3325 substitute_and_fold_engine.
3326 (copy_folder::get_value): Renamed from get_value.
3327 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3328 * tree-vrp.c (vrp_folder): New class derived from
3329 substitute_and_fold_engine.
3330 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3331 (vrp_folder::get_value): New member function.
3332 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3333 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3334 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3335 provide a class interface to folder/substitute routines.
3336 (ssa_prop_fold_stmt_fn): Remove typedef.
3337 (ssa_prop_get_value_fn): Likewise.
3338 (subsitute_and_fold): Remove prototype.
3339 (replace_uses_in): Likewise.
3340 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3341 Renamed from replace_uses_in. Call the virtual member function
3342 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3343 (substitute_and_fold_dom_walker): Remove initialization of
3344 data member entries for calbacks. Add substitute_and_fold_engine
3345 member and initialize it.
3346 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3347 member functions for get_value, replace_phi_args_in c
3348 replace_uses_in, and fold_stmt calls.
3349 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3350 substitute_and_fold. Remove assert. Update ctor call.
3352 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3353 (ssa_prop_visit_phi_fn): Likewise.
3354 (class ssa_propagation_engine): New class to provide an interface
3356 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3358 (ssa_prop_visit_phi): Likewise.
3359 (ssa_propagation_engine::simulate_stmt): Moved into class.
3360 Call visit_phi/visit_stmt from the class rather than via
3361 file scoped static variables.
3362 (ssa_propagation_engine::simulate_block): Moved into class.
3363 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3364 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3365 set file scoped statics for the visit_stmt/visit_phi callbacks.
3366 * tree-complex.c (complex_propagate): New class derived from
3367 ssa_propagation_engine.
3368 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3369 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3370 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3372 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3373 ssa_propagation_engine.
3374 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3375 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3376 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3377 * tree-ssa-copy.c (copy_prop): New class derived from
3378 ssa_propagation_engine.
3379 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3380 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3381 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3382 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3383 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3384 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3385 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3387 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3389 PR rtl-optimization/82778
3390 PR rtl-optimization/82597
3391 * compare-elim.c (struct comparison): Add in_a_setter field.
3392 (find_comparison_dom_walker::before_dom_children): Remove killed
3393 bitmap and df_simulate_find_defs call, instead walk the defs.
3394 Compute last_setter and initialize in_a_setter. Merge definitions
3395 with first initialization for a few variables.
3396 (try_validate_parallel): Use insn_invalid_p instead of
3397 recog_memoized. Return insn rather than just the pattern.
3398 (try_merge_compare): Fix up comment. Don't uselessly test if
3399 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3401 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3404 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3405 Alan Hayward <alan.hayward@arm.com>
3406 David Sherwood <david.sherwood@arm.com>
3408 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3409 aarch64_hard_regno_nregs to get the number of registers
3412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3413 Alan Hayward <alan.hayward@arm.com>
3414 David Sherwood <david.sherwood@arm.com>
3416 * config/aarch64/constraints.md (Upl): Rename to...
3418 * config/aarch64/aarch64.md
3419 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3422 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3423 Alan Hayward <alan.hayward@arm.com>
3424 David Sherwood <david.sherwood@arm.com>
3426 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3427 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3430 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3431 Alan Hayward <alan.hayward@arm.com>
3432 David Sherwood <david.sherwood@arm.com>
3434 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3435 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3436 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3438 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3439 of a single element.
3440 * config/aarch64/iterators.md: Add a comment above the permute
3441 unspecs to say that they are generated directly by
3442 aarch64_expand_vec_perm_const.
3443 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3445 2017-11-01 Nathan Sidwell <nathan@acm.org>
3447 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3449 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3451 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3452 and medany code models, and describe what they do.
3454 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3456 Revert accidental duplicate:
3458 * combine.c (can_change_dest_mode): Reject changes in
3459 REGMODE_NATURAL_SIZE.
3461 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3463 PR rtl-optimization/64682
3464 PR rtl-optimization/69567
3465 PR rtl-optimization/69737
3466 PR rtl-optimization/82683
3467 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3468 register mentioned in the note, drop the note, unless it came from I3,
3469 in which case it should go to I3 again.
3471 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3473 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3474 and return false if not.
3475 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3477 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3479 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3480 range is known to be empty.
3482 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3483 Alan Hayward <alan.hayward@arm.com>
3484 David Sherwood <david.sherwood@arm.com>
3486 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3487 and CONST_VECTOR_NUNITS instead of computing the number of units from
3488 the byte sizes of the vector and element.
3489 (simplify_binary_operation_1): Likewise.
3490 (simplify_const_binary_operation): Likewise.
3491 (simplify_ternary_operation): Likewise.
3493 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3494 Alan Hayward <alan.hayward@arm.com>
3495 David Sherwood <david.sherwood@arm.com>
3497 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3498 (int_mem_offset): ...this new function.
3499 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3500 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3501 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3504 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3505 Alan Hayward <alan.hayward@arm.com>
3506 David Sherwood <david.sherwood@arm.com>
3508 * lower-subreg.c (interesting_mode_p): New function.
3509 (compute_costs, find_decomposable_subregs, decompose_register)
3510 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3511 (resolve_clobber, dump_choices): Use it.
3513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3514 Alan Hayward <alan.hayward@arm.com>
3515 David Sherwood <david.sherwood@arm.com>
3517 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3519 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3520 Alan Hayward <alan.hayward@arm.com>
3521 David Sherwood <david.sherwood@arm.com>
3523 * alias.c (find_base_value, find_base_term): Only process integer
3524 truncations. Check the precision rather than the size.
3526 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3527 Alan Hayward <alan.hayward@arm.com>
3528 David Sherwood <david.sherwood@arm.com>
3530 * machmode.h (is_narrower_int_mode): New function
3531 * optabs.c (expand_float, expand_fix): Use it.
3532 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3534 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3535 Alan Hayward <alan.hayward@arm.com>
3536 David Sherwood <david.sherwood@arm.com>
3538 * rtl.h (narrower_subreg_mode): New function.
3539 * ira-color.c (update_costs_from_allocno): Use it.
3541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3542 Alan Hayward <alan.hayward@arm.com>
3543 David Sherwood <david.sherwood@arm.com>
3545 * optabs-query.h (convert_optab_p): New function, split out from...
3546 (convert_optab_handler): ...here.
3547 (widening_optab_handler): Delete.
3548 (find_widening_optab_handler): Remove permit_non_widening parameter.
3549 (find_widening_optab_handler_and_mode): Likewise. Provide an
3550 override that operates on mode class wrappers.
3551 * optabs-query.c (widening_optab_handler): Delete.
3552 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3553 parameter. Assert that the two modes are the same class and that
3554 the "from" mode is narrower than the "to" mode. Use
3555 convert_optab_handler instead of widening_optab_handler.
3556 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3557 instead of widening_optab_handler.
3558 * expr.c (expand_expr_real_2): Update calls to
3559 find_widening_optab_handler.
3560 * optabs.c (expand_widen_pattern_expr): Likewise.
3561 (expand_binop_directly): Take the insn_code as a parameter.
3562 (expand_binop): Only call find_widening_optab_handler for
3563 conversion optabs; use optab_handler otherwise. Update calls
3564 to find_widening_optab_handler and expand_binop_directly.
3565 Use convert_optab_handler instead of widening_optab_handler.
3566 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3567 find_widening_optab_handler and use scalar_mode rather than
3569 (convert_plusminus_to_widen): Likewise.
3571 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3572 Alan Hayward <alan.hayward@arm.com>
3573 David Sherwood <david.sherwood@arm.com>
3575 * machmode.h (fixed_size_mode): New class.
3576 * rtl.h (get_pool_mode): Return fixed_size_mode.
3577 * gengtype.c (main): Add fixed_size_mode.
3578 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3579 (get_raw_arg_mode): Likewise.
3580 * doc/tm.texi: Regenerate.
3581 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3582 * targhooks.c (default_get_reg_raw_mode): Likewise.
3583 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3584 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3585 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3586 (msp430_get_raw_result_mode): Likewise.
3587 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3588 * dbxout.c (dbxout_parms): Require fixed-size modes.
3589 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3590 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3591 * omp-low.c (lower_oacc_reductions): Likewise.
3592 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3593 (simplify_subreg): Update accordingly.
3594 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3595 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3596 that aren't fixed-size.
3597 (get_pool_mode): Return a fixed_size_mode.
3598 (output_constant_pool_2): Take a fixed_size_mode.
3600 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3601 Alan Hayward <alan.hayward@arm.com>
3602 David Sherwood <david.sherwood@arm.com>
3604 * doc/rtl.texi (vec_series): Document.
3605 (const): Say that the operand can be a vec_series.
3606 * rtl.def (VEC_SERIES): New rtx code.
3607 * rtl.h (const_vec_series_p_1): Declare.
3608 (const_vec_series_p): New function.
3609 * emit-rtl.h (gen_const_vec_series): Declare.
3610 (gen_vec_series): Likewise.
3611 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3612 (gen_vec_series): Likewise.
3613 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3614 * simplify-rtx.c (simplify_unary_operation): Handle negations
3616 (simplify_binary_operation_series): New function.
3617 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3618 (test_vector_ops_series): New function.
3619 (test_vector_ops): Call it.
3620 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3621 gen_const_vec_series.
3622 (altivec_lvsr): Likewise.
3623 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3625 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3626 Alan Hayward <alan.hayward@arm.com>
3627 David Sherwood <david.sherwood@arm.com>
3629 * doc/rtl.texi (const): Update description of address constants.
3630 Say that vector constants are allowed too.
3631 * common.md (E, F): Use CONSTANT_P instead of checking for
3633 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3634 checking for CONST_VECTOR.
3635 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3637 * expr.c (expand_expr_real_2): Check for vector modes instead
3638 of checking for CONST_VECTOR.
3639 * rtl.h (const_vec_p): New function.
3640 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3641 (unwrap_const_vec_duplicate): Handle them here too.
3643 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3644 David Malcolm <dmalcolm@redhat.com>
3645 Alan Hayward <alan.hayward@arm.com>
3646 David Sherwood <david.sherwood@arm.com>
3648 * rtl.h (vec_duplicate_p): New function.
3649 * selftest-rtl.c (assert_rtx_eq_at): New function.
3650 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3651 (assert_rtx_eq_at): Declare.
3652 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3653 * selftest-run-tests.c (selftest::run_tests): Call it.
3654 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3655 (simplify_unary_operation_1): Recursively handle vector duplicates.
3656 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3658 (simplify_subreg): Handle subregs of vector duplicates.
3659 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3660 (selftest::simplify_rtx_c_tests): New functions.
3662 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3663 Alan Hayward <alan.hayward@arm.com>
3664 David Sherwood <david.sherwood@arm.com>
3666 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3667 (gen_vec_duplicate): Likewise.
3668 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3670 (gen_const_vector): ...here.
3671 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3672 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3673 whose elements are all equal.
3674 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3675 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3676 (simplify_relational_operation): Likewise.
3677 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3679 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3680 (aarch64_expand_vector_init): Likewise.
3681 * config/arm/arm.c (neon_vdup_constant): Likewise.
3682 (neon_expand_vector_init): Likewise.
3683 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3684 (arm_block_set_unaligned_vect): Likewise.
3685 (arm_block_set_aligned_vect): Likewise.
3686 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3687 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3688 (expand_vec_perm_even_odd_pack): Likewise.
3689 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3690 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3691 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3692 gen_const_vec_duplicate.
3693 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3694 * config/mips/mips.c (mips_gen_const_int_vector): Use
3695 gen_const_vec_duplicate.
3696 (mips_expand_vector_init): Use CONST0_RTX.
3697 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3698 (define_split): Use gen_const_vec_duplicate.
3699 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3700 (define_split): Use gen_const_vec_duplicate.
3701 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3702 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3703 * config/spu/spu.c (spu_const): Likewise.
3705 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3706 Alan Hayward <alan.hayward@arm.com>
3707 David Sherwood <david.sherwood@arm.com>
3709 * combine.c (can_change_dest_mode): Reject changes in
3710 REGMODE_NATURAL_SIZE.
3712 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3714 * configure.ac (--enable-libssp): New.
3715 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3716 trying to determine target-specific default. Adjust indentation.
3717 * configure: Regenerated.
3718 * doc/install.texi (Configuration): Expand --disable-libssp
3721 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3723 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3726 2017-10-31 Martin Jambor <mjambor@suse.cz>
3729 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3731 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3733 * auto-profile.c (autofdo_source_profile::read): Use
3734 UNKNOWN_LOCATION rather than 0.
3735 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3736 (warning_at): ...this overload.
3737 (warning_at_rich_loc_n): Rename to...
3738 (warning_n): ...this overload.
3739 (error_at_rich_loc): Rename to...
3740 (error_at): ...this overload.
3741 (pedwarn_at_rich_loc): Rename to...
3742 (pedwarn): ...this overload.
3743 (permerror_at_rich_loc): Rename to...
3744 (permerror): ...this overload.
3745 (inform_at_rich_loc): Rename to...
3746 (inform): ...this overload.
3747 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3748 (diagnostic_n_impl_richloc): Rename to...
3749 (diagnostic_n_impl): ...this rich_location *-based decl.
3750 (inform_at_rich_loc): Rename to...
3751 (inform): ...this, and add an assertion.
3752 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3753 (warning_at_rich_loc): Rename to...
3754 (warning_at): ...this, and add an assertion.
3755 (warning_at_rich_loc_n): Rename to...
3756 (warning_n): ...this, and add an assertion.
3757 (warning_n): Update location_t-based implementation for removal of
3758 location_t-based diagnostic_n_impl.
3759 (pedwarn_at_rich_loc): Rename to...
3760 (pedwarn): ...this, and add an assertion.
3761 (permerror_at_rich_loc): Rename to...
3762 (permerror): ...this, and add an assertion.
3763 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3764 (error_at_rich_loc): Rename to...
3765 (error_at): ...this, and add an assertion.
3766 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3767 (driver::do_spec_on_infiles): Likewise.
3768 * substring-locations.c (format_warning_va): Update for renaming
3769 of inform_at_rich_loc.
3771 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3773 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3774 _Float<N>X built-in functions so that the variant without the
3775 "__builtin_" prefix is only enabled for the GNU C and Objective C
3776 languages when they are in non-strict ANSI/ISO mode.
3777 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3778 * target.def (floatn_builtin_p): Add a target hook to control
3779 whether _Float<N> and _Float<N>X built-in functions without the
3780 "__builtin_" prefix are enabled, and return true for C and
3781 Objective C in the default hook. Include langhooks.h in
3783 * targhooks.h (default_floatn_builtin_p): Likewise.
3784 * targhooks.c (default_floatn_builtin_p): Likewise.
3785 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3786 floatn_builtin_p target hook.
3787 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3789 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3790 Eric Botcazou <ebotcazou@adacore.com>
3792 PR rtl-optimization/81803
3793 * lra-constraints.c (curr_insn_transform): Also reload the whole
3794 register for a strict subreg no wider than a word if this is for
3795 a WORD_REGISTER_OPERATIONS target.
3797 2017-10-31 Jason Merrill <jason@redhat.com>
3799 * gdbinit.in: Skip over inlines from timevar.h.
3801 2017-10-31 Martin Liska <mliska@suse.cz>
3803 * doc/gcov.texi: Document new option.
3804 * gcov.c (print_usage): Likewise print it.
3805 (process_args): Support the argument.
3806 (format_count): New function.
3807 (format_gcov): Use the function.
3809 2017-10-31 Martin Liska <mliska@suse.cz>
3811 * gcov.c (struct name_map): do not use typedef.
3812 Define operator== and operator<.
3813 (name_search): Remove.
3814 (name_sort): Remove.
3815 (main): Do not allocate names.
3816 (process_file): Add vertical space.
3817 (generate_results): Use std::find.
3818 (release_structures): Do not release memory.
3819 (find_source): Use std::find.
3821 2017-10-31 Martin Liska <mliska@suse.cz>
3823 * gcov.c (struct line_info): Remove it's typedef.
3824 (line_info::line_info): Add proper ctor.
3825 (line_info::has_block): Do not use a typedef.
3826 (struct source_info): Do not use typedef.
3827 (circuit): Likewise.
3828 (get_cycles_count): Likewise.
3829 (output_intermediate_file): Iterate via vector iterator.
3830 (add_line_counts): Use std::vector methods.
3831 (accumulate_line_counts): Likewise.
3832 (output_lines): Likewise.
3834 2017-10-31 Martin Liska <mliska@suse.cz>
3836 * gcov.c (struct source_info): Remove typedef.
3837 (source_info::source_info): Add proper ctor.
3838 (accumulate_line_counts): Use struct, not it's typedef.
3839 (output_gcov_file): Likewise.
3840 (output_lines): Likewise.
3841 (main): Do not allocate an array.
3842 (output_intermediate_file): Use size of vector container.
3843 (process_file): Resize the vector.
3844 (generate_results): Do not preallocate, use newly added vector
3846 (release_structures): Do not release sources.
3847 (find_source): Use vector methods.
3848 (add_line_counts): Do not use typedef.
3850 2017-10-31 Martin Liska <mliska@suse.cz>
3852 * doc/gcov.texi: Document that.
3853 * gcov.c (add_line_counts): Mark lines with a non-executed
3855 (output_line_beginning): Handle such lines.
3856 (output_lines): Pass new argument.
3857 (output_intermediate_file): Print it in intermediate format.
3859 2017-10-31 Martin Liska <mliska@suse.cz>
3861 * color-macros.h: New file.
3862 * diagnostic-color.c: Factor out color related to macros to
3864 * doc/gcov.texi: Document -k option.
3865 * gcov.c (INCLUDE_STRING): Include string.h.
3866 (print_usage): Add -k option.
3867 (process_args): Parse it.
3868 (pad_count_string): New function.
3869 (output_line_beginning): Likewise.
3870 (DEFAULT_LINE_START): New macro.
3871 (output_lines): Support color output.
3873 2017-10-31 Martin Liska <mliska@suse.cz>
3875 PR gcov-profile/82633
3876 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3877 their interaction with GCOV infrastructure.
3878 * configure.ac: Add -fkeep-{inline,static}-functions to
3880 * configure: Regenerate.
3882 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3885 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3887 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3890 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3891 into a register if it does not fit into an immediate offset field.
3893 2017-10-31 Olivier Hainque <hainque@adacore.com>
3895 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3897 2017-10-31 Julia Koval <julia.koval@intel.com>
3899 * config.gcc: Add gfniintrin.h.
3900 * config/i386/gfniintrin.h: New.
3901 * config/i386/i386-builtin-types.def
3902 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3903 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3904 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3905 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3906 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3907 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3908 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3909 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3910 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3911 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3912 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3913 * config/i386/immintrin.h: Include gfniintrin.h.
3914 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3916 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3918 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3920 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3922 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3923 (arm_ashldi3_1bit): Remove pattern.
3924 (ashrdi3): Remove shift by 1 expansion.
3925 (arm_ashrdi3_1bit): Remove pattern.
3926 (lshrdi3): Remove shift by 1 expansion.
3927 (arm_lshrdi3_1bit): Remove pattern.
3928 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3929 cost of ashldi3 by 1.
3930 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3931 (<shift>di3_neon): Likewise.
3933 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3935 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3936 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3937 and (*aarch64_simd_mov<VQ:mode>).
3938 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3939 pattern alternative.
3940 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3941 attributes to match pattern alternative.
3943 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3945 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3946 * config/rs6000/emmintrin.h: New file.
3947 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3949 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3951 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3952 * (movdi_vfp_cortexa8): Remove pattern.
3954 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3956 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3958 (Specific, alpha*-dec-osf5.1): Remove.
3959 (Specific, mips-sgi-irix5): Remove.
3960 (Specific, mips-sgi-irix6): Remove.
3962 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3965 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3966 arguments to clear_bit_region_be.
3968 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3970 * gimplify.c: Include memmodel.h.
3972 2017-10-30 Martin Jambor <mjambor@suse.cz>
3974 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3975 condition whether loop should be executed at all.
3977 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3979 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3980 gimple folding of vec_madd() intrinsics.
3981 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3982 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3983 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3985 2017-10-30 Richard Biener <rguenther@suse.de>
3987 PR tree-optimization/82762
3989 2017-10-23 Richard Biener <rguenther@suse.de>
3991 PR tree-optimization/82129
3993 2017-08-01 Richard Biener <rguenther@suse.de>
3995 PR tree-optimization/81181
3996 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3997 (compute_antic): ... end of iteration here.
3999 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4001 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4002 -std=iso9899:2017 and -std=gnu17.
4003 * doc/standards.texi (C Language): Document C17 support.
4004 * doc/cpp.texi (Overview): Mention -std=c17.
4005 (Standard Predefined Macros): Document C11 and C17 values of
4006 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4007 * doc/extend.texi (Inline): Do not list individual options for
4008 standards newer than C99.
4009 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4011 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4014 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4016 * asan.c (asan_finish_file): Align asan globals array by shadow
4019 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4022 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4023 (struct store_immediate_info): Add bitregion_start and bitregion_end
4025 (store_immediate_info::store_immediate_info): Add brs and bre
4026 arguments and initialize bitregion_{start,end} from those.
4027 (struct merged_store_group): Add bitregion_start, bitregion_end,
4028 align_base and mask fields. Drop unnecessary struct keyword from
4029 struct store_immediate_info. Add do_merge method.
4030 (clear_bit_region_be): Use memset instead of loop storing zeros.
4031 (merged_store_group::do_merge): New method.
4032 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4033 stores as long as the surrounding bitregions have no gaps.
4034 (merged_store_group::merge_overlapping): Use do_merge.
4035 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4036 is byte aligned, rather than requiring that start and width are
4037 byte aligned. Drop unnecessary struct keyword from
4038 struct store_immediate_info. Allocate and populate also mask array.
4039 Make start of the arrays relative to bitregion_start rather than
4040 start and size them according to bitregion_{end,start} difference.
4041 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4042 struct store_immediate_info.
4043 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4044 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4045 struct keyword from struct store_immediate_info.
4046 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4047 between stores as long as the surrounding bitregions have no gaps.
4049 (struct split_store): Add orig non-static data member.
4050 (split_store::split_store): Initialize orig to false.
4051 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4052 if there is exactly a single original stmt. Change stmts argument
4053 to pointer from reference, if NULL, don't push anything to it. Add
4054 first argument, use it to optimize skipping over orig stmts that
4055 are known to be before bitpos already. Simplify.
4056 (split_group): Return unsigned int count how many stores are or
4057 would be needed rather than a bool. Add allow_unaligned argument.
4058 Change split_stores argument from reference to pointer, if NULL,
4059 only do a dry run computing how many stores would be produced.
4060 Rewritten algorithm to use both alignment and misalign if
4061 !allow_unaligned and handle bitfield stores with gaps.
4062 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4063 from bitregion_start instead of start. Compute allow_unaligned
4064 here, if true, do 2 split_group dry runs to compute which one
4065 produces fewer stores and prefer aligned if equal. Punt if
4066 new count is bigger or equal than original before emitting any
4067 statements, rather than during that. Remove no longer needed
4068 new_ssa_names tracking. Replace num_stmts with
4069 split_stores.length (). Use 32-bit stack allocated entries
4070 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4071 if possible. Handle bitfields with gaps.
4072 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4073 Compute bitregion_{start,end} for the stores and construct
4074 store_immediate_info with that. Formatting fixes.
4076 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4079 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4080 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4082 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4084 * gimplify.c: Include tm_p.h.
4086 * common.opt (gcoff): Re-add as ignored option.
4087 (gcoff1, gcoff2, gcoff3): Likewise.
4089 * Makefile.in (OBJS): Delete sdbout.o.
4090 (GTFILES): Delete $(srcdir)/sdbout.c.
4091 * debug.h: Delete sdb_debug_hooks.
4092 * final.c: Delete sdbout.h include.
4093 (final_scan_insn): Delete SDB_DEBUG check.
4094 (rest_of_clean_state): Likewise.
4095 * output.h: Delete sdb_begin_function_line.
4098 * toplev.c: Delete sdbout.h include.
4099 (process_options): Delete SDB_DEBUG check.
4100 * tree-core.h (tree_type_common): Delete pointer field of
4102 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4103 TYPE_SYMTAB_POINTER.
4104 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4105 (TYPE_SYMTAB_IS_POINTER): Delete.
4106 (TYPE_SYMTAB_IS_DIE): Renumber.
4107 * xcoffout.c: Refer to former sdbout.c file.
4108 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4110 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4111 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4112 (Debugging Options): Delete -gcoff.
4113 (-gstabs): Delete SDB reference.
4115 (-gcoff@var{level}): Delete.
4116 * doc/passes.texi (Debugging information output): Delete SDB and
4117 sdbout.c references.
4118 * doc/tm.texi: Regenerate.
4119 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4120 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4122 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4123 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4124 (SDB_DEBUGGING_INFO): Delete.
4125 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4126 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4127 * target.def (output_source_filename): Delete COFF reference.
4129 * common.opt (gcoff): Delete.
4130 (gxcoff+): Update Negative chain.
4131 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4133 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4134 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4135 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4137 (expand_function_start): Change sdb reference to past tense.
4138 (expand_function_end): Change sdb reference to past tense.
4139 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4140 * opts.c (debug_type_names): Delete coff entry.
4141 (common_handle_option): Delete OPT_gcoff case.
4142 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4144 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4145 * config/cris/cris.h: Delete SDB reference in comment.
4146 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4147 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4148 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4149 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4151 (ix86_expand_prologue): Likewise.
4152 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4153 * config/ia64/ia64.h: Likewise.
4154 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4155 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4157 * config/mmix/mmix.h: Likewise.
4158 * config/nds32/nds32.c: Likewise.
4159 * config/stormy/storym16.h: Likewise.
4160 * config/visium/visium.h: Likewise.
4161 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4163 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4165 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4166 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4167 FRAME_POINTER_REGNUM point at high end of local var area.
4169 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4171 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4172 Move comment around. Do not reset best_edge for a copiable
4173 destination if the copy would cause a partition change.
4174 (better_edge_p): Remove redundant check.
4176 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4178 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4180 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4182 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4183 for math functions that have _Float<N> and _Float<N>X variants.
4184 (mathfn_built_in_2): Add support for math functions that have
4185 _Float<N> and _Float<N>X variants.
4186 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4187 (expand_builtin_mathfn_ternary): Add support for fma with
4188 _Float<N> and _Float<N>X variants.
4189 (expand_builtin): Likewise.
4190 (fold_builtin_3): Likewise.
4191 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4192 create math function _Float<N> and _Float<N>X variants as external
4194 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4195 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4196 the __builtin_ prefix and if not strict ansi, without the prefix.
4197 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4198 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4199 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4200 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4201 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4202 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4203 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4204 function signatures for fma _Float<N> and _Float<N>X variants.
4205 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4206 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4207 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4208 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4209 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4210 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4211 * gencfn-macros.c (print_case_cfn): Add support for math functions
4212 that have _Float<N> and _Float<N>X variants.
4213 (print_define_operator_list): Likewise.
4214 (fltfn_suffixes): Likewise.
4216 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4217 for math functions that have _Float<N> and _Float<N>X variants.
4218 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4219 and _Float<N>X variants.
4220 (COPYSIGN): Likewise.
4223 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4224 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4226 (integer_valued_read_call_p): Likewise.
4227 * fold-const-call.c (fold_const_call_ss): Likewise.
4228 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4229 _Float<N> and _Float<N>X variants.
4230 (fold_const_call_ssss): Add support for fma _Float<N> and
4231 _Float<N>X variants.
4232 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4233 support for copysign and fma _Float<N> and _Float<N>X variants.
4234 (backprop::process_builtin_call_use): Likewise.
4235 * tree-call-cdce.c (can_test_argument_range); Add support for
4236 sqrt _Float<N> and _Float<N>X variants.
4237 (edom_only_function): Likewise.
4238 (get_no_error_domain): Likewise.
4239 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4240 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4241 copysign _Float<N> and _Float<N>X variants.
4242 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4243 handled by machine independent code.
4244 (FMAF128): Likewise.
4245 * doc/cpp.texi (Common Predefined Macros): Document defining
4246 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4247 fma _Float<N> and _Float<N>X variants.
4249 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4252 * config/i386/i386-modes.def (CCFPU): Remove definition.
4253 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4254 (ix86_cc_modes_compatible): Ditto.
4255 (ix86_expand_carry_flag_compare): Ditto.
4256 (ix86_expand_int_movcc): Ditto.
4257 (ix86_expand_int_addcc): Ditto.
4258 (ix86_reverse_condition): Ditto.
4259 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4260 Return true/false for unordered/ordered fp comparisons.
4261 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4262 (ix86_prepare_fp_compare_args): Update for rename.
4263 (ix86_expand_fp_compare): Update for rename. Generate unordered
4264 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4265 (ix86_expand_sse_compare_and_jump): Ditto.
4266 * config/i386/predicates.md (fcmov_comparison_operator):
4267 Remove CCFPU mode handling.
4268 (ix86_comparison_operator): Ditto.
4269 (ix86_carry_flag_operator): Ditto.
4270 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
4271 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
4272 (*cmpu<mode>_cc_i387): Ditto.
4273 (FPCMP): Remove mode iterator.
4274 (unord): Remove mode attribute.
4275 (unord_subst): New define_subst transformation
4276 (unord): New define_subst attribute.
4278 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
4279 (*cmpi<unord>xf_i387): Ditto.
4280 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
4281 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
4282 using unord_subst transformation.
4283 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
4284 (round_saeonly): Also handle CCFP mode.
4285 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
4286 Remove UNSPEC_SAHF unspec handling.
4288 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4290 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
4292 2017-10-27 Jeff Law <law@redhat.com>
4294 * gimple-ssa-sprintf.c: Include domwalk.h.
4295 (class sprintf_dom_walker): New class, derived from dom_walker.
4296 (sprintf_dom_walker::before_dom_children): New function.
4297 (struct call_info): Moved into sprintf_dom_walker class
4298 (compute_formath_length, handle_gimple_call): Likewise.
4299 (sprintf_length::execute): Call the dominator walker rather
4300 than walking the statements.
4302 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4303 gimple statement locations.
4304 (check_array_bounds): Corresponding changes. Get the statement's
4305 location directly from wi->stmt.
4307 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4310 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4312 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4314 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4315 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4317 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4320 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4321 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4322 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4323 maybe_get_pool_constant.
4324 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4327 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4329 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4330 2.26 caveat. Update gas and gld versions.
4331 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4334 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4336 * cgraph.h (set_malloc_flag): Declare.
4337 * cgraph.c (set_malloc_flag_1): New function.
4338 (set_malloc_flag): Likewise.
4339 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4340 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4342 (read_ipa_call_summary): Add support for reading is_return_callee.
4343 (write_ipa_call_summary): Stream is_return_callee.
4344 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4345 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4346 ipa-prop.h, ipa-fnsummary.h.
4347 (pure_const_names): Change to static.
4348 (malloc_state_e): Define.
4349 (malloc_state_names): Define.
4350 (funct_state_d): Add field malloc_state.
4351 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4352 (check_retval_uses): New function.
4353 (malloc_candidate_p): Likewise.
4354 (analyze_function): Add support for malloc attribute.
4355 (pure_const_write_summary): Stream malloc_state.
4356 (pure_const_read_summary): Add support for reading malloc_state.
4357 (dump_malloc_lattice): New function.
4358 (propagate_malloc): New function.
4359 (warn_function_malloc): New function.
4360 (ipa_pure_const::execute): Call propagate_malloc and
4361 ipa_free_fn_summary.
4362 (pass_local_pure_const::execute): Add support for malloc attribute.
4363 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4364 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4366 2017-10-27 Martin Liska <mliska@suse.cz>
4368 PR gcov-profile/82457
4369 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4370 for fork-like functions to be properly instrumented.
4372 2017-10-27 Richard Biener <rguenther@suse.de>
4375 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4376 info when we redirected EH.
4378 2017-10-26 Michael Collison <michael.collison@arm.com>
4380 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4382 (<optab>_trunchf<GPI:mode>2: New pattern.
4383 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4384 * config/aarch64/iterators.md (wv): New mode attribute.
4385 (vf, VF): New mode attributes.
4386 (vgp, VGP): New mode attributes.
4387 (s): Update attribute with SImode and DImode prefixes.
4389 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4391 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4392 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4393 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4394 (nios2_option_overide): Initialize it. Don't allow R0-relative
4395 addressing with PIC.
4396 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4397 (nios2_symbolic_constant_p): Likewise.
4398 (nios2_legitimate_address_p): Likewise.
4399 (nios2_r0rel_section_name_p): New.
4400 (nios2_symbol_ref_in_r0rel_data_p): New.
4401 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4402 (r0rel_constant_p): New.
4403 (nios2_print_operand_address): Handle r0rel_constant_p.
4404 (nios2_cdx_narrow_form_p): Likewise.
4405 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4406 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4407 (Nios II Options): Document -mr0rel-sec.
4409 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4411 * config/nios2/nios2.c: Include xregex.h.
4412 (nios2_gprel_sec_regex): New.
4413 (nios2_option_overide): Initialize it. Don't allow GP-relative
4414 addressing with PIC.
4415 (nios2_small_section_name_p): Check for regex match.
4416 * config/nios2/nios2.opt (mgprel-sec=): New option.
4417 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4418 (Nios II Options): Document -mgprel-sec.
4420 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4422 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4424 2017-10-26 Tom de Vries <tom@codesourcery.com>
4426 PR tree-optimization/82707
4427 * gimple.c (gimple_copy): Fix unsharing of
4428 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4430 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4432 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4433 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4434 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4435 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4436 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4437 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4438 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4439 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4440 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4441 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4442 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4443 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4444 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4445 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4446 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4447 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4448 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4450 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4452 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4454 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4455 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4456 warning to rs6000.c. Remove the Undocumented flag, since it has
4458 (-mabi=ibmlongdouble): Likewise.
4459 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4460 already set, set the default format for long double.
4461 (rs6000_debug_reg_global): Print whether long double is IBM or
4463 (rs6000_option_override_internal): Rework setting long double
4464 format. Only warn if the user is changing the long double default
4465 and they did not use -Wno-psabi.
4466 * doc/invoke.texi (PowerPC options): Update the documentation for
4467 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4469 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4470 Alan Hayward <alan.hayward@arm.com>
4471 David Sherwood <david.sherwood@arm.com>
4473 * rtl.h (wider_subreg_mode): New function.
4474 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4475 rather than an unsigned int *.
4476 * ira-color.c (regno_max_ref_width): Replace with...
4477 (regno_max_ref_mode): ...this new variable.
4478 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4479 Use wider_subreg_mode.
4480 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4481 rather than an unsigned int *.
4482 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4483 (process_alt_operands): Likewise.
4484 (invariant_p): Likewise.
4485 * lra-spills.c (assign_mem_slot): Likewise.
4486 (add_pseudo_to_slot): Likewise.
4487 * lra.c (collect_non_operand_hard_regs): Likewise.
4488 (add_regs_to_insn_regno_info): Likewise.
4489 * reload1.c (regno_max_ref_width): Replace with...
4490 (regno_max_ref_mode): ...this new variable.
4491 (reload): Update accordingly. Update call to
4492 ira_sort_regnos_for_alter_reg.
4493 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4494 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4495 (scan_paradoxical_subregs): Likewise.
4497 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4499 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4500 (aarch64_frame): Add emit_frame_chain boolean.
4501 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4502 Move eh_return case to aarch64_layout_frame.
4503 (aarch64_layout_frame): Initialize emit_frame_chain.
4504 (aarch64_expand_prologue): Use emit_frame_chain.
4506 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4508 * config/aarch64/aarch64.c (aarch64_layout_frame):
4509 Ensure LR is always stored at the bottom of the callee-saves.
4510 Remove rarely used frame layout which saves callee-saves at top of
4511 frame, so the store of LR can be used as a valid probe in all cases.
4513 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4515 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4516 Improve unaligned TImode/TFmode base/offset split.
4518 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4519 Alan Hayward <alan.hayward@arm.com>
4520 David Sherwood <david.sherwood@arm.com>
4522 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4523 * combine.c (find_single_use_1): Likewise.
4524 (expand_field_assignment): Likewise.
4525 (move_deaths): Likewise.
4526 * lra-constraints.c (simplify_operand_subreg): Likewise.
4527 (curr_insn_transform): Likewise.
4528 * lra.c (collect_non_operand_hard_regs): Likewise.
4529 (add_regs_to_insn_regno_info): Likewise.
4530 * rtlanal.c (reg_referenced_p): Likewise.
4531 (covers_regno_no_parallel_p): Likewise.
4533 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4535 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4536 Don't print any bits outside the precision of the value.
4537 * wide-int.cc (test_printing): Add some new tests.
4539 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4541 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4542 supports -xbrace_comment option.
4543 * configure: Regenerate.
4544 * config.in: Regenerate.
4545 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4546 (ASM_CPU_SPEC): Use it.
4548 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4550 * target.def (static_rtx_alignment): New hook.
4551 * targhooks.h (default_static_rtx_alignment): Declare.
4552 * targhooks.c (default_static_rtx_alignment): New function.
4553 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4554 * doc/tm.texi: Regenerate.
4555 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4556 instead of targetm.constant_alignment. Remove call to
4558 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4559 (cris_preferred_mininum_alignment): New function, split out from...
4560 (cris_constant_alignment): ...here.
4561 (cris_static_rtx_alignment): New function.
4562 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4564 (ix86_constant_alignment): ...here.
4565 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4566 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4567 (mmix_static_rtx_alignment): New function.
4568 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4569 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4571 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4574 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4575 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4577 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4579 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4580 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4581 mark as a sign-extending load.
4582 (local_pic_load_u): Define.
4584 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4587 * fold-const.c (operand_equal_for_comparison_p): Also return true
4588 if ARG0 is a simple variant of ARG1 with narrower precision.
4589 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4591 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4593 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4595 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4596 scatter_static, scatter_per_elt.
4597 * x86-tune-costs.h: Add new cost entries.
4599 2017-10-25 Richard Biener <rguenther@suse.de>
4601 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4602 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4603 class pass_fre): Move to ...
4604 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4605 class pass_fre): ... here and adjust for statistics.
4607 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4610 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4611 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4612 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4613 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4616 2017-10-25 Richard Biener <rguenther@suse.de>
4618 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4619 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4620 eliminate_push_avail, eliminate_insert): Move inside...
4621 (class eliminate_dom_walker): ... this class in preparation
4623 (fini_eliminate): Remove by merging with ...
4624 (eliminate): ... this function. Adjust for class changes.
4625 (pass_pre::execute): Remove fini_eliminate call.
4626 (pass_fre::execute): Likewise.
4628 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4631 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4632 (VPERMI2, VPERMI2I): New mode iterators.
4633 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4634 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4636 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4637 mode iterator. Remove 3 old define_insn patterns.
4638 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4639 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4640 VPERMI2 mode iterator, remove the other two expanders.
4641 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4642 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4643 instructions, remove the other two patterns.
4644 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4645 mode iterator, remove the other two patterns.
4646 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4647 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4648 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4649 and adjust argument order accordingly.
4650 (ix86_expand_vec_perm): Adjust caller.
4651 (expand_vec_perm_1): Likewise.
4652 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4653 (expand_vec_perm_vpermt2_vpshub2): ... this.
4654 (ix86_expand_vec_perm_const_1): Adjust caller.
4655 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4658 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4659 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4660 (vec_shl_<mode>): Remove unused expander.
4661 (avx512bw_<shift_insn><mode>3): New define_insn.
4662 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4663 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4665 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4668 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4671 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4673 PR rtl-optimization/82396
4674 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4675 (autopref_multipass_init): Simplify initialization.
4676 (autopref_rank_data): Simplify sort order.
4677 * gcc/sched-int.h (autopref_multipass_data_): Remove
4678 multi_mem_insn_p, min_offset and max_offset.
4680 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4683 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4684 Check special value of flag_omit_frame_pointer.
4685 (aarch64_can_eliminate): Likewise.
4686 (aarch64_override_options_after_change_1): Simplify handling of
4687 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4689 2017-10-24 Richard Biener <rguenther@suse.de>
4691 PR tree-optimization/82697
4692 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4693 zero for conditional load and unconditional store.
4695 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4697 * doc/install.texi: Document bootstrap-cet.
4699 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4702 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4703 ENDBR instruction at function entrance if function is only
4706 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4709 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4710 patterns to better describe from which operation the CF is computed.
4711 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4712 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4713 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4714 is 0, use _0 suffixed expanders instead of emitting a comparison
4717 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4719 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4720 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4722 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4725 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4726 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4727 * loop-iv.c (iv_get_reaching_def): Likewise.
4728 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4729 variable is promoted and the partition contains undefined values.
4731 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4733 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4735 (nios2_address_cost): Define.
4736 (nios2_legitimize_address): Recognize (exp + constant) directly.
4737 (TARGET_ADDRESS_COST): Define.
4739 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4741 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4742 (nios2_symbolic_memory_operand_p): Declare.
4743 (nios2_split_large_constant): Declare.
4744 (nios2_split_symbolic_memory_operand): Declare.
4745 * config/nios2/nios2.c: Adjust includes.
4746 (nios2_symbolic_constant_allowed): New.
4747 (nios2_symbolic_constant_p): New.
4748 (nios2_plus_symbolic_constant_p): New.
4749 (nios2_valid_addr_expr_p): Recognize addresses involving
4751 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4752 (nios2_symbolic_memory_operand_p): New.
4753 (nios2_large_constant_p): New.
4754 (nios2_split_large_constant): New.
4755 (nios2_split_plus_large_constant): New.
4756 (nios2_split_symbolic_memory_operand): New.
4757 (nios2_legitimize_address): Code refactoring. Handle addresses
4758 involving symbolic constants.
4759 (nios2_emit_move_sequence): Likewise.
4760 (nios2_print_operand): Improve error output.
4761 (nios2_print_operand_address): Handle LO_SUM.
4762 (nios2_cdx_narrow_form_p): Likewise.
4763 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4764 operands involving symbolic constants.
4765 (movhi_internal, movsi_internal): Likewise.
4766 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4767 (extendhisi2, extendqi<mode>2): Likewise.
4769 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4771 * tree-pass.h (PROP_rtl_split_insns): Define.
4772 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4774 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4776 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4778 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4781 * target.def (const_not_ok_for_debug_p): Default to
4782 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4783 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4784 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4785 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4786 which targetm.const_not_ok_for_debug_p returned true.
4787 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4789 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4791 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4792 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4793 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4795 (ix86_const_not_ok_for_debug_p): New function.
4796 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4797 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4799 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4802 * system.h: Conditionally include "unique-ptr.h" if
4803 INCLUDE_UNIQUE_PTR is defined.
4804 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4805 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4807 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4809 * config/rl78/rl78.md: New define_expand "subdi3".
4811 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4814 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4815 DF_REF_INSN if DF_REF_INSN_INFO is false.
4817 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4819 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4820 xmm_move instead of sse_move.
4821 (sse_store_index): New function.
4822 (ix86_register_move_cost): Be more sensible about mismatch stall;
4823 model AVX moves correctly; make difference between sse->integer and
4825 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4826 moves; make difference between SSE and AVX.
4827 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4828 and zmm_move. Increase size of sse load and store tables;
4829 add unaligned load and store tables; add ssemmx_to_integer.
4830 * x86-tune-costs.h: Update all entries according to real
4831 move latencies from Agner Fog's manual and chip documentation.
4833 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4836 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4837 * config/i386/constraints.md (Wf): New constraint.
4838 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4839 (cmp<dwi>_doubleword): Removed.
4840 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4841 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4842 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4843 expand with cmp<dwi>_doubleword. For LTU and GEU use
4844 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4846 * common.opt (gcolumn-info): Enable by default.
4847 * doc/invoke.texi (gcolumn-info): Document new default.
4849 2017-10-23 Richard Biener <rguenther@suse.de>
4851 PR tree-optimization/82672
4852 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4853 Fold the stmt if we propagated into it.
4855 2017-10-23 Richard Biener <rguenther@suse.de>
4857 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4858 (bitmap_remove_expr_from_set): ... this. All callers call this
4859 for non-constant values.
4860 (bitmap_set_subtract): Rename to...
4861 (bitmap_set_subtract_expressions): ... this. Adjust and
4863 (bitmap_set_contains_value): Remove superfluous check.
4864 (bitmap_set_replace_value): Inline into single caller ...
4865 (bitmap_value_replace_in_set): ... here and simplify.
4866 (dependent_clean): Merge into ...
4867 (clean): ... this using an overload. Adjust.
4868 (prune_clobbered_mems): Adjust.
4869 (compute_antic_aux): Likewise.
4870 (compute_partial_antic_aux): Likewise.
4872 2017-10-23 Richard Biener <rguenther@suse.de>
4874 PR tree-optimization/82129
4876 2017-08-01 Richard Biener <rguenther@suse.de>
4878 PR tree-optimization/81181
4879 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4880 (compute_antic): ... end of iteration here.
4882 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4884 * target.def (starting_frame_offset): New hook.
4885 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4886 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4887 * doc/tm.texi.in: Regenerate.
4888 * hooks.h (hook_hwi_void_0): Declare.
4889 * hooks.c (hook_hwi_void_0): New function.
4890 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4891 STARTING_FRAME_OFFSET.
4892 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4893 * reload1.c (reload): Likewise.
4894 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4895 instead of STARTING_FRAME_OFFSET.
4896 * function.c (try_fit_stack_local): Likewise.
4897 (assign_stack_local_1): Likewise
4898 (instantiate_virtual_regs): Likewise.
4899 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4900 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4901 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4902 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4903 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4904 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4905 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4906 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4907 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4908 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4909 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4910 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4911 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4912 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4913 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4914 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4915 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4916 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4917 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4918 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4919 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4920 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4921 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4922 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4923 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4924 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4925 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4926 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4927 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4928 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4929 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4930 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4931 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4932 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4933 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4934 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4935 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4936 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4937 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4938 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4939 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4940 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4941 return a HOST_WIDE_INT.
4942 (avr_builtin_setjmp_frame_value): Use it instead of
4943 STARTING_FRAME_OFFSET.
4944 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4945 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4946 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4948 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4949 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4950 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4951 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4952 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4953 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4954 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4955 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4956 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4957 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4958 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4959 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4961 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4962 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4963 * config/mips/mips.c (mips_compute_frame_info): Refer to
4964 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4965 (mips_starting_frame_offset): New function.
4966 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4967 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4968 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4969 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4970 and return a HOST_WIDE_INT.
4971 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4972 (mmix_initial_elimination_offset): Refer to
4973 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4974 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4975 * config/pa/pa.c (pa_starting_frame_offset): New function.
4976 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4977 (pa_expand_prologue): Likewise.
4978 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4979 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4980 !FRAME_GROWS_DOWNWARD handling to...
4981 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4982 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4983 !FRAME_GROWS_DOWNWARD handling to...
4984 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4985 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4986 !FRAME_GROWS_DOWNWARD handling to...
4987 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4988 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4990 (rs6000_starting_frame_offset): New function.
4991 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4992 !FRAME_GROWS_DOWNWARD handling to...
4993 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4994 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4995 !FRAME_GROWS_DOWNWARD handling to...
4996 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4997 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4998 !FRAME_GROWS_DOWNWARD handling to...
4999 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5000 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5001 (rs6000_starting_frame_offset): New function.
5002 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5003 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5004 * config/vax/vax.c (vax_starting_frame_offset): New function.
5005 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5006 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5007 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5008 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5009 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5010 * system.h (STARTING_FRAME_OFFSET): Poison.
5012 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5014 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5015 SCALAR_TYPE_MODE instead of TYPE_MODE.
5017 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5018 Alan Hayward <alan.hayward@arm.com>
5019 David Sherwood <david.sherwood@arm.com>
5021 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5023 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5024 Alan Hayward <alan.hayward@arm.com>
5025 David Sherwood <david.sherwood@arm.com>
5027 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5029 2017-10-23 Richard Biener <rguenther@suse.de>
5031 PR tree-optimization/82129
5032 * tree-ssa-pre.c (bitmap_set_and): Remove.
5033 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5034 canonicalizing expressions in the set to those with lowest
5035 ID rather than taking that from the first edge.
5037 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5039 * combine.c (rtx_equal_for_field_assignment_p): Use
5040 byte_lowpart_offset.
5042 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5043 Alan Hayward <alan.hayward@arm.com>
5044 David Sherwood <david.sherwood@arm.com>
5046 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5047 to a SUBREG_PROMOTED_VAR.
5049 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5050 Alan Hayward <alan.hayward@arm.com>
5051 David Sherwood <david.sherwood@arm.com>
5053 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5054 (expand_debug_source_expr): Likewise.
5055 * combine.c (combine_simplify_rtx): Likewise.
5056 * cse.c (fold_rtx): Likewise.
5057 * optabs.c (expand_float): Likewise.
5058 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5059 (simplify_binary_operation_1): Likewise.
5061 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5062 Alan Hayward <alan.hayward@arm.com>
5063 David Sherwood <david.sherwood@arm.com>
5065 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5066 (record_promoted_value): Likewise.
5067 * expr.c (expand_expr_real_2): Likewise.
5068 * ree.c (update_reg_equal_equiv_notes): Likewise.
5069 (combine_set_extension): Likewise.
5070 * rtlanal.c (low_bitmask_len): Likewise.
5071 * simplify-rtx.c (neg_const_int): Likewise.
5072 (simplify_binary_operation_1): Likewise.
5074 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5075 Alan Hayward <alan.hayward@arm.com>
5076 David Sherwood <david.sherwood@arm.com>
5078 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5079 * regcprop.c (maybe_mode_change): Likewise.
5080 * reload1.c (alter_reg): Likewise.
5082 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5084 * inchash.h (inchash::hash::add_wide_int): New function.
5085 * lto-streamer-out.c (hash_tree): Use it.
5087 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5089 * inchash.h (inchash::hash::add_wide_int): Rename to...
5090 (inchash::hash::add_hwi): ...this.
5091 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5092 (polymorphic_call_target_hasher::hash): Likewise.
5093 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5094 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5095 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5096 * lto-streamer-out.c (hash_tree): Likewise.
5097 * optc-save-gen.awk: Likewise.
5098 * tree.c (add_expr): Likewise.
5100 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5103 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5104 for ordered inequality comparisons even with TARGET_IEEE_FP.
5106 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5109 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5110 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5111 Expand with cmp<dwi>_doubleword.
5113 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5115 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5116 List CET intrinsics.
5117 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5118 specific to -fcf-protection option.
5120 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5122 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5123 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5124 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5125 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5126 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5127 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5128 (extra_objs): Add cet.o for Linux/x86 targets.
5129 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5130 * config/i386/cet.c: New file.
5131 * config/i386/cetintrin.h: Likewise.
5132 * config/i386/t-cet: Likewise.
5133 * config/i386/cpuid.h (bit_SHSTK): New.
5134 (bit_IBT): Likewise.
5135 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5136 pass IBT and SHSTK bits.
5137 * config/i386/i386-builtin-types.def
5138 (VOID_FTYPE_UNSIGNED_PVOID): New.
5139 (VOID_FTYPE_UINT64_PVOID): Likewise.
5140 * config/i386/i386-builtin.def: Add CET intrinsics.
5141 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5142 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5143 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5144 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5146 * config/i386/i386.c (rest_of_insert_endbranch): New.
5147 (pass_data_insert_endbranch): Likewise.
5148 (pass_insert_endbranch): Likewise.
5149 (make_pass_insert_endbranch): Likewise.
5150 (ix86_notrack_prefixed_insn_p): Likewise.
5151 (ix86_target_string): Add -mibt, -mshstk flags.
5152 (ix86_option_override_internal): Add flag_cf_protection
5154 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5155 (ix86_print_operand): Add 'notrack' prefix output.
5156 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5157 (ix86_expand_builtin): Expand CET intrinsics.
5158 (x86_output_mi_thunk): Add 'endbranch' instruction.
5159 * config/i386/i386.h (TARGET_IBT): New.
5160 (TARGET_IBT_P): Likewise.
5161 (TARGET_SHSTK): Likewise.
5162 (TARGET_SHSTK_P): Likewise.
5163 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5164 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5165 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5166 (builtin_setjmp_setup): New pattern.
5167 (builtin_longjmp): Likewise.
5168 (rdssp<mode>): Likewise.
5169 (incssp<mode>): Likewise.
5170 (saveprevssp): Likewise.
5171 (rstorssp): Likewise.
5172 (wrss<mode>): Likewise.
5173 (wruss<mode>): Likewise.
5174 (setssbsy): Likewise.
5175 (clrssbsy): Likewise.
5176 (nop_endbr): Likewise.
5177 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5179 * config/i386/immintrin.h: Include <cetintrin.h>.
5180 * config/i386/linux-common.h
5181 (file_end_indicate_exec_stack_and_cet): New prototype.
5182 (TARGET_ASM_FILE_END): New.
5184 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5186 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5187 latencies instead of having separate table; make difference between
5188 integer and float costs.
5189 * i386.h (processor_costs): Remove scalar_stmt_cost,
5190 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5191 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5193 * x86-tune-costs.h: Remove entries which has been removed in
5194 procesor_costs from all tables; make cond_taken_branch_cost
5195 and cond_not_taken_branch_cost COST_N_INSNS based.
5197 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5199 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5201 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5203 * config/i386/i386.md (isa): Remove fma_avx512f.
5204 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5205 <avx512>_fmadd_<mode>_mask3<round_name>,
5206 <avx512>_fmsub_<mode>_mask<round_name>,
5207 <avx512>_fmsub_<mode>_mask3<round_name>,
5208 <avx512>_fnmadd_<mode>_mask<round_name>,
5209 <avx512>_fnmadd_<mode>_mask3<round_name>,
5210 <avx512>_fnmsub_<mode>_mask<round_name>,
5211 <avx512>_fnmsub_<mode>_mask3<round_name>,
5212 <avx512>_fmaddsub_<mode>_mask<round_name>,
5213 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5214 <avx512>_fmsubadd_<mode>_mask<round_name>,
5215 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5216 (*vec_widen_umult_even_v16si<mask_name>,
5217 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5218 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5220 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5222 * extend.texi: Add 'nocf_check' documentation.
5223 * gimple.texi: Add second parameter to
5224 gimple_build_call_from_tree.
5225 * invoke.texi: Add -fcf-protection documentation.
5226 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5228 2017-10-20 Richard Biener <rguenther@suse.de>
5230 PR tree-optimization/82473
5231 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5232 the largest input type.
5234 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5236 * c-attribs.c (handle_nocf_check_attribute): New function.
5237 (c_common_attribute_table): Add 'nocf_check' handling.
5238 * gimple-parser.c: Add second argument NULL to
5239 gimple_build_call_from_tree.
5240 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5241 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5243 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5244 * common.opt: Add fcf-protection flag.
5245 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5246 * flag-types.h: Add enum cf_protection_level.
5247 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5248 Add 'nocf_check' attribute propagation to gimple call.
5249 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5250 (gimple_build_call_from_tree): Update prototype.
5251 (gimple_call_nocf_check_p): New function.
5252 (gimple_call_set_nocf_check): Likewise.
5253 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5254 * ipa-icf.c: Add nocf_check attribute in statement hash.
5255 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5256 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5257 * toplev.c (process_options): Add flag_cf_protection handling.
5259 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5261 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5263 2017-10-20 Richard Biener <rguenther@suse.de>
5265 PR tree-optimization/82603
5266 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5267 remove false predicated stores.
5269 2017-10-20 Richard Biener <rguenther@suse.de>
5271 * graphite-isl-ast-to-gimple.c
5272 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
5273 Remove return value and simplify, dump copied stmt after lhs
5275 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
5276 Reduce dump verbosity.
5277 (gsi_insert_earliest): Likewise.
5278 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
5279 * graphite.c (print_global_statistics): Adjust dumping.
5280 (print_graphite_scop_statistics): Likewise.
5281 (print_graphite_statistics): Do not dump loops here.
5282 (graphite_transform_loops): But here.
5284 2017-10-20 Nicolas Roche <roche@adacore.com>
5286 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
5287 * configure: Regenerate.
5289 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5292 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
5293 functions when optimizing replace GIMPLE_RETURN stmts with
5294 calls to __builtin_unreachable ().
5297 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5298 for -fsanitize=thread link of executables.
5299 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
5300 link of executables.
5303 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5305 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5306 define_insns for logical vector shifts to use VI248_AVX512BW
5307 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5308 condition, useless isa and prefix attributes. Change the first
5309 2 of these define_insns to ...
5310 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5311 define_insn for avx512vl.
5312 (<shift_insn><mode>3): ... and this, new define_insn without
5313 masking for non-avx512vl.
5316 * config/i386/sse.md (*andnot<mode>3,
5317 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5318 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5319 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5320 not applied use empty suffix even for TARGET_AVX512VL.
5321 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5322 is applied, supply evex,evex or evex,evex,evex instead of just
5325 2017-10-20 Julia Koval <julia.koval@intel.com>
5327 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5328 (OPTION_MASK_ISA_GFNI_UNSET): New.
5329 (ix86_handle_option): Handle OPT_mgfni.
5330 * config/i386/cpuid.h (bit_GFNI): New.
5331 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5332 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5333 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5334 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5335 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5336 * config/i386/i386.opt: Add mgfni.
5338 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5340 * config/msp430/msp430.c (msp430_option_override): Disable
5341 -fdelete-null-pointer-checks.
5342 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5344 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5346 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5347 of x87 and SSE instructions.
5349 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5351 * asan.c (create_cond_insert_point): Do not update edge count.
5352 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5353 (afdo_propagate_circuit): Likewise.
5354 (afdo_calculate_branch_prob): Likewise.
5355 (afdo_annotate_cfg): Likewise.
5356 * basic-block.h (struct edge_def): Remove count.
5357 (edge_def::count): New accessor.
5358 * bb-reorder.c (rotate_loop): Update.
5359 (find_traces_1_round): Update.
5360 (connect_traces): Update.
5361 (sanitize_hot_paths): Update.
5362 * cfg.c (unchecked_make_edge): Update.
5363 (make_single_succ_edge): Update.
5364 (check_bb_profile): Update.
5365 (dump_edge_info): Update.
5366 (update_bb_profile_for_threading): Update.
5367 (scale_bbs_frequencies_int): Update.
5368 (scale_bbs_frequencies_gcov_type): Update.
5369 (scale_bbs_frequencies_profile_count): Update.
5370 (scale_bbs_frequencies): Update.
5371 * cfganal.c (connect_infinite_loops_to_exit): Update.
5372 * cfgbuild.c (compute_outgoing_frequencies): Update.
5373 (find_many_sub_basic_blocks): Update.
5374 * cfgcleanup.c (try_forward_edges): Update.
5375 (try_crossjump_to_edge): Update
5376 * cfgexpand.c (expand_gimple_cond): Update
5377 (expand_gimple_tailcall): Update
5378 (construct_exit_block): Update
5379 * cfghooks.c (verify_flow_info): Update
5380 (redirect_edge_succ_nodup): Update
5381 (split_edge): Update
5382 (make_forwarder_block): Update
5383 (duplicate_block): Update
5384 (account_profile_record): Update
5385 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5386 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5387 * cfgloopmanip.c (scale_loop_profile): Update.
5389 (lv_adjust_loop_entry_edge): Update.
5390 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5391 (force_nonfallthru_and_redirect): Update.
5392 (purge_dead_edges): Update.
5393 (rtl_flow_call_edges_add): Update.
5394 * cgraphunit.c (init_lowered_empty_function): Update.
5395 (cgraph_node::expand_thunk): Update.
5396 * gimple-pretty-print.c (dump_probability): Update.
5397 (dump_edge_probability): Update.
5398 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5399 * haifa-sched.c (sched_create_recovery_edges): Update.
5400 * hsa-gen.c (convert_switch_statements): Update.
5401 * ifcvt.c (dead_or_predicable): Update.
5402 * ipa-inline-transform.c (inline_transform): Update.
5403 * ipa-split.c (split_function): Update.
5404 * ipa-utils.c (ipa_merge_profiles): Update.
5405 * loop-doloop.c (add_test): Update.
5406 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5407 * lto-streamer-in.c (input_cfg): Update.
5408 (input_function): Update.
5409 * lto-streamer-out.c (output_cfg): Update.
5410 * modulo-sched.c (sms_schedule): Update.
5411 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5412 * predict.c (maybe_hot_edge_p): Update.
5413 (unlikely_executed_edge_p): Update.
5414 (probably_never_executed_edge_p): Update.
5415 (dump_prediction): Update.
5416 (drop_profile): Update.
5417 (propagate_unlikely_bbs_forward): Update.
5418 (determine_unlikely_bbs): Update.
5419 (force_edge_cold): Update.
5420 * profile.c (compute_branch_probabilities): Update.
5421 * reg-stack.c (better_edge): Update.
5422 * shrink-wrap.c (handle_simple_exit): Update.
5423 * tracer.c (better_p): Update.
5424 * trans-mem.c (expand_transaction): Update.
5425 (split_bb_make_tm_edge): Update.
5426 * tree-call-cdce.c: Update.
5427 * tree-cfg.c (gimple_find_sub_bbs): Update.
5428 (gimple_split_edge): Update.
5429 (gimple_duplicate_sese_region): Update.
5430 (gimple_duplicate_sese_tail): Update.
5431 (gimple_flow_call_edges_add): Update.
5432 (insert_cond_bb): Update.
5433 (execute_fixup_cfg): Update.
5434 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5435 * tree-complex.c (expand_complex_div_wide): Update.
5436 * tree-eh.c (lower_resx): Update.
5437 (unsplit_eh): Update.
5438 (cleanup_empty_eh_move_lp): Update.
5439 * tree-inline.c (copy_edges_for_bb): Update.
5440 (freqs_to_counts): Update.
5441 (copy_cfg_body): Update.
5442 * tree-ssa-dce.c (remove_dead_stmt): Update.
5443 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5444 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5445 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5446 (unloop_loops): Update.
5447 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5448 * tree-ssa-loop-split.c (connect_loops): Update.
5449 (split_loop): Update.
5450 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5451 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5452 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5453 * tree-ssa-reassoc.c (branch_fixup): Update.
5454 * tree-ssa-tail-merge.c (replace_block_by): Update.
5455 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5456 (compute_path_counts): Update.
5457 (update_profile): Update.
5458 (recompute_probabilities): Update.
5459 (update_joiner_offpath_counts): Update.
5460 (estimated_freqs_path): Update.
5461 (freqs_to_counts_path): Update.
5462 (clear_counts_path): Update.
5463 (ssa_fix_duplicate_block_edges): Update.
5464 (duplicate_thread_path): Update.
5465 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5466 (case_bit_test_cmp): Update.
5467 (collect_switch_conv_info): Update.
5468 (gen_inbound_check): Update.
5469 (do_jump_if_equal): Update.
5470 (emit_cmp_and_jump_insns): Update.
5471 * tree-tailcall.c (decrease_profile): Update.
5472 (eliminate_tail_call): Update.
5473 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5474 (vect_do_peeling): Update.
5475 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5476 * ubsan.c (ubsan_expand_null_ifn): Update.
5477 (ubsan_expand_ptr_ifn): Update.
5478 * value-prof.c (gimple_divmod_fixed_value): Update.
5479 (gimple_mod_pow2): Update.
5480 (gimple_mod_subtract): Update.
5481 (gimple_ic): Update.
5482 (gimple_stringop_fixed_value): Update.
5484 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5487 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5489 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5491 PR rtl-optimization/82395
5492 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5493 based on non_spilled_static_chain_regno_p.
5495 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5497 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5498 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5499 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5501 2017-10-19 Martin Sebor <msebor@redhat.com>
5503 PR tree-optimization/82596
5504 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5506 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5508 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5509 (get_mem_refs_of_builtin_call): Likewise.
5510 * builtins.c (expand_builtin_apply): Adjust call to
5511 allocate_dynamic_stack_space.
5512 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5513 the third argument to allocate_dynamic_stack_space, otherwise -1.
5514 (expand_builtin): Deal with all alloca variants.
5515 (is_inexpensive_builtin): Likewise.
5516 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5517 * calls.c (special_function_p): Deal with all alloca variants.
5518 (initialize_argument_information): Adjust call to
5519 allocate_dynamic_stack_space.
5520 (expand_call): Likewise.
5521 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5522 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5523 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5524 use it for the stack usage computation.
5525 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5526 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5527 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5528 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5529 (in_loop_p): Remove first argument and useless check.
5530 (pass_walloca::execute): Remove useless test and adjust call to above.
5531 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5532 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5533 (gimplify_call_expr): Deal with all alloca variants.
5534 * hsa-gen.c (gen_hsa_alloca): Likewise.
5535 (gen_hsa_insns_for_call): Likewise.
5536 * ipa-pure-const.c (special_builtin_state): Likewise.
5537 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5538 * tree-object-size.c (alloc_object_size): Likewise.
5539 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5540 (call_may_clobber_ref_p_1): Likewise.
5541 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5542 (ccp_fold_stmt): Likewise.
5543 (optimize_stack_restore): Likewise.
5544 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5545 (mark_all_reaching_defs_necessary_1): Likewise.
5546 (propagate_necessity): Likewise.
5547 (eliminate_unnecessary_stmts): Likewise.
5548 * tree.c (build_common_builtin_nodes): Build
5549 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5550 (build_alloca_call_expr): New function.
5551 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5552 (CASE_BUILT_IN_ALLOCA): Likewise.
5553 (build_alloca_call_expr): Declare.
5554 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5556 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5559 * dwarf2out.c (new_die_raw): New static inline function.
5560 (new_die): Use it to create the DIE.
5561 (add_AT_external_die_ref): Likewise.
5562 (clone_die): Likewise.
5563 (clone_as_declaration): Likewise.
5564 (dwarf2out_vms_debug_main_pointer): Likewise.
5565 (base_type_die): Likewise. Remove early return for corner cases.
5566 Do not call add_pubtype on the DIE here.
5567 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5568 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5569 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5570 native order exists for base types, attach the DIE manually and call
5571 add_pubtype on it. Do not equate a reverse order DIE to the type.
5573 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5575 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5576 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5578 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5580 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5582 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5583 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5584 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5585 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5586 * opts.c (sanitizer_opts): Add builtin.
5587 * ubsan.c (instrument_builtin): New function.
5588 (pass_ubsan::execute): Call it.
5589 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5590 * doc/invoke.texi: Document -fsanitize=builtin.
5592 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5593 builtins, store max (log2 (align), 0) into uchar field instead of
5594 align into uptr field.
5595 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5596 store uchar 0 field instead of uptr 0 field.
5597 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5598 instead of passing one address of struct with 2 locations pass
5599 two addresses of structs with 1 location each.
5600 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5601 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5602 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5603 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5604 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5605 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5606 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5607 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5609 2017-10-19 Martin Liska <mliska@suse.cz>
5612 * file-find.c (remove_prefix): Remove.
5613 * file-find.h (remove_prefix): Likewise.
5614 * gcc-ar.c: Remove smartness of lookup.
5616 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5618 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5619 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5620 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5622 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5625 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5626 (setcc + and to xor + setcc): New peephole2.
5628 2017-10-19 Tom de Vries <tom@codesourcery.com>
5630 * doc/sourcebuild.texi (Test Directives, Variants of
5631 dg-require-support): Add dg-require-stack-size.
5633 2017-10-19 Martin Liska <mliska@suse.cz>
5636 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5637 that have a large alignment.
5638 (gimplify_target_expr): Likewise.
5640 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5642 PR rtl-optimization/82602
5643 * ira.c (rtx_moveable_p): Return false for volatile asm.
5645 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5648 * config/i386/i386-modes.def (CCGZ): New CC mode.
5649 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5650 * config/i386/predicates.md (ix86_comparison_operator):
5652 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5653 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5654 with double-word subtraction.
5655 (put_condition_code): Handle CCGZmode.
5657 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5659 * wide-int.cc (debug (const wide_int &)): New.
5660 (debug (const wide_int *)): New.
5661 (debug (const widest_int &)): New.
5662 (debug (const widest_int *)): New.
5664 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5667 * lra-constraints.c (curr_insn_transform): Use non-input operand
5668 instead of output one for matched reload.
5670 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5672 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5673 (tree-ssa-loop-ivopts.h): New header file.
5674 (struct builtin_info): New fields.
5675 (classify_builtin_1): Compute and record base and offset parts for
5676 memset builtin partition by calling strip_offset.
5677 (offset_cmp, fuse_memset_builtins): New functions.
5678 (finalize_partitions): Fuse adjacent memset partitions by calling
5680 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5681 Expose the interface.
5682 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5684 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5686 PR tree-optimization/82574
5687 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5688 that data reference must be executed exactly once per iteration
5689 against the outermost loop in nest.
5690 (classify_partition): Update call to above function.
5692 2017-10-18 Richard Biener <rguenther@suse.de>
5694 PR tree-optimization/82591
5695 * graphite.c (graphite_transform_loops): Move code gen message
5697 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5698 Here. Handle scop_to_isl_ast failing.
5699 (scop_to_isl_ast): Limit the number of ISL operations.
5701 2017-10-18 Richard Biener <rguenther@suse.de>
5703 * graphite-isl-ast-to-gimple.c
5704 (translate_isl_ast_to_gimple::set_rename): Simplify.
5705 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5706 (graphite_copy_stmts_from_block): ... here.
5707 (copy_bb_and_scalar_dependences): Simplify.
5708 (add_parameters_to_ivs_params): Canonicalize.
5709 (generate_entry_out_of_ssa_copies): Simplify.
5710 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5711 by passing in ISL dimension.
5712 (parameter_index_in_region_1): Rename to ...
5713 (parameter_index_in_region): ... this.
5714 (extract_affine): Adjust assert, pass down parameter index.
5715 (add_param_constraints): Use range-info when available.
5716 (build_scop_context): Adjust.
5717 * sese.c (new_sese_info): Adjust.
5718 (free_sese_info): Likewise.
5719 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5720 Remove unused typedefs.
5721 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5723 2017-10-18 Martin Liska <mliska@suse.cz>
5725 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5727 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5729 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5730 (s390_sched_init): Do not reset s390_sched_state if we entered the
5731 current basic block via a fallthru edge and all others are unlikely.
5733 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5735 * config/s390/s390.c (NUM_SIDES): New variable.
5736 (LONGRUNNING_THRESHOLD): New variable.
5737 (LATENCY_FACTOR): New variable.
5738 (s390_sched_score): Decrease score for long-running instructions on
5740 (s390_sched_variable_issue): Perform bookkeeping for long-running
5743 2017-10-18 Richard Biener <rguenther@suse.de>
5745 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5746 Simplify with removal of the parameter rename map.
5747 (set_rename): Likewise.
5748 (should_copy_to_new_region): Likewise.
5749 (graphite_copy_stmts_from_block): Likewise.
5750 (copy_bb_and_scalar_dependences): Remove initialization of
5751 unused copied_bb_map.
5753 (copy_internal_parameters): Likewise.
5754 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5755 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5756 Use INTEGRAL_TYPE_P.
5757 (parameter_index_in_region_1): Rename to ...
5758 (assign_parameter_index_in_region): ... this. Assert we have
5759 a parameter we handle.
5760 (scan_tree_for_params): Adjust.
5761 * sese.h (parameter_rename_map_t): Remove.
5762 (struct sese_info_t): Remove unused parameter_rename_map and
5763 copied_bb_map members.
5764 * sese.c (new_sese_info): Adjust.
5765 (free_sese_info): Likewise.
5767 2017-10-18 Martin Liska <mliska@suse.cz>
5770 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5771 on an abnormal edge.
5773 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5775 * doc/invoke.texi (ffunction-sections and fdata-sections):
5778 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5780 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5781 the use statement can throw internally.
5783 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5785 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5786 any RTX present on the RHS of a SET.
5787 * compare-elim.c (try_eliminate_compare): Restore comment.
5789 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5791 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5792 may be also called on tcc_type nodes.
5793 * langhooks.c (lhd_tree_size): Likewise.
5795 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5797 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5798 format_warning_at_substring.
5799 (maybe_warn): Convert source_range * param to a location_t. Pass
5800 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5801 (format_directive): Remove code to extract source_ranges and
5802 source_range * in favor of just a location_t.
5803 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5805 * substring-locations.c (format_warning_va): Convert
5806 source_range * param to a location_t.
5807 (format_warning_at_substring): Likewise.
5808 * substring-locations.h (format_warning_va): Likewise.
5809 (format_warning_at_substring): Likewise.
5811 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5813 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5815 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5816 and scatter/gather ops.
5818 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5819 vec_gather_load and vec_scatter_store.
5820 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5821 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5822 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5823 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5824 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5825 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5827 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5829 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5830 Detect FTST instruction and handle its register pops. Only pop
5831 second operand if can_pop_second_op is true.
5832 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5833 set can_pop_second_op to false in the compare_for_stack_reg call.
5835 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5836 output_fp_compare for stack register operands.
5837 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5838 instructions here. Do not emit stack register pops here. Assert
5839 that FCOMPP pops next to top stack register. Rewrite function.
5841 2017-10-17 Nathan Sidwell <nathan@acm.org>
5844 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5845 use DECL_ASSEMBLER_NAME_RAW.
5848 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5851 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5852 Wilco Dijkstra <wilco.dijkstra@arm.com>
5854 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5855 converstion to Pmode from the buf_addr.
5857 2017-10-17 Richard Biener <rguenther@suse.de>
5859 * graphite-dependences.c (scop_get_reads_and_writes): Change
5860 output parameters to references.
5862 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5864 PR 71026/tree-optimization
5865 * fold-const.c (distribute_real_division): Removed.
5866 (fold_binary_loc): Remove calls to distribute_real_divison.
5868 2017-10-17 Richard Biener <rguenther@suse.de>
5870 * graphite-scop-detection.c
5871 (scop_detection::stmt_has_simple_data_refs_p): Always use
5872 the full nest as region.
5873 (try_generate_gimple_bb): Likewise.
5874 * sese.c (scalar_evolution_in_region): Simplify now that
5875 SCEV can handle instantiation in regions.
5876 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5877 in the non-loop part of a function if requested.
5879 2017-10-17 Richard Biener <rguenther@suse.de>
5881 PR tree-optimization/82563
5882 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5884 (graphite_regenerate_ast_isl): Call it.
5885 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5887 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5889 PR tree-optimization/82549
5890 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5891 Formatting fixes. Instead of calling make_bit_field_ref with negative
5894 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5896 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5897 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5898 _mm_maskz_reduce_ss): New.
5899 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5900 __builtin_ia32_reducess_mask): Ditto..
5901 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5902 * config/i386/sse.md (reduces<mode>): Renamed to ...
5903 (reduces<mode><mask_scalar_name>): ... this.
5904 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5906 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5907 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5909 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5911 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5912 * selftest-run-tests.c (selftest::run_tests): Call
5913 selftest::unique_ptr_tests_cc_tests.
5914 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5915 * unique-ptr-tests.cc: New file.
5917 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5920 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5922 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5923 (make_hard_regno_born, make_hard_regno_dead): Update
5924 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5926 2017-10-16 Jeff Law <law@redhat.com>
5928 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5930 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5932 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5934 2017-10-16 Olivier Hainque <hainque@adacore.com>
5936 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5937 with_cpu if we were configured for an e500v2 target cpu name.
5939 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5941 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5942 * doc/invoke.texi: Document +nodsp as a valid extension for
5945 2017-10-16 Martin Liska <mliska@suse.cz>
5947 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5948 (test_set_range): Likewise.
5949 (test_range_functions): Rename to ...
5950 (test_bit_in_range): ... this.
5951 (sbitmap_c_tests): Add new test.
5953 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5955 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5957 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5958 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5960 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5962 * config/aarch64/aarch64-builtins.c
5963 (aarch64_types_quadopu_lane_qualifiers): New.
5964 (TYPES_QUADOPU_LANE): New.
5965 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5966 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5967 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5968 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5969 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5970 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5971 (Vdottype, DOTPROD): New.
5972 (sur): Add SDOT and UDOT.
5974 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5976 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5977 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5978 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5980 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5981 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5982 Enable TARGET_DOTPROD.
5983 (cortex-a75.cortex-a55): Likewise.
5984 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5986 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5988 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5989 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5991 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5993 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5994 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5995 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5996 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5997 * config/arm/types.md (neon_dot, neon_dot_q): New.
5998 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6000 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6002 * config/arm/arm.h (TARGET_DOTPROD): New.
6003 * config/arm/arm.c (arm_arch_dotprod): New.
6004 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6005 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6006 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6007 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6008 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6009 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6010 * doc/invoke.texi (armv8.2-a): Document dotprod
6012 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6014 * i386.c (ix86_vec_cost): New function.
6015 (ix86_rtx_costs): Handle vector operations better.
6016 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6017 * x86-tune-costs.h: Add new costs to all tables.
6019 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6021 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6023 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6024 divsd, sqrtss and sqrtsd
6025 * x86-tune-costs.h: Add new entries to all costs.
6026 (znver1_cost): Fix to match real instruction latencies.
6028 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6029 Michael Collison <michael.collison@arm.com>
6031 * compare-elim.c: Include emit-rtl.h.
6032 (can_merge_compare_into_arith): New function.
6033 (try_validate_parallel): Likewise.
6034 (try_merge_compare): Likewise.
6035 (try_eliminate_compare): Call the above when no previous clobber
6037 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6040 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6044 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6045 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6049 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6050 to be any operand_equal_p operands. For & (B - 1) require
6051 B to be power of 2. Recognize
6052 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6054 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6057 * optabs.c (expand_memory_blockage): Fix call of
6058 targetm.have_memory_blockage.
6060 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6063 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6064 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6065 objects to extra_objs instead of overwriting it.
6067 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6069 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6070 Use any_fp_register_operand as operand[3] predicate. Simplify
6071 equality test for operands[2] and operands[4] memory location.
6072 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6073 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6074 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6075 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6076 any_fp_register_operand as operand[1] predicate. Simplify
6077 equality test for operands[0] and operands[3] memory location.
6078 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6079 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6080 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6082 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6084 * target-insns.def: Add memory_blockage.
6085 * optabs.c (expand_memory_blockage): New function.
6086 (expand_asm_memory_barrier): Rename ...
6087 (expand_asm_memory_blockage): ... to this.
6088 (expand_mem_thread_fence): Call expand_memory_blockage
6089 instead of expand_asm_memory_barrier.
6090 (expand_mem_singnal_fence): Ditto.
6091 (expand_atomic_load): Ditto.
6092 (expand_atomic_store): Ditto.
6093 * doc/md.texi (Standard Pattern Names For Generation):
6094 Document memory_blockage instruction pattern.
6096 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6098 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6099 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6100 * config/rl78/rl78.md: New define_expand "adddi3".
6102 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6104 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6107 2017-10-13 Jeff Law <law@redhat.com>
6109 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6111 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6114 * internal-fn.c (expand_mul_overflow): If both operands have
6115 the same highpart of -1 or 0 and the topmost bit of lowpart
6116 is different, overflow is if res <= 0 rather than res < 0.
6118 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6120 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6121 TARGET_P9_VECTOR code for unaligned_load case.
6123 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6125 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6127 2017-10-13 Nathan Sidwell <nathan@acm.org>
6129 * tree-core.h (tree_contains_struct): Make bool.
6130 * tree.c (tree_contains_struct): Likewise.
6131 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6132 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6133 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6134 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6135 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6137 2017-10-13 Richard Biener <rguenther@suse.de>
6139 * graphite-isl-ast-to-gimple.c
6140 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6141 parameters and dominance check.
6142 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6143 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6144 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6145 Do not update SSA form here or do intermediate IL verification.
6146 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6147 (graphite_initialize): Remove check on the number of loops in
6148 the function and inline into graphite_transform_loops.
6149 (graphite_finalize): Inline into graphite_transform_loops.
6150 (graphite_transform_loops): Perform SSA update and IL verification
6152 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6154 2017-10-13 Richard Biener <rguenther@suse.de>
6156 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6157 graphite_expression_type_precision): Avoid global constructor
6159 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6160 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6161 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6162 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6163 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6164 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6166 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6169 * config/i386/i386.h (ix86_red_zone_size): New.
6170 * config/i386/i386.md (push peephole2s): Replace
6171 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6173 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6174 Alan Hayward <alan.hayward@arm.com>
6175 David Sherwood <david.sherwood@arm.com>
6177 * combine.c (can_change_dest_mode): Reject changes in
6178 REGMODE_NATURAL_SIZE.
6180 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6181 Alan Hayward <alan.hayward@arm.com>
6182 David Sherwood <david.sherwood@arm.com>
6184 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6185 (expand_debug_source_expr): Likewise.
6186 * combine.c (combine_simplify_rtx): Likewise.
6187 * cse.c (fold_rtx): Likewise.
6188 * fwprop.c (canonicalize_address): Likewise.
6189 * targhooks.c (default_shift_truncation_mask): Likewise.
6191 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6192 Alan Hayward <alan.hayward@arm.com>
6193 David Sherwood <david.sherwood@arm.com>
6195 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6196 (widened_mode): Likewise.
6197 (expand_unop): Likewise.
6198 * ree.c (transform_ifelse): Likewise.
6199 (merge_def_and_ext): Likewise.
6200 (combine_reaching_defs): Likewise.
6201 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6203 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6204 Alan Hayward <alan.hayward@arm.com>
6205 David Sherwood <david.sherwood@arm.com>
6207 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6208 * combine.c (gen_lowpart_for_combine): Likewise.
6209 * dwarf2out.c (rtl_for_decl_location): Likewise.
6210 * final.c (alter_subreg): Likewise.
6211 * rtlhooks.c (gen_lowpart_general): Likewise.
6212 (gen_lowpart_if_possible): Likewise.
6214 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6215 Alan Hayward <alan.hayward@arm.com>
6216 David Sherwood <david.sherwood@arm.com>
6218 * calls.c (expand_call): Use subreg_lowpart_offset.
6219 * cse.c (cse_insn): Likewise.
6220 * regcprop.c (copy_value): Likewise.
6221 (copyprop_hardreg_forward_1): Likewise.
6223 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6226 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6227 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6228 =Q constraints to +Q and into insn condition add check
6229 that operands[0] and operands[1] are equal.
6230 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6231 =Q constraints to +Q and into insn condition add check
6232 that operands[0] is equal to either operands[1] or operands[2].
6235 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6236 instead of handling MINUS_EXPR twice (once for each argument),
6237 canonicalize operand order and handle just once, use rtype where
6238 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6241 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6242 any values of __C while still being pattern recognizable as a simple
6245 2017-10-13 Richard Biener <rguenther@suse.de>
6247 PR tree-optimization/82451
6249 2017-10-02 Richard Biener <rguenther@suse.de>
6251 PR tree-optimization/82355
6252 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6253 a mapping for the enclosing loop but avoid generating one for
6255 (copy_bb_and_scalar_dependences): Remove premature codegen
6256 error on PHIs in blocks duplicated into multiple places.
6257 * graphite-scop-detection.c
6258 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6259 in the region use it as loop and nest to analyze the DR in.
6260 (try_generate_gimple_bb): Likewise.
6261 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6262 (add_loop_constraints): For blocks in a loop not in the region
6263 create a dimension with a single iteration.
6264 * sese.h (gbb_loop_at_index): Remove assert.
6266 * cfgloop.c (loop_preheader_edge): For the loop tree root
6267 return the single successor of the entry block.
6268 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6269 Reset the SCEV hashtable and niters.
6270 * graphite-scop-detection.c
6271 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
6272 assert that we only have POLYNOMIAL_CHREC that vary in loops
6273 contained in the region.
6274 (scop_detection::graphite_can_represent_expr): Adjust.
6275 (scop_detection::stmt_has_simple_data_refs_p): For loops
6276 not in the region set loop to NULL. The nest is now the
6277 entry edge to the region.
6278 (try_generate_gimple_bb): Likewise.
6279 * sese.c (scalar_evolution_in_region): Adjust for
6280 instantiate_scev change.
6281 * tree-data-ref.h (graphite_find_data_references_in_stmt):
6282 Make nest parameter the edge into the region.
6283 (create_data_ref): Likewise.
6284 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
6285 entry edge into a region and adjust instantiate_scev calls.
6286 (create_data_ref): Likewise.
6287 (graphite_find_data_references_in_stmt): Likewise.
6288 (find_data_references_in_stmt): Pass the loop preheader edge
6289 from the nest argument.
6290 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
6291 parameter the edge into the region.
6292 (instantiate_parameters): Use the loop preheader edge as entry.
6293 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
6295 (get_instantiated_value_entry): Make instantiate_below parameter
6296 the edge into the region.
6297 (instantiate_scev_name): Likewise. Adjust dominance checks,
6298 when we cannot use loop-based instantiation instantiate by
6299 walking use-def chains.
6300 (instantiate_scev_poly): Adjust.
6301 (instantiate_scev_binary): Likewise.
6302 (instantiate_scev_convert): Likewise.
6303 (instantiate_scev_not): Likewise.
6304 (instantiate_array_ref): Remove.
6305 (instantiate_scev_3): Likewise.
6306 (instantiate_scev_2): Likewise.
6307 (instantiate_scev_1): Likewise.
6308 (instantiate_scev_r): Do not blindly handle N-operand trees.
6309 Do not instantiate array-refs. Handle all constants and invariants.
6310 (instantiate_scev): Make instantiate_below parameter
6311 the edge into the region.
6312 (resolve_mixers): Use the loop preheader edge for the region
6313 parameter to instantiate_scev_r.
6314 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6316 2017-10-13 Richard Biener <rguenther@suse.de>
6318 PR tree-optimization/82525
6319 * graphite-isl-ast-to-gimple.c
6320 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6322 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6323 Fail code generation when we cannot represent the isl integer.
6324 (binary_op_to_tree): Elide modulo operations that are no-ops
6325 in the type we code generate. Remove now superfluous code
6328 2017-10-13 Richard Biener <rguenther@suse.de>
6330 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6331 (scop_detection::harmful_loop_in_region): Remove premature
6332 IV type restriction.
6333 (scop_detection::graphite_can_represent_scev): We can handle
6334 pointer IVs just fine.
6336 2017-10-13 Alan Modra <amodra@gmail.com>
6338 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6339 "Clobbers and Scratch Registers". Add paragraph on
6340 alternative to clobbers for scratch registers and OpenBLAS
6343 2017-10-13 Alan Modra <amodra@gmail.com>
6345 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6346 example of a memory input for a string of known length. Move
6347 commentary out of table. Add a number of new examples
6348 covering array memory inputs.
6350 2017-10-12 Martin Liska <mliska@suse.cz>
6352 PR tree-optimization/82493
6353 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6354 (test_range_functions): New function.
6355 (sbitmap_c_tests): Likewise.
6356 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6357 * selftest.h (sbitmap_c_tests): New function.
6359 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6361 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6363 * config/rs6000/amo.h: Fix spacing issue.
6365 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6368 * config/i386/i386.md (*ashl<mode>3_mask_1,
6369 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6370 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6373 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6375 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6376 (profile_probability): Set max_probability
6377 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6379 * profile-count.c (profile_probability::differs_from_p): Do not
6380 rely on max_probaiblity == 10000
6382 2017-10-12 Jeff Law <law@redhat.com>
6384 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6387 2017-10-12 Martin Sebor <msebor@redhat.com>
6391 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6392 (handle_alias_pairs): Call it.
6393 * common.opt (-Wattribute-alias): New option.
6394 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6395 * doc/invoke.texi (-Wattribute-alias): Document.
6397 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6400 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6402 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6404 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6405 (make_hard_regno_born, make_hard_regno_dead): Update
6406 bb_killed_pseudos and bb_gen_pseudos.
6408 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6410 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6412 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6414 * config/alpha/alpha.c (alpha_split_conditional_move):
6415 Use std::swap instead of manually swapping.
6416 (alpha_stdarg_optimize_hook): Ditto.
6417 (alpha_canonicalize_comparison): Ditto.
6419 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6421 * tree-loop-distribution.c (struct builtin_info): New struct.
6422 (struct partition): Refactor fields into struct builtin_info.
6423 (partition_free): Free struct builtin_info.
6424 (build_size_arg_loc, build_addr_arg_loc): Delete.
6425 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6426 information from struct builtin_info.
6427 (find_single_drs): New function refactored from classify_partition.
6428 Also moved builtin validity checks to this function.
6429 (compute_access_range, alloc_builtin): New functions.
6430 (classify_builtin_st, classify_builtin_ldst): New functions.
6431 (classify_partition): Refactor code into functions find_single_drs,
6432 classify_builtin_st and classify_builtin_ldst.
6433 (distribute_loop): Don't do runtime alias check when distributing
6435 (find_seed_stmts_for_distribution): New function.
6436 (pass_loop_distribution::execute): Refactor code finding seed
6437 stmts into above function. Support distribution for the innermost
6438 two-level loop nest. Adjust dump information.
6440 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6442 * tree-loop-distribution.c: Adjust the general comment.
6443 (NUM_PARTITION_THRESHOLD): New macro.
6444 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6445 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6446 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6447 in call to build_partition_graph.
6448 (finalize_partitions): New parameter. Make loop distribution more
6449 conservative by fusing more partitions.
6450 (distribute_loop): Don't do runtime alias check in case of loop nest
6452 (find_seed_stmts_for_distribution): New function.
6453 (prepare_perfect_loop_nest): New function.
6454 (pass_loop_distribution::execute): Refactor code finding seed stmts
6455 and loop nest into above functions. Support loop nest distribution.
6456 Adjust dump information accordingly.
6458 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6460 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6461 and set PTYPE_SEQUENTIAL for merged partition.
6463 2017-10-12 Richard Biener <rguenther@suse.de>
6465 PR tree-optimization/69728
6467 2017-09-19 Richard Biener <rguenther@suse.de>
6469 PR tree-optimization/69728
6470 * graphite-sese-to-poly.c (schedule_error): New global.
6471 (add_loop_schedule): Handle empty domain by failing the
6473 (build_original_schedule): Handle schedule_error.
6475 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6476 domain by returning an unchanged schedule.
6478 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6480 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6481 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6483 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6485 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6488 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6491 * expr.c (store_field): Don't optimize away bitsize == 0 store
6492 from CALL_EXPR with addressable return type.
6494 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6496 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6497 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6498 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6499 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6500 TARGET_ISEL instead of TARGET_ISEL<sel>.
6502 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6504 * config/rs6000/rs6000.c
6505 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6507 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6509 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6511 (reg_or_cint_operand): Fix comment.
6512 (reg_or_zero_operand): New predicate.
6513 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6514 * config/rs6000/rs6000.c (output_isel): Delete.
6515 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6516 instead of reg_or_cint_operand. Output instruction directly (not via
6518 (isel_unsigned_<mode>): Ditto.
6519 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6520 gpc_reg_operand. Add an instruction alternative for this. Output
6521 instruction directly.
6522 (*isel_reversed_unsigned_<mode>): Ditto.
6524 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6526 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6527 (TARGET_CANONICALIZE_COMPARISON): Define.
6529 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6532 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6533 Check whether the dest is REG before adding REG_EQUIV note.
6535 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6538 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6540 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6541 (make_hard_regno_born, make_hard_regno_dead): Update
6542 bb_killed_pseudos and bb_gen_pseudos.
6544 2017-10-11 Nathan Sidwell <nathan@acm.org>
6546 * incpath.h (enum incpath_kind): Name enum, prefix values.
6547 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6548 * incpath.c (heads, tails): Use INC_MAX.
6549 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6550 (merge_include_chains, split_quote_chain,
6551 register_include_chains): Update incpath_kind names.
6552 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6553 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6555 (add_framework_path, darwin_register_objc_includes): Likewise.
6556 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6558 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6560 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6561 Do not use float_operator operator predicate.
6562 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6563 * config/i386/predicates.md (float_operator): Remove predicate.
6565 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6567 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6568 (*jccxf_i387): Ditto.
6569 (*jcc<mode>_i387): Ditto.
6570 (*jccu<mode>_i387): Ditto.
6571 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6572 (*jcc_*_i387 splitters): Remove.
6573 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6574 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6575 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6578 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6580 * profile-count.h (slow_safe_scale_64bit): New function.
6581 (safe_scale_64bit): New inline.
6582 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6583 * profile-count.c: Include wide-int.h
6584 (slow_safe_scale_64bit): New.
6586 2017-10-11 Nathan Sidwell <nathan@acm.org>
6588 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6589 HAS_DECL_ASSEMBLER_NAME_P.
6590 * gimple-expr.c (gimple_decl_printable_name: Check
6591 HAS_DECL_ASSEMBLER_NAME_P too.
6592 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6593 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6594 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6595 * passes.c (rest_of_decl_compilation): Check
6596 HAS_DECL_ASSEMBLER_NAME_P too.
6597 * recog.c (verify_changes): Likewise.
6598 * tree-pretty-print.c (dump_decl_name): Likewise.
6599 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6601 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6602 (SET_DECL_ASSEMBLER_NAME): Use it.
6603 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6604 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6605 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6607 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6609 * config.gcc (i386, x86_64): Add extra objects.
6610 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6611 (ix86_min_insn_size): Declare.
6612 (ix86_issue_rate): Declare.
6613 (ix86_adjust_cost): Declare.
6614 (ia32_multipass_dfa_lookahead): Declare.
6615 (ix86_macro_fusion_p): Declare.
6616 (ix86_macro_fusion_pair_p): Declare.
6617 (ix86_bd_has_dispatch): Declare.
6618 (ix86_bd_do_dispatch): Declare.
6619 (ix86_core2i7_init_hooks): Declare.
6620 (ix86_atom_sched_reorder): Declare.
6621 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6622 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6623 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6624 (rip_relative_addr_p): Rename to ...
6625 (ix86_rip_relative_addr_p): ... this one; export.
6626 (memory_address_length): Update.
6627 (ix86_issue_rate): Move to x86-tune-sched.c.
6628 (ix86_flags_dependent): Move to x86-tune-sched.c.
6629 (ix86_agi_dependent): Move to x86-tune-sched.c.
6630 (exact_dependency_1): Move to x86-tune-sched.c.
6631 (exact_store_load_dependency): Move to x86-tune-sched.c.
6632 (ix86_adjust_cost): Move to x86-tune-sched.c.
6633 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6634 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6635 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6636 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6637 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6638 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6639 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6640 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6641 (min_insn_size): Rename to ...
6642 (ix86_min_insn_size): ... this one; export.
6643 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6644 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6645 (core2i7_first_cycle_multipass_backtrack): Move to
6646 x86-tune-sched-core.c.
6647 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6648 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6649 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6650 (ix86_avoid_jump_mispredicts): Update.
6651 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6652 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6653 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6654 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6655 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6656 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6657 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6658 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6659 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6660 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6661 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6662 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6663 (BIG): Move to ix86-tune-sched-bd.c.
6664 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6665 (enum insn_path): Move to ix86-tune-sched-bd.c.
6666 (get_mem_group): Move to ix86-tune-sched-bd.c.
6667 (is_cmp): Move to ix86-tune-sched-bd.c.
6668 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6669 (is_branch): Move to ix86-tune-sched-bd.c.
6670 (is_prefetch): Move to ix86-tune-sched-bd.c.
6671 (init_window): Move to ix86-tune-sched-bd.c.
6672 (allocate_window): Move to ix86-tune-sched-bd.c.
6673 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6674 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6675 (process_end_window): Move to ix86-tune-sched-bd.c.
6676 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6677 (find_constant): Move to ix86-tune-sched-bd.c.
6678 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6679 (has_immediate): Move to ix86-tune-sched-bd.c.
6680 (get_insn_path): Move to ix86-tune-sched-bd.c.
6681 (get_insn_group): Move to ix86-tune-sched-bd.c.
6682 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6683 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6684 (add_insn_window): Move to ix86-tune-sched-bd.c.
6685 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6686 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6687 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6688 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6689 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6690 (do_dispatch): Move to ix86-tune-sched-bd.c.
6691 (has_dispatch): Move to ix86-tune-sched-bd.c.
6692 * config/i386/t-i386: Add new object files.
6693 * config/i386/x86-tune-costs.h: New file.
6694 * config/i386/x86-tune-sched-atom.c: New file.
6695 * config/i386/x86-tune-sched-bd.c: New file.
6696 * config/i386/x86-tune-sched-core.c: New file.
6697 * config/i386/x86-tune-sched.c: New file.
6699 2017-10-11 Liu Hao <lh_mouse@126.com>
6701 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6702 the generic version below instead.
6703 (should_colorize): Recognize Windows consoles as terminals
6705 * pretty-print.c [__MINGW32__] (write_all): New function.
6706 [__MINGW32__] (find_esc_head): Likewise.
6707 [__MINGW32__] (find_esc_terminator): Likewise.
6708 [__MINGW32__] (eat_esc_sequence): Likewise.
6709 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6711 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6714 2017-10-11 Richard Biener <rguenther@suse.de>
6716 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6717 Properly call analyze_scalar_evolution with the loop of the stmt.
6719 2017-10-11 Richard Biener <rguenther@suse.de>
6721 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6722 * tree-core.h (tree_base): Add chrec_var union member.
6723 * tree.h (CHREC_VAR): Remove.
6724 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6725 * tree-chrec.h (build_polynomial_chrec): Adjust.
6726 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6727 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6729 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6731 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6732 * match.pd: ... here.
6733 ((T) X == (T) Y): Relax condition.
6735 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6737 PR tree-optimization/82472
6738 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6740 (break_alias_scc_partitions): Update postorder number.
6742 2017-10-11 Martin Liska <mliska@suse.cz>
6745 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6747 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6749 2017-10-11 Martin Liska <mliska@suse.cz>
6751 * print-rtl.c (print_insn): Move declaration of idbuf
6752 to same scope as name.
6754 2017-10-11 Martin Liska <mliska@suse.cz>
6759 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6762 2017-10-11 Martin Liska <mliska@suse.cz>
6765 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6768 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6770 * config.gcc (default_gnu_indirect_function): Default to yes for
6771 arm*-*-linux* with glibc.
6773 2017-10-11 Richard Biener <rguenther@suse.de>
6775 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6776 default-defs and types we do not want to analyze.
6777 (interpret_loop_phi): Replace unreachable code with an assert.
6778 (compute_scalar_evolution_in_loop): Remove and inline ...
6779 (analyze_scalar_evolution_1): ... here, replacing condition with
6780 what makes the intent clearer. Remove handling of cases
6781 get_scalar_evolution now handles.
6783 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6785 PR rtl-optimization/81434
6786 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6787 comment for main loop. In sched_group_found if, also add checks for
6788 pass and min_cost_group.
6790 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6792 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6793 (rs6000_insn_cost): New function.
6794 * config/rs6000/rs6000.md (cost): New attribute.
6796 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6797 H.J. Lu <hongjiu.lu@intel.com>
6801 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6802 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6803 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6804 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6805 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6806 OPTION_MASK_ISA_MMX and some other bit set require both
6807 mmx and the ISAs without the mmx bit.
6808 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6809 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6810 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6811 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6812 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6813 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6814 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6815 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6816 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6817 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6818 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6819 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6820 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6821 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6822 Add OPTION_MASK_ISA_MMX.
6824 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6826 * config.gcc (armv7*-*-freebsd*): New target.
6827 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6829 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6831 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6832 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6835 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6837 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6838 a template parameter.
6839 (WIDE_INT_REF_FOR): Update accordingly.
6840 * tree.h (wi::int_traits <const_tree>): Delete.
6841 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6842 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6843 (wi::tree_to_wide_ref): New typedef.
6844 (wi::to_wide): New function.
6845 * calls.c (get_size_range): Use wi::to_wide when operating on
6847 * cgraph.c (cgraph_node::create_thunk): Likewise.
6848 * config/i386/i386.c (ix86_data_alignment): Likewise.
6849 (ix86_local_alignment): Likewise.
6850 * dbxout.c (stabstr_O): Likewise.
6851 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6852 * expr.c (const_vector_from_tree): Likewise.
6853 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6854 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6855 (fold_negate_expr_1, int_const_binop_1, const_binop)
6856 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6857 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6858 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6859 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6860 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6861 (fold_not_const, round_up_loc): Likewise.
6862 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6863 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6864 (alloca_call_type): Likewise.
6865 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6866 * godump.c (go_output_typedef): Likewise.
6867 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6868 * internal-fn.c (get_min_precision): Likewise.
6869 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6870 * ipa-polymorphic-call.c
6871 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6872 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6873 (ipa_modify_call_arguments): Likewise.
6874 * match.pd: Likewise.
6875 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6876 * print-tree.c (print_node_brief, print_node): Likewise.
6877 * stmt.c (expand_case): Likewise.
6878 * stor-layout.c (layout_type): Likewise.
6879 * tree-affine.c (tree_to_aff_combination): Likewise.
6880 * tree-cfg.c (group_case_labels_stmt): Likewise.
6881 * tree-data-ref.c (dr_analyze_indices): Likewise.
6882 (prune_runtime_alias_test_list): Likewise.
6883 * tree-dump.c (dequeue_and_dump): Likewise.
6884 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6885 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6886 * tree-pretty-print.c (dump_generic_node): Likewise.
6887 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6888 (simple_iv_with_niters): Likewise.
6889 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6890 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6891 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6892 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6893 (refine_value_range_using_guard, number_of_iterations_ne_max)
6894 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6895 (get_cst_init_from_scev, record_nonwrapping_iv)
6896 (scev_var_range_cant_overflow): Likewise.
6897 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6898 * tree-ssa-pre.c (compute_avail): Likewise.
6899 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6900 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6901 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6902 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6903 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6904 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6905 (dump_case_nodes, try_switch_expansion): Likewise.
6906 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6907 (vect_do_peeling): Likewise.
6908 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6909 * tree-vect-stmts.c (vectorizable_load): Likewise.
6910 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6911 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6912 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6913 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6914 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6915 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6916 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6917 (evrp_dom_walker::before_dom_children): Likewise.
6918 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6919 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6920 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6921 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6922 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6923 (get_range_pos_neg): Likewise.
6924 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6925 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6926 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6927 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6928 * config/avr/avr.c (avr_fold_builtin): Likewise.
6929 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6930 * config/msp430/msp430.c (msp430_attr): Likewise.
6931 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6932 * config/powerpcspe/powerpcspe-c.c
6933 (altivec_resolve_overloaded_builtin): Likewise.
6934 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6935 (rs6000_expand_ternop_builtin): Likewise.
6936 * config/rs6000/rs6000-c.c
6937 (altivec_resolve_overloaded_builtin): Likewise.
6938 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6939 (rs6000_expand_ternop_builtin): Likewise.
6940 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6942 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6944 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6945 when copying loop nest with only one inner loop.
6947 2017-10-10 Richard Biener <rguenther@suse.de>
6949 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6950 blocks if SCEV is active.
6951 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6953 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6954 (scev_initialize): Assert we are not yet initialized.
6956 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6958 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6959 inner loop's exit stmt by making it always exit the loop, otherwise
6960 we would generate an infinite empty loop.
6962 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6964 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6965 renaming variables in new preheader if it's deleted.
6967 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6969 * tree-loop-distribution.c (struct partition): Remove unused field
6970 loops of the structure.
6971 (partition_alloc, partition_free): Ditto.
6972 (build_rdg_partition_for_vertex): Ditto.
6974 2017-10-09 Jeff Law <law@redhat.com>
6976 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6977 return type to match prototype and documentation.
6979 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6981 * config/rs6000/rs6000.c (processor_costs): Move to ...
6982 * config/rs6000/rs6000.h: ... here.
6983 (rs6000_cost): Declare.
6985 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6987 * except.c (setjmp_fn): New global variable.
6988 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6989 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6990 if DONT_USE_BUILTIN_SETJMP is defined.
6992 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6994 * target.def (insn_cost): New hook.
6995 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6996 * doc/tm.texi: Regenerate.
6997 * rtlanal.c (insn_cost): Use the new hook.
6999 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7001 * combine.c (combine_validate_cost): Compute the new insn_cost,
7002 not just pattern_cost.
7003 (try_combine): Adjust comment.
7005 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7007 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7009 * combine.c (uid_insn_cost): Adjust comment.
7010 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7012 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7013 * dse.c (find_shift_sequence): Ditto.
7014 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7015 (bb_valid_for_noce_process_p): Use pattern_cost.
7016 * rtl.h (insn_rtx_cost): Delete.
7017 (pattern_cost): New prototype.
7018 (insn_cost): New prototype.
7019 * rtlanal.c (insn_rtx_cost): Rename to...
7020 (pattern_cost): ... this.
7023 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7025 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7026 (*jcc<mode>_0_r_i387): Ditto.
7027 (*jccxf_r_i387): Ditto.
7028 (*jcc<mode>_r_i387): Ditto.
7029 (*jccu<mode>_r_i387): Ditto.
7030 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7031 (*jcc): Rename from *jcc_1.
7033 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7035 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7036 deferred rescans after the lvx/stvx recombination pre-pass.
7038 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7040 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7041 memory operation instruction support.
7042 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7043 (rs6000-ibm-aix[789]*): Likewise.
7044 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7045 Document new functions.
7047 2017-10-09 Richard Biener <rguenther@suse.de>
7049 PR tree-optimization/82397
7050 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7051 equality only for semantically equal trees.
7053 2017-10-09 Richard Biener <rguenther@suse.de>
7055 PR tree-optimization/82449
7056 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7057 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7058 allow constant addresses.
7059 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7062 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7064 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7067 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7070 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7073 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7076 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7078 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7081 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7082 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7084 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7086 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7087 (WI_BINARY_PREDICATE_RESULT): Likewise.
7088 (wi::binary_traits::operator_result): New type.
7089 (wi::binary_traits::predicate_result): Likewise.
7090 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7091 (generic_wide_int::operator==, generic_wide_int::operator!=)
7092 (generic_wide_int::operator&, generic_wide_int::and_not)
7093 (generic_wide_int::operator|, generic_wide_int::or_not)
7094 (generic_wide_int::operator^, generic_wide_int::operator+
7095 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7097 (operator~, unary operator-, operator==, operator!=, operator&)
7098 (operator|, operator^, operator+, binary operator-, operator*): New
7100 * expr.c (get_inner_reference): Use wi::bit_and_not.
7101 * fold-const.c (fold_binary_loc): Likewise.
7102 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7103 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7104 (bit_value_binop): Likewise.
7105 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7106 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7107 (extract_range_from_binary_expr_1): Likewise.
7108 (masked_increment): Likewise.
7109 (simplify_bit_ops_using_ranges): Likewise.
7111 2017-10-09 Martin Jambor <mjambor@suse.cz>
7114 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7115 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7116 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7118 (hsa_fixup_mov_insn_type): New function.
7119 (hsa_op_with_type::get_in_type): Use it.
7120 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7121 immediates in an assert.
7122 (hsa_op_with_type::extend_int_to_32bit): New method.
7123 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7124 types. Convert to dest type if necessary.
7125 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7126 (reg_for_gimple_ssa): Pass false as min32int to
7127 hsa_type_for_scalar_tree_type.
7128 (gen_hsa_addr): Fixup type when creating addresable temporary.
7129 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7130 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7131 necessary. Call hsa_fixup_mov_insn_type.
7132 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7133 extend operands and convert to dest type if necessary.
7134 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7135 to dest type if necessary.
7136 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7137 if conversion nt necessary and size matches.
7138 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7139 to dest type if necessary.
7140 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7141 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7143 (gen_hsa_clrsb): Likewise.
7144 (gen_hsa_ffs): Likewise.
7145 (gen_hsa_divmod): Extend operands and convert to dest type if
7147 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7149 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7151 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7152 Remove empty default arguments. Use a brace block as output
7154 (conditional return): Ditto.
7156 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7157 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7159 (group_ending_nop): Ditto.
7160 (doloop_end): Ditto.
7161 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7162 (splitters for those): Ditto.
7164 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7166 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7167 a conditional jump (and the compare for it) so that pc_rtx is the
7169 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7170 for the deleted and renamed ctr<mode>_internal[234] patterns.
7171 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7172 Delete second conditional return pattern.
7173 (ctr<mode>_internal2): Delete this second bdnz pattern.
7174 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7175 (ctr<mode>_internal4): Delete this second bdz pattern.
7177 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7179 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7180 (always_initialized_rtx_for_ssa_name_p): New predicate.
7181 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7182 (finish_out_of_ssa): Free new field of SA.
7183 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7184 * tree-ssa-coalesce.c: Include tree-ssa.h.
7185 (get_parm_default_def_partitions): Remove extern keyword.
7186 (get_undefined_value_partitions): New function.
7187 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7188 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7190 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7192 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7194 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7196 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7198 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7199 for targets that preffer 128bit.
7201 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7203 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7205 2017-10-08 Olivier Hainque <hainque@adacore.com>
7207 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7208 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7210 (thumb_set_return_address): Likewise.
7212 2017-10-08 Olivier Hainque <hainque@adacore.com>
7214 * common/config/arm/arm-common.c (arm_except_unwind_info):
7215 Handle DWARF2_UNWIND_INFO.
7217 2017-10-07 Michael Collison <michael.collison@arm.com>
7219 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7222 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7224 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7225 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7226 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7227 defined, force the creation of a new block for a dispatch label.
7229 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7231 * invoke.texi (Wsuggest-attribute=cold): Document.
7232 * common.opt (Wsuggest-attribute=cold): New
7233 * ipa-pure-const.c (warn_function_cold): New function.
7234 * predict.c (compute_function_frequency): Use it.
7235 * predict.h (warn_function_cold): Declare.
7237 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7239 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7242 2017-10-06 Martin Liska <mliska@suse.cz>
7244 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7245 keyword for member functions.
7246 (struct sanopt_tree_couple): New struct.
7247 (struct sanopt_tree_couple_hash): New function.
7248 (struct sanopt_ctx): Add new hash_map.
7249 (has_dominating_ubsan_ptr_check): New function.
7250 (record_ubsan_ptr_check_stmt): Likewise.
7251 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7252 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7253 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7255 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7258 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7259 aarch64_simd_valid_immediate on CONST_VECTORs.
7260 (aarch64_reg_or_bic_imm): Likewise.
7262 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7264 PR rtl-optimization/82396
7265 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7267 2017-10-06 Sebastian Pop <sebpop@gmail.com>
7269 * graphite-dependences.c (scop_get_reads): Move code to...
7270 (scop_get_must_writes): Move code to...
7271 (scop_get_may_writes): Move code to...
7272 (scop_get_reads_and_writes): ... here.
7273 (scop_get_dependences): Call scop_get_reads_and_writes.
7275 2017-10-06 Jakub Jelinek <jakub@redhat.com>
7277 PR tree-optimization/82434
7278 * fold-const.h (can_native_encode_type_p,
7279 can_native_encode_string_p): Remove.
7280 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
7281 don't encode anything, just return what would be otherwise returned.
7282 (native_encode_fixed, native_encode_complex, native_encode_vector):
7284 (native_encode_string): Likewise. Inline by hand
7285 can_native_encode_string_p.
7286 (can_native_encode_type_p): Remove.
7287 (can_native_encode_string_p): Remove.
7288 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
7289 STRING_CSTs using can_native_encode_string_p, test all
7290 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
7291 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
7292 argument from native_encode_expr.
7293 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
7294 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
7297 2017-10-06 Richard Biener <rguenther@suse.de>
7299 PR tree-optimization/82397
7300 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
7301 operand_equal_p but rely on data_ref_compare_tree for detecting
7303 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7304 to match up with dr_group_sort_cmp.
7306 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7309 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7311 * config/s390/s390-builtin-types.def: Regenerate.
7313 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7316 * config/s390/s390-builtin-types.def: Regenerate.
7317 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7318 Change flag from B_VXE to B_VX.
7319 (s390_vec_min_dbl): Remove B_VXE flag.
7321 2017-10-06 Richard Biener <rguenther@suse.de>
7323 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7324 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7325 translate_isl_ast_to_gimple::is_valid_rename,
7326 translate_isl_ast_to_gimple::get_rename,
7327 translate_isl_ast_to_gimple::get_def_bb_for_const,
7328 translate_isl_ast_to_gimple::get_new_name,
7329 translate_isl_ast_to_gimple::collect_all_ssa_names,
7330 translate_isl_ast_to_gimple::copy_loop_phi_args,
7331 translate_isl_ast_to_gimple::collect_all_ssa_names,
7332 translate_isl_ast_to_gimple::copy_loop_phi_args,
7333 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7334 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7335 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7336 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7337 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7338 translate_isl_ast_to_gimple::copy_cond_phi_args,
7339 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7340 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7341 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7342 translate_isl_ast_to_gimple::rename_uses,
7343 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7344 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7345 (set_rename_for_each_def): Likewise.
7346 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7347 here. Handle rewriting SCEV analyzable uses here.
7348 (copy_bb_and_scalar_dependences): Generate code for PHI
7350 (graphite_regenerate_ast_isl): Adjust.
7351 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7352 (add_write, add_read): New functions.
7353 (build_cross_bb_scalars_def): Use it and simplify.
7354 (build_cross_bb_scalars_use): Likewise.
7355 (graphite_find_cross_bb_scalar_vars): Inline into...
7356 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7357 simulating out-of-SSA. Compute liveout and add dependencies.
7358 (build_scops): Force an empty entry block.
7359 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7361 (sese_build_liveouts): Declare.
7362 (sese_trivially_empty_bb_p): Likewise.
7363 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7364 compute liveout and debug_liveout.
7365 (sese_bad_liveouts_use): Remove.
7366 (sese_reset_debug_liveouts_bb): Likewise.
7367 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7368 (sese_build_liveouts): Build liveout and debug_liveout and store
7370 (new_sese_info): Adjust.
7371 (free_sese_info): Likewise.
7372 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7373 do not build liveout here.
7374 (move_sese_in_condition): Adjust region entry.
7375 (scev_analyzable_p): Match up with chrec_apply requirements.
7376 (sese_trivially_empty_bb_p): New.
7377 * tree-into-ssa.c (get_reaching_def): Properly support generating
7378 default-defs for incremental rewrite of anonymous names.
7380 2017-10-06 Richard Biener <rguenther@suse.de>
7382 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7383 precision do not perform modulo reduction.
7385 2017-10-06 Richard Biener <rguenther@suse.de>
7387 PR tree-optimization/82436
7388 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7389 conservatively choose the vectorization factor when checking
7390 whether we can perform the required load permutation.
7391 (vect_transform_slp_perm_load): Assert when we may not fail.
7393 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7395 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7396 message for incompatible -msdata=* and -mcall-* options.
7398 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7400 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7401 rate for post-reload scheduling.
7403 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7405 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7407 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7409 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7410 to improve monte carlo in scimark.
7412 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7414 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7415 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7416 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7417 pentium4_cost, nocona_cost): Set reassociation width to 1.
7418 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7419 width to 2 for fp operations and 1 otherwise.
7420 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7422 (atom_cost): Set reassociation width to 2.
7423 (slm_cost, generic_cost): Set fp reassociation width
7424 to 2 and 1 otherwise.
7425 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7426 (core_cost): Set fp reassociation width to 4 and vector to 2.
7427 (ix86_reassociation_width): Rewrite using cost table; special case
7428 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7429 and TARGET_AVX128_OPTIMAL.
7430 * config/i386/i386.h (processor_costs): Add
7431 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7432 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7433 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7434 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7435 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7436 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7438 2017-10-05 Nathan Sidwell <nathan@acm.org>
7440 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7442 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7444 * config/arm/arm.c (arm_test_fpu_data): New.
7445 (arm_run_selftests): Call arm_test_fpu_data.
7447 2017-10-04 Nathan Sidwell <nathan@acm.org>
7449 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7451 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7453 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7455 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7456 check type for aarch64_simd_valid_immediate.
7457 (aarch64_output_simd_mov_immediate): Update prototype.
7458 (aarch64_simd_valid_immediate): Update prototype.
7459 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7460 support for ORR-immediate.
7461 (and<mode>3): modified pattern to add support for BIC-immediate.
7462 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7463 now checks for valid immediate for BIC and ORR based on new enum
7465 (aarch64_output_simd_mov_immediate): Function now used to output
7466 BIC/ORR imm as well based on new enum argument.
7467 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7469 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7470 (aarch64_reg_or_bic_imm): Likewise.
7472 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7474 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7475 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7477 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7481 PR rtl-optimization/82396
7482 * haifa-sched.c (autopref_multipass_init): Simplify
7484 (autopref_rank_data): Simplify sort order.
7485 * sched-int.h (autopref_multipass_data_): Remove
7486 multi_mem_insn_p, min_offset and max_offset.
7488 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7490 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7492 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7494 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7497 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7499 * doc/sourcebuild.texi: Document vect_long_mult.
7501 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7503 PR tree-optimization/82413
7504 * fold-const.c (build_range_check): Use widest_int when comparing
7505 the maximum ETYPE value with HIGH.
7507 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7509 PR rtl-optimization/82396
7510 * haifa-sched.c (autopref_multipass_init): Simplify
7512 (autopref_rank_data): Simplify sort order.
7513 * sched-int.h (autopref_multipass_data_): Remove
7514 multi_mem_insn_p, min_offset and max_offset.
7516 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7518 PR tree-optimization/82381
7519 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7520 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7523 PR tree-optimization/82374
7524 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7525 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7526 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7527 current_function_decl to the new decl.
7529 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7531 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7532 helper macro for IEEE float128 hardware built-in functions.
7533 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7535 (TRUNCF128_ODD): Likewise.
7536 (ADDF128_ODD): Likewise.
7537 (SUBF128_ODD): Likewise.
7538 (MULF128_ODD): Likewise.
7539 (DIVF128_ODD): Likewise.
7540 (FMAF128_ODD): Likewise.
7541 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7542 UNSPEC_TRUNC_ROUND_TO_ODD.
7543 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7544 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7545 floating point round to odd instructions.
7546 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7547 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7548 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7549 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7550 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7551 (trunc<mode>sf2_hw): Change the truncate with round to odd
7552 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7553 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7554 to odd hardware instructions.
7555 (sub<mode>3_odd): Likewise.
7556 (mul<mode>3_odd): Likewise.
7557 (div<mode>3_odd): Likewise.
7558 (sqrt<mode>2_odd): Likewise.
7559 (fma<mode>4_odd): Likewise.
7560 (fms<mode>4_odd): Likewise.
7561 (nfma<mode>4_odd): Likewise.
7562 (nfms<mode>4_odd): Likewise.
7563 (trunc<mode>df2_odd): Change the truncate with round to odd
7564 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7566 * doc/extend.texi (PowerPC built-in functions): Update documentation
7567 for existing IEEE float128-bit built-in functions. Add built-in
7568 functions that generate the IEEE 128-bit floating point round to
7571 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7573 PR rtl-optimization/77729
7574 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7575 to (X&(C1&~C2))|C2 transformations.
7577 2017-10-03 Martin Jambor <mjambor@suse.cz>
7579 PR tree-optimization/82363
7580 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7581 mismatch, mark lacc written regardless of racc.
7583 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7585 PR tree-optimization/82381
7586 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7587 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7588 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7589 fallthrough into reassoc_stmt_dominates_stmt_p.
7592 * combine.c (combine_instructions): Don't combine in unreachable
7595 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7598 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7599 function to not use the have_cpu variable. Do not set cpu_index,
7600 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7602 (rs6000_valid_attribute_p): Remove duplicate initializations of
7603 old_optimize and func_optimize.
7604 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7605 (rs6000_activate_target_options): Make global.
7606 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7609 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7611 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7612 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7613 Set *poffset to 0 and *psize and *pmax_size to -1 if
7614 *poffset + *psize overflows in HOST_WIDE_INT.
7616 PR tree-optimization/82387
7617 PR tree-optimization/82388
7618 PR tree-optimization/82389
7619 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7620 instead of live_bytes non-NULL.
7622 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7625 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7628 2017-10-02 Richard Biener <rguenther@suse.de>
7630 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7631 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7632 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7634 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7636 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7637 requested precision matches the type's.
7638 * calls.c (alloc_max_size): Calculate the new candidate size as
7639 a widest_int and use wi::to_widest when comparing it with the
7640 current candidate size.
7641 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7642 zero rather than integer_zero_node.
7643 * match.pd: Check for a no-op conversion before using wi::add
7644 rather than after. Use tree_to_uhwi when summing small shift
7645 counts into an unsigned int.
7647 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7648 Alan Hayward <alan.hayward@arm.com>
7649 David Sherwood <david.sherwood@arm.com>
7652 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7653 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7654 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7655 POINTER_AND_FP_REGS.
7657 2017-10-02 Richard Biener <rguenther@suse.de>
7659 PR tree-optimization/82355
7660 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7661 a mapping for the enclosing loop but avoid generating one for
7663 (copy_bb_and_scalar_dependences): Remove premature codegen
7664 error on PHIs in blocks duplicated into multiple places.
7665 * graphite-scop-detection.c
7666 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7667 in the region use it as loop and nest to analyze the DR in.
7668 (try_generate_gimple_bb): Likewise.
7669 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7670 (add_loop_constraints): For blocks in a loop not in the region
7671 create a dimension with a single iteration.
7672 * sese.h (gbb_loop_at_index): Remove assert.
7674 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7676 * omp-expand.c (adjust_context_scope): New function.
7677 (expand_parallel_call): Call adjust_context_scope.
7679 2017-10-01 Jeff Law <law@redhat.com>
7681 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7682 dom_opt_dom_walker class with direct access to private members.
7683 Add comments. Call test_for_singularity.
7684 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7685 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7686 m_dummy_cond anymore.
7687 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7689 (pass_dominator:execute): Build the dummy_cond here and pass it
7690 to the dom_opt_dom_walker ctor.
7691 (test_for_singularity): New function.
7693 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7694 Maya Rashish <coypu@sdf.org>
7696 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7697 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7698 (alpha*-*-netbsd*) Use nbsd_tm_file.
7699 (arm*-*-netbsdelf*) Likewise.
7700 (i[34567]86-*-netbsdelf*) Likewise.
7701 (x86_64-*-netbsd*) Likewise.
7702 (mips*-*-netbsd*) Likewise.
7703 (powerpc-*-netbsd*) Likewise.
7704 (sh*-*-netbsd*) Likewise.
7705 (sparc-*-netbsdelf*) Likewise.
7706 (sparc64-*-netbsd*) Likewise.
7707 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7709 (vax-*-netbsdelf*) Likewise.
7710 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7711 (UINT_FAST8_TYPE) Likewise.
7712 (INT_FAST16_TYPE) Check CHAR_FAST16.
7713 (UINT_FAST16_TYPE) Likewise.
7715 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7718 * config/i386/i386.md
7719 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7720 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7721 *divmodsi4_zext_2): New define_insn_and_split.
7722 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7723 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7724 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7725 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7726 New define_insn_and_split.
7727 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7728 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7729 operands[1] having DImode when mode is SImode.
7731 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7732 always SImode for DIV and MOD in REG_EQUAL notes.
7734 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7737 * match.pd: Fix handling of NaNs in pattern.
7739 2017-09-29 Jeff Law <law@redhat.com>
7741 * sbitmap.c (bitmap_bit_in_range_p): New function.
7742 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7743 * tree-ssa-dse.c (live_bytes_read): New function.
7744 (dse_classify_store): Ignore reads of dead bytes.
7746 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7747 typos and whitespace errors.
7748 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7749 * config/s390/s390.c (s390_emit_prologue): Likewise.
7751 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7754 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7755 with a symbol for LRA.
7757 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7759 PR rtl-optimization/82338
7760 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7762 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7764 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7765 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7766 (qsort_chk): Declare.
7767 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7768 (qsort_chk): New function.
7770 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7772 PR tree-optimization/82337
7773 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7774 phi definition if the PHI result appears in an abnormal PHI.
7775 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7776 basis appears in an abnormal PHI.
7778 2017-09-29 Richard Biener <rguenther@suse.de>
7780 * graphite-isl-ast-to-gimple.c
7781 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7782 (binary_op_to_tree): Use it.
7783 (get_rename_from_scev): Likewise.
7784 (copy_loop_phi_nodes): Likewise.
7785 (copy_bb_and_scalar_dependences): Likewise.
7786 (translate_pending_phi_nodes): Likewise.
7788 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7791 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7792 for movabsq $(i32 << shift), r64.
7794 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7796 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7797 index when encoding %esp as %rsp to avoid 0x67 prefix.
7799 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7801 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7802 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7804 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7806 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7807 Extensions with more than 16 double VFP registers.
7808 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7809 to_clear_mask and all code related to it. Replace the remaining
7810 entry by a sbitmap and adapt code accordingly.
7812 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7814 * brig-builtins.def: Change pure attributes to const.
7816 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7818 * config.gcc (default_gnu_indirect_function): Default to yes for
7819 sparc*-*-linux* with glibc.
7821 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7823 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7824 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7825 when creating .init_array and .fini_array sections with priority
7828 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7831 * config/aarch64/aarch64.c
7832 (aarch64_builtin_support_vector_misalignment): Always return false
7833 when misalignment is unknown.
7835 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7837 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7838 this function to return false if the definition used by the swap
7839 instruction is artificial, or if the memory address from which the
7840 constant value is loaded is not represented by a base address held
7841 in a register or if the base address register is a frame or stack
7842 pointer. Additionally, return false if the base address of the
7843 loaded constant is a SYMBOL_REF but is not considered to be a
7845 (replace_swapped_load_constant): New function.
7846 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7847 loaded constant vector with a load of a swapped constant vector.
7849 2017-09-27 Carl Love <cel@us.ibm.com>
7851 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7852 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7853 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7856 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7858 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7859 first, always call autopref_rank_data otherwise.
7861 2017-09-27 Richard Biener <rguenther@suse.de>
7863 * graphite-scop-detection.c (find_scop_parameters): Move
7864 loop bound handling ...
7865 (gather_bbs::before_dom_children): ... here, avoiding the need
7866 to build scop_info->loop_nest.
7867 (record_loop_in_sese): Remove.
7868 * sese.h (sese_info_t::loop_nest): Remove.
7869 * sese.c (new_sese_info): Do not allocate loop_nest.
7870 (free_sese_info): Do not free loop_nest.
7872 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7875 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7876 lhs from calls if the lhs has addressable type.
7878 2017-09-27 Richard Biener <rguenther@suse.de>
7880 * graphite.h (scop::max_alias_set): New member.
7881 * graphite-scop-detection.c: Remove references to non-existing
7882 --param in comments.
7883 (build_alias_sets): Record the maximum alias set used for drs.
7884 (build_scops): Support zero as unlimited for
7885 --param graphite-max-arrays-per-scop.
7886 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7888 (build_poly_sr_1): ... here. Compute alias set based on the
7889 maximum alias set used for drs rather than
7890 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7892 2017-09-27 Richard Biener <rguenther@suse.de>
7894 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7895 --param loop-block-tile-size=0 to disable tiling.
7897 2017-09-27 Richard Biener <rguenther@suse.de>
7899 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7900 (graphite-max-nb-scop-params): Document special value zero.
7901 * domwalk.h (dom_walker::STOP): New symbolical constant.
7902 (dom_walker::dom_walker): Add optional parameter for bb to
7904 (dom_walker::~dom_walker): Declare.
7905 (dom_walker::before_dom_children): Document STOP return value.
7906 (dom_walker::m_user_bb_to_rpo): New member.
7907 (dom_walker::m_bb_to_rpo): Likewise.
7908 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7909 mapping here if not provided by the user.
7910 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7911 provided by the user.
7912 (dom_walker::STOP): Define.
7913 (dom_walker::walk): Do not compute bb to RPO mapping here.
7914 Support STOP return value from before_dom_children to stop
7916 * graphite-optimize-isl.c (optimize_isl): If the schedule
7917 is the same still generate code if -fgraphite-identity
7918 or -floop-parallelize-all are given.
7919 * graphite-scop-detection.c: Include cfganal.h.
7920 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7922 (gather_bbs::before_dom_children): Return STOP for BBs
7924 (build_scops): Compute bb to RPO mapping and pass it to
7925 the domwalk. Treat --param graphite-max-nb-scop-params=0
7926 as not limiting the number of params.
7927 * graphite.c (graphite_initialize): Remove limit on the
7928 number of basic-blocks in a function.
7929 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7930 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7931 default value of 10.
7933 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7935 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7936 Adjust code to eliminate needing to do the shift right 32-bits
7937 operation after XSCVDPSPN.
7939 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7941 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7942 ((X / Y) != 0 -> X >= Y): Likewise.
7944 2017-09-26 Carl Love <cel@us.ibm.com>
7946 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7947 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7948 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7949 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7950 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7951 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7952 definitions and overloading.
7953 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7954 statement for P9V_BUILTIN_XST_LEN_R.
7955 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7956 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7957 define_expand and define_insn for the instructions and builtins.
7958 * doc/extend.texi: Update the built-in documentation file for the new
7960 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7961 define_insn for the instructions
7963 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7966 * gcc/config/netbsd-protos.h: New file.
7967 * gcc/config/netbsd.c: New file.
7968 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7969 * gcc/config/t-netbsd: New file.
7970 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7971 (tmake_file) Add t-netbsd.
7972 (extra_objs) Add netbsd.o.
7974 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7978 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7980 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7982 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7983 sign extension from a vector register to a GPR by doing a 32-bit
7984 direct move and then an EXTSW.
7985 (extendsi<mode>2 splitter): Likewise.
7986 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7987 right or vector extract after doing XSCVDPSPN. Use
7988 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7990 (movdi_from_sf_zero_ext): Likewise.
7991 (reload_gpr_from_vsxsf): Likewise.
7992 (p8_mfvsrd_4_disf): Delete, no longer used.
7993 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7994 then needing to move the SFmode to a GPR to use the XSCVDPSP
7995 instruction instead of FRSP and XSCVDPSPN.
7996 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7997 it is adjacent to the other XSCVSPDP insns.
7998 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7999 SFmode to be in traditional Altivec registers.
8000 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8001 (vsx_xscvspdpn): Likewise.
8002 (vsx_xscvspdpn_scalar): Likewise.
8004 2017-09-26 Martin Jambor <mjambor@suse.cz>
8006 * tree-sra.c (compare_access_positions): Put integral types first,
8007 stabilize sorting of integral types, remove conditions putting
8008 non-full-precision integers last.
8009 (sort_and_splice_var_accesses): Disable scalarization if a
8010 non-integert would be represented by a non-full-precision integer.
8012 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8014 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8015 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8016 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8017 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8018 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8019 conditionals inside the function instead of around it. Call
8020 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8021 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8023 2017-09-26 Richard Biener <rguenther@suse.de>
8025 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8027 (scop_detection::build_scop_breadth): ... this. Removed.
8028 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8029 (scop_detection::harmful_stmt_in_bb): Likewise.
8030 (scop_detection::graphite_can_represent_stmt): Likewise.
8031 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8032 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8033 (scop_detection::can_represent_loop_1): ... this. Removed.
8034 (scop_detection::harmful_loop_in_region): Simplify after inlining
8035 the above and remove more quadraticness.
8036 (build_scops): Adjust.
8037 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8040 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8043 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8044 REGNO (base) == SP_REG if base is a REG.
8047 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8048 if it is different SSA_NAME.
8049 (optimize_range_tests_cmp_bitwise): New function.
8050 (optimize_range_tests): Call it.
8052 2017-09-26 Richard Biener <rguenther@suse.de>
8054 PR tree-optimization/82321
8055 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8056 for the def being inside the loop.
8058 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8060 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8062 * config/s390/s390-builtins.def: Fix constraint on op4.
8064 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8066 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8067 independent expanders.
8068 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8069 ("vec_ordered", "vec_unordered"): New expanders.
8071 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8073 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8076 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8078 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8079 vec_unpacks_lo_v16qi.
8080 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8082 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8084 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8085 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8086 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8088 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8090 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8092 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8094 ("*vec_slb<mode>"): New insn pattern.
8095 ("vec_shr_<mode>"): New expander.
8096 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8097 and force the shift count operand to V16QImode.
8098 ("vec_srb<mode>"): Set shift count mode to V16QI.
8100 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8102 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8103 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8104 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8106 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8109 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8111 2017-09-26 Richard Biener <rguenther@suse.de>
8113 PR tree-optimization/82320
8114 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8117 2017-09-25 Jeff Law <law@redhat.com>
8119 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8120 prototype for new argument.
8121 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8122 mostly extracted from rs6000_emit_allocate_stack.
8123 (rs6000_emit_probe_stack_range_stack_clash): New function.
8124 (rs6000_emit_allocate_stack): Call
8125 rs6000_emit_probe_stack_range_stack_clash as needed.
8126 (rs6000_emit_probe_stack_range): Add additional argument
8127 to call to gen_probe_stack_range{si,di}.
8128 (output_probe_stack_range): New.
8129 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8130 (output_probe_stack_range_stack_clash): New.
8131 (rs6000_emit_prologue): Emit notes into dump file as requested.
8132 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8133 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8134 Add additional operand and pass it to output_probe_stack_range.
8136 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8138 PR tree-optimization/82163
8139 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8140 (checking_verify_loop_closed_ssa): New parameter.
8141 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8142 (check_loop_closed_ssa_stmt): Delete.
8143 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8144 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8145 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8148 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8150 * brig-builtins.def: Treat HSAIL barrier builtins as
8151 setjmp/longjump style functions.
8153 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8155 * target.def (constant_alignment): New hook.
8156 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8157 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8158 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8159 * doc/tm.texi: Regenerate.
8160 * targhooks.h (default_constant_alignment): Declare.
8161 (constant_alignment_word_strings): Likewise.
8162 * targhooks.c (default_constant_alignment): New function.
8163 (constant_alignment_word_strings): Likewise.
8164 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8165 instead of CONSTANT_ALIGNMENT.
8166 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8167 (force_const_mem): Likewise.
8168 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8169 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8170 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8171 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8172 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8174 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8175 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8176 constant_alignment_word_strings.
8177 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8178 (CONSTANT_ALIGNMENT): Likewise.
8179 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8180 (arm_constant_alignment): New function.
8181 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8182 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8183 constant_alignment_word_strings.
8184 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8185 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8186 constant_alignment_word_strings.
8187 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8188 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8189 (cris_constant_alignment): New function.
8190 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8191 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8192 (epiphany_constant_alignment): New function.
8193 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8194 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8195 constant_alignment_word_strings.
8196 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8197 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8198 constant_alignment_word_strings.
8199 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8200 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8201 constant_alignment_word_strings.
8202 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8203 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8204 * config/i386/i386.c (ix86_constant_alignment): Make static.
8205 Use the same interface as the target hook.
8206 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8207 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8208 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8209 constant_alignment_word_strings.
8210 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8211 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8212 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8213 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8214 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8215 constant_alignment_word_strings.
8216 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8217 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8218 constant_alignment_word_strings.
8219 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8220 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8221 constant_alignment_word_strings.
8222 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8223 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8225 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8226 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8227 * config/mips/mips.c (mips_constant_alignment): New function.
8228 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8229 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8230 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8231 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8232 (mmix_constant_alignment): Make static. Use the same interface
8234 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8235 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8236 constant_alignment_word_strings.
8237 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8238 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8239 constant_alignment_word_strings.
8240 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8241 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8242 constant_alignment_word_strings.
8243 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8244 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8245 (rs6000_constant_alignment): New function.
8246 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8247 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8248 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8249 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8250 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8251 (rs6000_constant_alignment): New function.
8252 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8253 * config/s390/s390.c (s390_constant_alignment): New function.
8254 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8255 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8256 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8257 constant_alignment_word_strings.
8258 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8259 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8260 (sparc_constant_alignment): New function.
8261 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8262 * config/spu/spu.c (spu_constant_alignment): New function.
8263 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8264 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8265 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8266 constant_alignment_word_strings.
8267 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8268 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8269 constant_alignment_word_strings.
8270 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
8271 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8272 constant_alignment_word_strings.
8273 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
8274 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8275 (visium_constant_alignment): New function.
8276 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
8277 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8278 (xtensa_constant_alignment): New function.
8279 * system.h (CONSTANT_ALIGNMENT): Poison.
8281 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
8283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8284 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
8285 (rs6000_builtin_valid_without_lhs): New helper function.
8286 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8287 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
8289 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8291 * target.h (vec_perm_indices): Use unsigned short rather than
8293 (auto_vec_perm_indices): Likewise.
8294 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
8295 Use unsigned int rather than unsigned char.
8296 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
8298 2017-09-25 Richard Biener <rguenther@suse.de>
8300 * cfgloop.h (sort_sibling_loops): Declare.
8301 * cfgloop.c (sort_sibling_loops_cmp): New helper.
8302 (sort_sibling_loops): New function sorting the sibling loop list
8304 * graphite.c (graphite_transform_loops): Sort sibling loops.
8306 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8308 * target.def (vec_perm_const_ok): Change sel parameter to
8310 * optabs-query.c (can_vec_perm_p): Update accordingly.
8311 * doc/tm.texi: Regenerate.
8312 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8313 auto_vec_perm_indices and remove separate nelt field.
8314 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8315 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8316 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8317 (aarch64_expand_vec_perm_const): Update accordingly.
8318 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8319 to vec_perm_indices.
8320 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8321 auto_vec_perm_indices and remove separate nelt field.
8322 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8323 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8324 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8326 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8327 to vec_perm_indices.
8328 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8329 sel to vec_perm_indices.
8330 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8331 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8332 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8334 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8337 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8340 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8341 on the FUNCTION_DECL function context if it has a DIE that is a
8344 2017-09-25 Richard Biener <rguenther@suse.de>
8346 PR tree-optimization/82285
8347 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8350 2017-09-25 Tom de Vries <tom@codesourcery.com>
8354 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8357 2017-09-25 Richard Biener <rguenther@suse.de>
8359 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8360 ISL errors other than isl_error_quota happen. Dump if the
8361 schedule is the same.
8362 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8363 errors instead of aborting inside ISL.
8365 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8368 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8369 of libgcc_eh for m64.
8370 * config/i386/darwin64.h: Likewise.
8372 2017-09-25 Richard Biener <rguenther@suse.de>
8375 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8376 attribute for incomplete types nor twice for complete ones.
8378 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8381 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8382 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8385 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8388 * config/i386/i386.opt (mprefer-avx256): Use
8389 ix86_target_flags variable.
8390 * config/i386/i386.c (ix86_target_string): Move
8391 -mprefer-avx256 to flag2_opts.
8393 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8396 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8397 and x != -1 | y != -1 into (x & y) != -1.
8399 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8401 * config.gcc: Add new case statement to set
8402 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8403 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8404 s390x-*-linux* case statements. Added aarch64 to the list of
8405 supported architectures.
8407 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8409 PR tree-optimization/82289
8410 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8411 STMT_VINFO_RELEVANT_P.
8413 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8414 Alan Hayward <alan.hayward@arm.com>
8415 David Sherwood <david.sherwood@arm.com>
8417 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8418 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8419 (extract_range_from_binary_expr_1): Don't call
8420 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8422 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8423 Alan Hayward <alan.hayward@arm.com>
8424 David Sherwood <david.sherwood@arm.com>
8426 * target.def (preferred_vector_alignment): New hook.
8427 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8429 * doc/tm.texi: Regenerate.
8430 * targhooks.h (default_preferred_vector_alignment): Declare.
8431 * targhooks.c (default_preferred_vector_alignment): New function.
8432 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8434 (DR_TARGET_ALIGNMENT): New macro.
8435 (aligned_access_p): Update commentary.
8436 (vect_known_alignment_in_bytes): New function.
8437 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8439 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8440 Calculate the misalignment based on the target alignment rather than
8442 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8443 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8444 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8445 the target alignment, rather than masking the element misalignment
8446 with the number of elements in a vector. Also use the target
8447 alignment when calculating the maximum number of peels.
8448 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8449 instead of TYPE_ALIGN_UNIT.
8450 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8451 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8452 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8453 (vect_create_data_ref_ptr): Likewise.
8454 (vect_setup_realignment): Realign by ANDing with
8455 -DR_TARGET_MISALIGNMENT.
8456 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8457 the number of peels based on DR_TARGET_ALIGNMENT.
8458 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8459 with the guaranteed alignment boundary when deciding whether
8461 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8462 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8463 (ensure_base_align): Remove stmt_info parameter. Get the
8464 target base alignment from DR_TARGET_ALIGNMENT.
8465 (vectorizable_store): Update call accordingly. Interpret
8466 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8468 (vectorizable_load): Likewise.
8470 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8471 Alan Hayward <alan.hayward@arm.com>
8472 David Sherwood <david.sherwood@arm.com>
8474 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8475 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8476 (vect_enhance_data_refs_alignment): Likewise.
8478 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8480 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8481 error. Only quit immediately if parsing is complete.
8482 (BEGIN): Initialize fatal_err and parse_done.
8483 (begin fpu, end fpu): Check number of arguments.
8484 (begin arch, end arch): Likewise.
8485 (begin cpu, end cpu): Likewise.
8486 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8487 (optalias): Likewise.
8489 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8491 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8492 * config/arm/arm-isa.h: Delete. Move definitions to ...
8493 * arm-cpus.in: ... here. Use new feature and fgroup values.
8494 * config/arm/arm.c (arm_option_override): Use lower case for feature
8496 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8497 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8498 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8500 (print_isa_bits_for): New function.
8501 (gen_isa): New function.
8502 (gen_comm_data): Use print_isa_bits_for.
8503 (define feature): New keyword.
8504 (define fgroup): New keyword.
8505 * config/arm/t-arm (TM_H): Remove.
8506 (GTM_H): Add arm-isa.h.
8507 (arm-isa.h): Add rule to generate file.
8508 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8509 case for feature bit names.
8511 2017-09-22 Richard Biener <rguenther@suse.de>
8513 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8515 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8516 print of no dependency loops ...
8517 * graphite.c (graphite_transform_loops): ... here.
8518 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8520 (same_close_phi_node, remove_duplicate_close_phi,
8521 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8522 (canonicalize_loop_closed_ssa): ... here and simplify.
8523 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8524 (optimize_isl): Use dump_printf_loc to tell when we stopped
8525 optimizing because of an ISL timeout.
8527 2017-09-22 Richard Biener <rguenther@suse.de>
8529 PR tree-optimization/82291
8530 * tree-if-conv.c (predicate_mem_writes): Make sure to
8531 remove writes in blocks predicated with false.
8533 2017-09-22 Richard Biener <rguenther@suse.de>
8535 * sese.c: Include cfganal.h.
8536 (if_region_set_false_region): Remove.
8537 (create_if_region_on_edge): Likewise.
8538 (move_sese_in_condition): Re-implement without destroying
8541 2017-09-22 Richard Biener <rguenther@suse.de>
8543 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8544 Verify both BBs contain loop PHI nodes before dispatching to
8546 (graphite_regenerate_ast_isl): Do not recompute dominators,
8547 do not verify three times. Restructure for clarity.
8548 * graphite-scop-detection.c (same_close_phi_node,
8549 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8550 defined_in_loop_p, canonicalize_loop_closed_ssa,
8551 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8552 checking and SSA rewrite, move to ...
8553 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8554 (graphite_initialize): Do not pass in ctx, do not reset the
8555 SCEV cache, compute only dominators.
8556 (graphite_transform_loops): Allocate ISL ctx after
8557 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8558 Maintain post-dominators only around build_scops.
8559 * sese.c (if_region_set_false_region): Make static. Free
8560 and recompute dominators.
8561 (move_sese_in_condition): Assert we don't get called with
8562 post-dominators computed.
8563 * sese.h (if_region_set_false_region): Remove.
8565 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8567 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8568 mode attribute for TARGET_AVX512VL.
8570 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8572 * config/i386/i386.opt (mprefer-avx256): New option.
8573 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8575 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8576 for TARGET_PREFER_AVX256.
8577 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8579 2017-09-21 Jeff Law <law@redhat.com>
8581 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8582 Fix dump output if the only stack space is for pushed registers.
8584 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8586 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8589 2017-09-21 Martin Sebor <msebor@redhat.com>
8592 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8593 code (in C++) or code that triggers warnings.
8595 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8597 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8599 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8601 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8602 * sched-rgn.c: Ditto.
8603 * sel-sched-ir.c: Ditto.
8605 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8607 * toplev.h (set_random_seed): Adjust return type.
8608 * toplev.c (init_local_tick): Move eager initialization of random_seed
8609 to get_random_seed. Adjust comment.
8610 (init_random_seed): Inline to get_random_seed, delete.
8611 (get_random_seed): Initialize random_seed lazily.
8612 (set_random_seed): Do not return previous value.
8613 (print_switch_value): Do not call get_random_seed.
8615 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8617 * cgraph.c (delete_function_version): New, broken out from...
8618 (cgraph_node::delete_function_version): ...here. Rename to
8619 cgraph_node::delete_function_version_by_decl. Update all uses.
8620 (cgraph_node::remove): Call delete_function_version.
8622 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8625 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8626 VAR_DECLs to which addressable non-volatile parameters are mapped
8627 and for id->retvar after the return value assignment. Clear
8628 id->retval and id->retbnd after inlining.
8630 2017-09-21 Richard Biener <rguenther@suse.de>
8632 PR tree-optimization/82276
8633 PR tree-optimization/82244
8634 * tree-vrp.c (build_assert_expr_for): Set
8635 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8637 (remove_range_assertions): Revert earlier change.
8639 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8642 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8644 2017-09-21 Richard Biener <rguenther@suse.de>
8646 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8647 Restore valid IL after code generation errors.
8648 * graphite.c (graphite_transform_loops): Diagnose code
8649 generation issues as MSG_MISSED_OPTIMIZATION and continue
8650 with processing SCOPs.
8652 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8653 Alan Hayward <alan.hayward@arm.com>
8654 David Sherwood <david.sherwood@arm.com>
8656 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8657 rather than choosing between plus_constant and gen_rtx_<CODE>.
8658 * expr.c (emit_push_insn): Likewise.
8659 (expand_expr_real_2): Likewise.
8661 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8662 Alan Hayward <alan.hayward@arm.com>
8663 David Sherwood <david.sherwood@arm.com>
8665 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8667 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8668 Alan Hayward <alan.hayward@arm.com>
8669 David Sherwood <david.sherwood@arm.com>
8671 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8672 calling tree_to_uhwi.
8674 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8675 Alan Hayward <alan.hayward@arm.com>
8676 David Sherwood <david.sherwood@arm.com>
8678 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8679 INTEGER_CST rather than a negative test for ADDR_EXPR.
8681 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8682 Alan Hayward <alan.hayward@arm.com>
8683 David Sherwood <david.sherwood@arm.com>
8685 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8686 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8688 2017-09-21 Richard Biener <rguenther@suse.de>
8690 PR tree-optimization/71351
8691 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8692 graphite_create_new_loop_guard): Remove, fold remaining parts
8694 (translate_isl_ast_node_for): ... here and simplify.
8696 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8699 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8700 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8701 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8702 alternative always use QI mode, for -Os imov (=R,R) alternative
8703 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8706 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8707 Jeff Law <law@redhat.com>
8709 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8710 (allocate_stack_space): New function, partially extracted from
8712 (s390_emit_prologue): Track offset to most recent stack probe.
8713 Code to allocate space moved into allocate_stack_space.
8714 Dump actions when no stack is allocated.
8715 (s390_prologue_plus_offset): New function.
8716 (s390_emit_stack_probe): Likewise.
8718 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8720 * common.opt (Wa, Wl, Wp, g, gz=): Add
8722 (gno-column-info): Remove.
8723 (gcolumn-info): Drop RejectNegative.
8725 (gno-record-gcc-switches): Remove.
8726 (grecord-gcc-switches): Drop RejectNegative.
8727 (gno-split-dwarf): Remove.
8728 (gsplit-dwarf): Drop RejectNegative.
8729 (gno-strict-dwarf): Remove.
8730 (gstrict-dwarf): Drop RejectNegative.
8731 * config/darwin.opt (gfull, gused): Add RejectNegative.
8732 * dwarf2out.c (gen_producer_string): Drop
8733 gno-record-gcc-switches handler.
8734 * optc-gen.awk: Add g to prefixes with negative forms.
8735 * opts-common.c (remapping_prefix_p): New.
8736 (find_opt): Check it.
8737 (generate_canonical_option): Test g prefix.
8738 (option_map): Add -gno- mapping.
8739 (add_misspelling_candidates): Check remapping_prefix_p.
8741 2017-09-20 Jeff Law <law@redhat.com>
8743 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8744 thinko in stack clash protection support.
8746 * explow.c (compute_stack_clash_protection_loop_data): Use
8747 CONST_INT_P instead of explicit test. Verify object is a
8748 CONST_INT_P before looking at INTVAL.
8749 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8750 instead of explicit test.
8752 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8755 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8756 address instead of to r1 and r11.
8758 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8760 * config.gcc: Support "knm".
8761 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8762 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8764 * config/i386/i386.c (m_KNM): Define.
8765 (processor_target_table): Add "knm".
8767 (ix86_option_override_internal): Add "knm".
8768 (ix86_issue_rate): Add PROCESSOR_KNM.
8769 (ix86_adjust_cost): Ditto.
8770 (ia32_multipass_dfa_lookahead): Ditto.
8771 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8772 (fold_builtin_cpu): Add M_INTEL_KNM.
8773 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8774 (processor_type): Add PROCESSOR_KNM.
8775 * config/i386/x86-tune.def: Add m_KNM.
8776 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8778 2017-09-20 Richard Biener <rguenther@suse.de>
8780 PR tree-optimization/80213
8781 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8782 are allowed in empty BBs as well.
8783 (canonicalize_loop_closed_ssa): Also look for other complex
8785 (scop_detection::get_sese): Include the loop-closed PHI block
8787 (scop_detection::merge_sese): Remove code adding extra blocks.
8788 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8789 (build_scops): Assert the final returned scop is invalid.
8791 2017-09-20 Richard Biener <rguenther@suse.de>
8793 PR tree-optimization/82264
8794 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8796 (vn_phi_lookup): Likewise.
8797 (vn_phi_insert): Likewise.
8799 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8801 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8802 that fits into uhwi or shwi, add DW_AT_const_value regardless
8803 of early_dwarf without going through RTL, using add_AT_unsigned
8806 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8807 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8808 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8809 (DEBUG_MACRO_SECTION): Likewise.
8810 (DEBUG_LTO_MACRO_SECTION): Likewise.
8811 (DEBUG_STR_DWO_SECTION): Likewise.
8812 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8813 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8814 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8815 (DEBUG_LTO_LINE_STR_SECTION): Define.
8816 (init_sections_and_labels): Initialize debug_line_str_section
8817 variable. Initialize debug_loc_section for -gdwarf-5 to
8818 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8820 2017-09-20 Richard Biener <rguenther@suse.de>
8822 * graphite-sese-to-poly.c (extract_affine): Properly handle
8823 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8825 2017-09-20 Richard Biener <rguenther@suse.de>
8827 PR tree-optimization/81373
8828 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8829 Force SESE live-out defs to be handled even if they are
8832 2017-09-19 Jeff Law <law@redhat.com>
8834 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8835 nothing for stack adjustments with REG_STACK_CHECK.
8836 * sched-deps.c (parse_add_or_inc): Reject insns with
8837 REG_STACK_CHECK from dependency breaking.
8838 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8839 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8840 * reg-notes.def (STACK_CHECK): New note.
8842 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8843 (ix86_expand_prologue): Dump stack clash info as needed.
8844 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8846 * function.c (dump_stack_clash_frame_info): New function.
8847 * function.h (dump_stack_clash_frame_info): Prototype.
8848 (enum stack_clash_probes): New enum.
8850 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8851 flag_stack_clash_protection.
8852 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8853 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8854 (arm_frame_pointer_required): Likewise.
8855 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8856 (ia64_expand_prologue): Likewise.
8857 * config/mips/mips.c (mips_expand_prologue): Likewise.
8858 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8859 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8860 (sparc_flat_expand_prologue): Likewise.
8861 * config/spu/spu.c (spu_expand_prologue): Likewise.
8863 * explow.c: Include "params.h".
8864 (anti_adjust_stack_and_probe_stack_clash): New function.
8865 (get_stack_check_protect): Likewise.
8866 (compute_stack_clash_protection_loop_data): Likewise.
8867 (emit_stack_clash_protection_loop_start): Likewise.
8868 (emit_stack_clash_protection_loop_end): Likewise.
8869 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8870 Use anti_adjust_stack_and_probe_stack_clash.
8871 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8872 (emit_stack_clash_protection_loop_start): Likewise.
8873 (emit_stack_clash_protection_loop_end): Likewise.
8874 * rtl.h (get_stack_check_protect): Prototype.
8875 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8876 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8877 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8879 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8881 * doc/tm.texi: Rebuilt.
8882 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8883 get_stack_check_protect.
8884 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8885 * config/arm/arm.c (arm_expand_prologue): Likewise.
8886 (arm_frame_pointer_required): Likewise.
8887 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8888 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8889 * config/mips/mips.c (mips_expand_prologue): Likewise.
8890 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8891 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8892 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8893 (sparc_flat_expand_prologue): Likewise.
8895 * common.opt (-fstack-clash-protection): New option.
8896 * flag-types.h (enum stack_check_type): Note difference between
8897 -fstack-check= and -fstack-clash-protection.
8898 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8899 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8900 * toplev.c (process_options): Issue warnings/errors for cases
8901 not handled with -fstack-clash-protection.
8902 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8903 (-fstack-check): Note additional problem with -fstack-check=generic.
8904 Note that -fstack-check is primarily for Ada and refer users
8905 to -fstack-clash-protection for stack-clash-protection.
8906 Document new params for stack clash protection.
8908 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8910 * config/i386/i386.c (ix86_split_long_move): Do not handle
8911 address used for LEA in a special way.
8913 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8915 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8917 2017-09-19 Martin Sebor <msebor@redhat.com>
8920 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8921 of incompatible types.
8923 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8926 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8927 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8928 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8930 2017-09-19 Richard Biener <rguenther@suse.de>
8932 PR tree-optimization/82244
8933 * tree-vrp.c (remove_range_assertions): Do not propagate
8934 a constant to abnormals but replace the assert with a copy.
8936 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8938 PR rtl-optimization/57878
8939 PR rtl-optimization/68988
8940 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8941 avoidance test involving non_reload_pseudos. Move frequency test
8942 below the general fragmentation avoidance test.
8944 2017-09-19 Richard Biener <rguenther@suse.de>
8946 PR tree-optimization/69728
8947 * graphite-sese-to-poly.c (schedule_error): New global.
8948 (add_loop_schedule): Handle empty domain by failing the
8950 (build_original_schedule): Handle schedule_error.
8952 2017-09-19 Richard Biener <rguenther@suse.de>
8954 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8955 Do not iterate to sibling loops but only to siblings of inner
8958 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8961 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8962 that the registers are different.
8964 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8966 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8967 to processor_model and "amdfam17h" to arch_names_table.
8968 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8970 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8973 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8975 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8976 Alan Hayward <alan.hayward@arm.com>
8977 David Sherwood <david.sherwood@arm.com>
8979 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8981 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8983 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8984 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8985 vect_schedule_slp_instance.
8986 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8987 Update call to vect_slp_analyze_node_operations. Simplify return
8989 (vect_slp_analyze_bb_1): Update call accordingly.
8990 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8991 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8992 (vect_schedule_slp): Update call accordingly.
8994 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8995 Alan Hayward <alan.hayward@arm.com>
8996 David Sherwood <david.sherwood@arm.com>
8998 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8999 with types that aren't in fact scalar.
9001 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9003 * tree-vect-slp.c (vect_record_max_nunits): New function,
9005 (vect_build_slp_tree_1): ...here.
9006 (vect_build_slp_tree_2): Call it for phis too.
9008 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9010 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9011 to vect_get_vec_def_for_operand when getting the mask operand.
9013 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9014 Alan Hayward <alan.hayward@arm.com>
9015 David Sherwood <david.sherwood@arm.com>
9017 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9020 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9021 Alan Hayward <alan.hayward@arm.com>
9022 David Sherwood <david.sherwood@arm.com>
9024 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9025 calculation for vector booleans.
9027 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9028 Alan Hayward <alan.hayward@arm.com>
9029 David Sherwood <david.sherwood@arm.com>
9031 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9033 (vect_transform_stmt): ...here.
9034 (vect_analyze_stmt): Use it instead of calling
9035 vectorizable_live_operation directly.
9037 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9039 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9040 non-SIMT targets in acc vector loops.
9042 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9044 * configure.ac: Add arc and check if assembler supports gdwarf2.
9045 * configure: Regenerate.
9047 2017-09-18 Richard Biener <rguenther@suse.de>
9049 PR tree-optimization/82220
9050 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9051 epilogue niters from the min_profitable_iters compute.
9053 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9056 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9057 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9058 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9060 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9063 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9064 switching to a new text section.
9066 2017-09-18 Richard Biener <rguenther@suse.de>
9068 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9070 (build_alias_set): Reject aliases with no access function.
9072 2017-09-18 Richard Biener <rguenther@suse.de>
9074 PR tree-optimization/79622
9075 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9077 (build_cross_bb_scalars_use): Likewise.
9079 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9081 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9083 2017-09-18 Alan Modra <amodra@gmail.com>
9086 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9087 stack_pointer_rtx for count 0. Update comments. Break up
9088 large rtl expression.
9090 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9092 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9093 Increase to 20 bytes.
9094 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9095 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9096 or avx version of the stub.
9098 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9101 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9102 compute the minimum stack alignment. Also update preferred stack
9103 boundary for leaf functions.
9105 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9107 PR tree-optimization/82228
9108 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9111 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9113 * common/config/nds32/nds32-common.c
9114 (nds32_option_optimization_table): Refine formatting.
9115 (nds32_option_optimization_table): Use -fsched-pressure and
9116 -fomit-frame-pointer for specific optimization level.
9118 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9120 * config/nds32/nds32.c: Refine formatting and comments.
9121 * config/nds32/nds32.h: Likewise.
9122 * config/nds32/nds32.md: Likewise.
9123 * config/nds32/nds32-cost.c: Likewise.
9124 * config/nds32/nds32-isr.c: Likewise.
9125 * config/nds32/nds32-md-auxiliary.c: Likewise.
9126 * config/nds32/nds32-multiple.md: Likewise.
9127 * config/nds32/nds32-predicates.c: Likewise.
9129 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9130 Jakub Jelinek <jakub@redhat.com>
9132 Add support for -std=c++2a.
9133 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9135 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9137 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9140 * doc/extend.texi (Common Function Attributes): Add
9141 references to ARM, AArch64, and S/390 specific attributes.
9142 (Function Specific Option Pragmas): Add AArch64 and S/390
9143 to list of back ends that support the target pragma.
9145 2017-09-15 Nathan Sidwell <nathan@acm.org>
9147 * doc/standards.texi: Fix C++17 description. Update URLs for
9150 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9152 * common.opt (Wcast-align=strict): New warning option.
9153 * doc/invoke.texi: Document -Wcast-align=strict.
9155 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9157 * cgraph.h (cgraph_thunk_info): Add comments.
9158 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9159 assert for VIRTUAL_* arguments stricter.
9161 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9163 PR tree-optimization/71026
9164 * match.pd: Move RDIV patterns from fold-const.c
9165 * fold-const.c (distribute_real_division): Removed.
9166 (fold_binary_loc): Remove calls to distribute_real_divison.
9168 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9170 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9171 c++1z and gnu++1z as deprecated. Change other references to
9172 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9173 Change -Wc++1z-compat to -Wc++17-compat.
9174 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9175 * dwarf2out.c (highest_c_language): Handle C++17.
9176 (gen_compile_unit_die): Likewise.
9178 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9180 PR rtl-optimization/82192
9181 * combine.c (make_extraction): Don't look through non-paradoxical
9182 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9185 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9186 Alan Hayward <alan.hayward@arm.com>
9187 David Sherwood <david.sherwood@arm.com>
9189 * target.def (function_arg_offset): New hook.
9190 * targhooks.h (default_function_arg_offset): Declare.
9191 * targhooks.c (default_function_arg_offset): New function.
9192 * function.c (locate_and_pad_parm): Use
9193 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9194 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9195 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9196 * doc/tm.texi: Regenerate.
9197 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9198 * config/spu/spu.c (spu_function_arg_offset): New function.
9199 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9200 * system.h (FUNCTION_ARG_OFFSET): Poison.
9202 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9203 Alan Hayard <alan.hayward@arm.com>
9204 David Sherwood <david.sherwood@arm.com>
9206 * target.def (truly_noop_truncation): New hook.
9207 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9208 than TRULY_NOOP_TRUNCATION.
9209 * hooks.h (hook_bool_uint_uint_true): Declare.
9210 * hooks.c (hook_bool_uint_uint_true): New function.
9211 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9212 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9213 * doc/tm.texi: Regenerate.
9214 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9215 rather than TRULY_NOOP_TRUNCATION in comments.
9216 (simplify_comparison): Likewise.
9217 (record_truncated_value): Likewise.
9218 * expmed.c (extract_bit_field_1): Likewise.
9219 (extract_split_bit_field): Likewise.
9220 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9221 instead of TRULY_NOOP_TRUNCATION.
9222 * function.c (assign_parm_setup_block): Likewise.
9223 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9224 * rtlhooks.c: Include target.h.
9225 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9226 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9227 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9228 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9229 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9230 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9231 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9232 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9233 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9234 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9235 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9236 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9237 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9238 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9239 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9240 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9241 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9242 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9243 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9244 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9245 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9246 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9247 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9248 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9249 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9250 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9251 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9252 rather than TRULY_NOOP_TRUNCATION in comments.
9253 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9254 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9255 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9256 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9257 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9258 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9259 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9260 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9261 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9262 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9263 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9264 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9265 rather than TRULY_NOOP_TRUNCATION in comments.
9266 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9267 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9268 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
9269 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
9270 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
9271 TRULY_NOOP_TRUNCATION condition.
9272 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
9273 (TRULY_NOOP_TRUNCATION): Delete.
9274 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
9275 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
9276 * config/spu/spu.c (spu_truly_noop_truncation): New function.
9277 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9278 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
9279 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
9280 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
9281 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9282 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9283 rather than TRULY_NOOP_TRUNCATION in comments.
9284 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
9285 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
9286 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
9287 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
9288 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
9289 * system.h (TRULY_NOOP_TRUNCATION): Poison.
9291 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
9294 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
9295 (*cmp_ior): Likewise.
9296 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
9297 (*ior_scc_scc_cmp): Likewise.
9298 (*and_scc_scc): Likewise.
9299 (*and_scc_scc_cmp): Likewise.
9301 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9302 Alan Hayard <alan.hayward@arm.com>
9303 David Sherwood <david.sherwood@arm.com>
9305 * target.def (can_change_mode_class): New hook.
9306 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9307 (hard_regno_nregs): Likewise.
9308 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9309 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9310 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9311 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9312 (LOAD_EXTEND_OP): Update accordingly.
9313 * doc/tm.texi: Regenerate.
9314 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9315 CANNOT_CHANGE_MODE_CLASS.
9316 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9317 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9318 * combine.c (simplify_set): Update accordingly.
9319 * emit-rtl.c (validate_subreg): Likewise.
9320 * recog.c (general_operand): Likewise.
9321 * regcprop.c (mode_change_ok): Likewise.
9322 * reload1.c (choose_reload_regs): Likewise.
9323 (inherit_piecemeal_p): Likewise.
9324 * rtlanal.c (simplify_subreg_regno): Likewise.
9325 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9326 instead of CANNOT_CHANGE_MODE_CLASS.
9327 (reload_cse_simplify_operands): Likewise.
9328 * reload.c (push_reload): Use targetm.can_change_mode_class
9329 instead of CANNOT_CHANGE_MODE_CLASS.
9330 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9331 REG_CANNOT_CHANGE_MODE_P.
9332 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9333 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9335 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9336 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9337 (arm_can_change_mode_class): New function.
9338 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9339 than CANNOT_CHANGE_MODE_CLASS in comments.
9340 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9341 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9342 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9343 (ix86_can_change_mode_class): ...this new function, inverting the
9344 sense of the return value.
9345 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9346 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9347 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9348 (ia64_can_change_mode_class): New function.
9349 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9350 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9351 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9352 (m32c_can_change_mode_class): ...this new function, inverting the
9353 sense of the return value.
9354 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9355 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9356 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9357 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9358 (mips_can_change_mode_class): ...this new function, inverting the
9359 sense of the return value.
9360 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9361 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9362 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9363 (msp430_can_change_mode_class): New function.
9364 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9365 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9366 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9367 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9368 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9369 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9370 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9371 (pa_cannot_change_mode_class): Replace with...
9372 (pa_can_change_mode_class): ...this new function, inverting the
9373 sense of the return value.
9374 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9375 than CANNOT_CHANGE_MODE_CLASS in comments.
9376 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9377 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9378 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9379 (pdp11_cannot_change_mode_class): Replace with...
9380 (pdp11_can_change_mode_class): ...this new function, inverting the
9381 sense of the return value.
9382 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9383 * config/powerpcspe/powerpcspe-protos.h
9384 (rs6000_cannot_change_mode_class_ptr): Delete.
9385 * config/powerpcspe/powerpcspe.c
9386 (rs6000_cannot_change_mode_class_ptr): Delete.
9387 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9388 (rs6000_option_override_internal): Assign to
9389 targetm.can_change_mode_class instead of
9390 rs6000_cannot_change_mode_class_ptr.
9391 (rs6000_cannot_change_mode_class): Replace with...
9392 (rs6000_can_change_mode_class): ...this new function, inverting the
9393 sense of the return value.
9394 (rs6000_debug_cannot_change_mode_class): Replace with...
9395 (rs6000_debug_can_change_mode_class): ...this new function.
9396 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9397 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9398 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9399 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9400 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9402 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9403 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9404 (rs6000_option_override_internal): Assign to
9405 targetm.can_change_mode_class instead of
9406 rs6000_cannot_change_mode_class_ptr.
9407 (rs6000_cannot_change_mode_class): Replace with...
9408 (rs6000_can_change_mode_class): ...this new function, inverting the
9409 sense of the return value.
9410 (rs6000_debug_cannot_change_mode_class): Replace with...
9411 (rs6000_debug_can_change_mode_class): ...this new function.
9412 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9413 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9414 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9415 (s390_can_change_mode_class): ...this new function, inverting the
9416 sense of the return value.
9417 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9418 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9419 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9420 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9421 (sh_cannot_change_mode_class): Replace with...
9422 (sh_can_change_mode_class): ...this new function, inverting the
9423 sense of the return value.
9424 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9425 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9426 (sparc_can_change_mode_class): New function.
9427 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9428 * config/spu/spu.c (spu_can_change_mode_class): New function.
9429 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9430 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9431 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9432 (visium_can_change_mode_class): New function.
9433 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9435 2017-09-15 Richard Biener <rguenther@suse.de>
9437 PR tree-optimization/82217
9438 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9439 but not undefined case.
9441 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9444 * postreload.c (reload_cse_simplify_operands): Skip
9445 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9447 2017-09-15 Richard Biener <rguenther@suse.de>
9449 PR tree-optimization/68823
9450 * graphite-scop-detection.c (build_alias_set): If we have a
9451 possible dependence check whether we can handle them by just
9452 looking at the DRs DR_ACCESS_FNs.
9453 (build_scops): If build_alias_set fails, fail the SCOP.
9455 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9457 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9458 to support float128 built-in functions that require the ISA 3.0
9460 (BU_FLOAT128_3_HW): Likewise.
9461 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9463 (FMAF128): Likewise.
9465 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9466 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9467 floating point instructions.
9468 (rs6000_invalid_builtin): Likewise.
9469 (rs6000_builtin_mask_names): Likewise.
9470 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9471 (RS6000_BTM_FLOAT128_HW): Likewise.
9472 (RS6000_BTM_COMMON): Likewise.
9473 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9475 * doc/extend.texi (RS/6000 built-in functions): Document the
9476 IEEE 128-bit floating point square root and fused multiply-add
9479 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9481 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9482 reg (r2) isn't in the set of registers defined in the prologue.
9484 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9485 Alan Hayward <alan.hayward@arm.com>
9486 David Sherwood <david.sherwood@arm.com>
9488 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9489 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9490 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9491 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9492 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9494 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9495 max_vectorization_factor.
9496 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9498 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9499 Alan Hayward <alan.hayward@arm.com>
9500 David Sherwood <david.sherwood@arm.com>
9502 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9503 (vect_worthwhile_without_simd_p): Declare.
9504 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9505 (vectorizable_reduction): Use it.
9506 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9507 (vectorizable_operation): Likewise.
9509 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9510 Alan Hayward <alan.hayward@arm.com>
9511 David Sherwood <david.sherwood@arm.com>
9513 * tree-vectorizer.h (vect_get_num_copies): New function.
9514 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9515 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9516 (vectorizable_induction): Likewise.
9517 (vectorizable_live_operation): Likewise.
9518 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9519 (vectorizable_bswap): Likewise.
9520 (vectorizable_call): Likewise.
9521 (vectorizable_conversion): Likewise.
9522 (vectorizable_assignment): Likewise.
9523 (vectorizable_shift): Likewise.
9524 (vectorizable_operation): Likewise.
9525 (vectorizable_store): Likewise.
9526 (vectorizable_load): Likewise.
9527 (vectorizable_condition): Likewise.
9528 (vectorizable_comparison): Likewise.
9529 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9531 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9532 Alan Hayward <alan.hayward@arm.com>
9533 David Sherwood <david.sherwood@arm.com>
9535 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9536 of vect_init_vector.
9538 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9539 Alan Hayward <alan.hayward@arm.com>
9540 David Sherwood <david.sherwood@arm.com>
9542 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9543 an inline wrapper that provides a location.
9544 (gimple_build_vector): Likewise.
9545 * gimple-fold.c (gimple_build_vector_from_val): New function.
9546 (gimple_build_vector): Likewise.
9547 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9548 functions to build the initial value. Always return a gimple value.
9549 (get_initial_defs_for_reduction): Likewise. Only compute
9551 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9552 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9553 (vectorizable_induction): Use gimple_build_vector rather than
9556 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9557 Alan Hayward <alan.hayward@arm.com>
9558 David Sherwood <david.sherwood@arm.com>
9560 * target.h (vec_perm_indices): New typedef.
9561 (auto_vec_perm_indices): Likewise.
9562 * optabs-query.h: Include target.h
9563 (can_vec_perm_p): Take a vec_perm_indices *.
9564 * optabs-query.c (can_vec_perm_p): Likewise.
9565 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9566 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9567 * tree-vect-generic.c (lower_vec_perm): Likewise.
9568 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9569 (vect_grouped_load_supported): Likewise.
9570 (vect_shift_permute_load_chain): Likewise.
9571 (vect_permute_store_chain): Use auto_vec_perm_indices.
9572 (vect_permute_load_chain): Likewise.
9573 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9574 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9575 Update uses of can_vec_perm_p.
9576 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9577 mode with a number of elements. Take a vec_perm_indices *.
9578 (vect_create_epilog_for_reduction): Update accordingly.
9579 Use auto_vec_perm_indices.
9580 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9581 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9582 (vect_transform_slp_perm_load): Likewise.
9583 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9584 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9585 (vect_gen_perm_mask_checked): Likewise.
9586 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9587 (vect_gen_perm_mask_checked): Likewise.
9588 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9589 (vectorizable_store): Likewise.
9590 (vectorizable_load): Likewise.
9591 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9592 (vectorizable_bswap): Likewise.
9594 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9595 Alan Hayward <alan.hayward@arm.com>
9596 David Sherwood <david.sherwood@arm.com>
9598 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9599 * tree.c (build_vector): Likewise.
9600 (build_vector_from_ctor): Update accordingly.
9601 (build_vector_from_val): Likewise.
9602 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9603 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9604 * tree-vect-generic.c (add_rshift): Likewise.
9605 (expand_vector_divmod): Likewise.
9606 (optimize_vector_constructor): Likewise.
9607 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9608 (vect_transform_slp_perm_load): Likewise.
9609 (vect_schedule_slp_instance): Likewise.
9610 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9611 (vectorizable_call): Likewise.
9612 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9613 * expmed.c (make_tree): Likewise.
9614 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9615 a vector passed to build_vector.
9616 (fold_convert_const): Likewise.
9617 (exact_inverse): Likewise.
9618 (fold_ternary_loc): Likewise.
9619 (fold_relational_const): Likewise.
9620 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9621 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9622 (const_unop): Likewise. Store the reduction accumulator in a
9623 variable rather than an array.
9624 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9625 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9626 the new vector, rather than constructing it after the input arrays.
9627 (native_interpret_vector): Use auto_vec<tree> when building
9628 a vector passed to build_vector. Add elements in order.
9629 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9630 auto_vec<tree> when building a vector passed to build_vector.
9631 (vect_create_epilog_for_reduction): Likewise.
9632 (vectorizable_induction): Likewise.
9633 (get_initial_def_for_reduction): Likewise. Fix indentation of
9635 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9637 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9638 passed to build_vector.
9640 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9641 Alan Hayward <alan.hayward@arm.com>
9642 David Sherwood <david.sherwood@arm.com>
9644 * tree-core.h (tree_base::u): Add an "nelts" field.
9645 (tree_vector): Use VECTOR_CST_NELTS as the length.
9646 * tree.c (tree_size): Likewise.
9647 (make_vector): Initialize VECTOR_CST_NELTS.
9648 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9649 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9650 TYPE_VECTOR_SUBPARTS.
9651 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9652 as the number of units, setting it from VECTOR_CST_NELTS.
9653 (const_vector_from_tree): Likewise.
9654 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9655 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9656 (fold_negate_expr_1): Likewise.
9657 (fold_convert_const): Likewise.
9658 (const_binop): Likewise. Differentiate the number of output and
9660 (const_unop): Likewise.
9661 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9662 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9663 in cases that did the opposite.
9665 2017-09-14 Richard Biener <rguenther@suse.de>
9667 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9670 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9672 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9674 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9677 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9678 if and where to split a bb, except for splitting before debug insn
9679 sequences followed by non-label real insn. Delete debug insns
9680 in between basic blocks.
9682 * combine.c (make_compound_operation_int): Formatting fixes.
9684 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9685 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9686 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9687 * config/sol2.h (LINK_EH_SPEC): Likewise.
9688 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9689 * config/s390/linux.h (LINK_SPEC): Likewise.
9690 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9691 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9692 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9693 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9694 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9695 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9696 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9697 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9699 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9701 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9703 (ENDFILE_LINUX_SPEC): Likewise.
9704 (LINK_EH_SPEC): Likewise.
9705 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9706 (LINK_OS_LINUX_SPEC32): Likewise.
9707 (LINK_OS_LINUX_SPEC64): Likewise.
9708 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9709 (LINK_OS_LINUX_SPEC): Likewise.
9711 2017-09-13 Martin Liska <mliska@suse.cz>
9714 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9715 CASE_HIGH is NULL_TREE.
9717 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9718 Alan Hayward <alan.hayward@arm.com>
9719 David Sherwood <david.sherwood@arm.com>
9721 * target.def (secondary_memory_needed): New hook.
9722 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9723 instead of SECONDARY_MEMORY_NEEDED.
9724 (secondary_memory_needed_mode): Likewise.
9725 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9726 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9727 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9728 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9729 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9730 * doc/tm.texi: Regenerate.
9731 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9732 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9733 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9734 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9735 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9736 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9737 mode argument first and change the reg_class arguments to reg_class_t.
9738 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9739 Make static. Update the call to inline_secondary_memory_needed.
9740 (ix86_register_move_cost): Update the call to
9741 inline_secondary_memory_needed.
9742 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9743 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9745 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9746 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9748 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9749 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9750 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9751 and match hook interface. Add comment from mips.h.
9752 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9753 * config/mmix/mmix.md (truncdfsf2): Refer to
9754 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9756 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9757 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9758 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9759 (pa_secondary_memory_needed): New function.
9760 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9761 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9762 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9763 (pdp11_secondary_memory_needed): Make static and match hook interface.
9764 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9765 * config/powerpcspe/powerpcspe-protos.h
9766 (rs6000_secondary_memory_needed_ptr): Delete.
9767 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9769 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9770 (rs6000_option_override_internal): Assign to
9771 targetm.secondary_memory_needed rather than
9772 rs6000_secondary_memory_needed_ptr.
9773 (rs6000_secondary_memory_needed): Match hook interface.
9774 (rs6000_debug_secondary_memory_needed): Likewise.
9775 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9776 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9777 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9778 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9779 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9780 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9782 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9783 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9784 (rs6000_option_override_internal): Assign to
9785 targetm.secondary_memory_needed rather than
9786 rs6000_secondary_memory_needed_ptr.
9787 (rs6000_secondary_memory_needed): Match hook interface.
9788 (rs6000_debug_secondary_memory_needed): Likewise.
9789 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9790 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9791 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9792 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9793 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9794 (sparc_secondary_memory_needed): New function.
9795 * lra-constraints.c (check_and_process_move): Refer to
9796 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9798 (curr_insn_transform): Likewise.
9799 (process_alt_operands): Use targetm.secondary_memory_needed
9800 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9801 (check_secondary_memory_needed_p): Likewise.
9802 (choose_split_class): Likewise.
9803 * reload.c: Unconditionally include code that was previously
9804 conditional on SECONDARY_MEMORY_NEEDED.
9805 (push_secondary_reload): Use targetm.secondary_memory_needed
9806 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9807 (push_reload): Likewise.
9808 * reload1.c: Unconditionally include code that was previously
9809 conditional on SECONDARY_MEMORY_NEEDED.
9810 (choose_reload_regs): Use targetm.secondary_memory_needed
9811 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9812 (gen_reload): Likewise.
9813 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9815 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9816 Alan Hayward <alan.hayward@arm.com>
9817 David Sherwood <david.sherwood@arm.com>
9819 * target.def (secondary_memory_needed_mode): New hook:
9820 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9821 * targhooks.h (default_secondary_memory_needed_mode): New function.
9822 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9823 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9824 * doc/tm.texi: Regenerate.
9825 * lra-constraints.c (check_and_process_move): Use
9826 targetm.secondary_memory_needed_mode instead of
9827 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9828 (curr_insn_transform): Likewise.
9829 * reload.c (get_secondary_mem): Likewise.
9830 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9831 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9833 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9834 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9835 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9836 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9837 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9839 * config/powerpcspe/powerpcspe-protos.h
9840 (rs6000_secondary_memory_needed_mode): Delete.
9841 * config/powerpcspe/powerpcspe.c
9842 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9843 (rs6000_secondary_memory_needed_mode): Make static.
9844 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9845 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9847 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9849 (rs6000_secondary_memory_needed_mode): Make static.
9850 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9851 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9852 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9853 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9854 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9856 (sparc_secondary_memory_needed_mode): New function.
9857 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9859 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9861 * config/aarch64/constraints.md (Umq): New constraint.
9862 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9864 (mov<mode>): Update condition.
9866 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9868 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9869 when bitposition is the same.
9871 2017-09-13 Richard Biener <rguenther@suse.de>
9873 * dwarf2out.c (output_die_symbol): Remove.
9874 (output_die): Do not output a DIEs symbol.
9876 2017-09-13 Richard Biener <rguenther@suse.de>
9879 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9880 default-def to avoid breaking iterator update with the weird
9881 interaction with cgraph_update_edges_for_call_stmt_node.
9883 2017-09-13 Richard Biener <rguenther@suse.de>
9885 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9886 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9887 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9888 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9890 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9892 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9893 Disable pc relative literal load irrespective of
9894 TARGET_FIX_ERR_A53_84341 for default.
9896 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9898 * config/sparc/sparc.c (output_return): Output the source location of
9899 the insn in the delay slot, if any.
9900 (output_sibcall): Likewise.
9902 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9905 * common.opt (-static-pie): New alias.
9906 (shared): Negate static-pie.
9907 (-no-pie): Update help text.
9909 (static-pie): New option.
9910 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9911 -static-pie support.
9912 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9913 (LINK_EH_SPEC): Likewise.
9914 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9915 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9916 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9917 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9918 (init_gcc_specs): Likewise.
9919 (init_spec): Likewise.
9920 (display_help): Update help message for -pie.
9921 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9924 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9926 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9927 (movdi_aarch64): Likewise.
9928 (movti_aarch64): Likewise.
9930 2017-09-12 Simon Wright <simon@pushface.org>
9933 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9934 calculation of the minor version, always output as 0.
9936 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9939 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9940 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9941 on it early, rather than manual conversion late. For
9942 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9943 instead of performing manual conversion.
9945 2017-09-12 Carl Love <cel@us.ibm.com>
9947 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9948 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9950 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9951 VMULOSW): Add definitions.
9952 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9953 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9954 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9956 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9958 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9960 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9962 (movti_aarch64): Likewise.
9963 (movdf_aarch64): Likewise.
9964 (movtf_aarch64): Likewise.
9965 (load_pairdi): Likewise.
9966 (store_pairdi): Likewise.
9967 (load_pairdf): Likewise.
9968 (store_pairdf): Likewise.
9969 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9970 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9971 (ldr_got_small_<mode>): Likewise.
9972 (ldr_got_small_28k_<mode>): Likewise.
9973 (ldr_got_tiny): Likewise.
9974 * config/aarch64/iterators.md (ldst_sz): New.
9975 (ldpstp_sz): Likewise.
9976 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9978 (thunderx_load): Split load_8 to load_16.
9979 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9981 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9982 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9983 (xgene1_store_pair): Split store_8 to store_16.
9984 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9985 (falkor_st_0_st_sd): Split store_8 to store_16.
9987 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9989 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9990 and store1/2/3/4 to store_4/8/12/16.
9991 * config/aarch64/aarch64.md: Update for rename.
9992 * config/arm/arm.md: Likewise.: Likewise.
9993 * config/arm/arm.c: Likewise.
9994 * config/arm/thumb1.md: Likewise.
9995 * config/arm/thumb2.md: Likewise.
9996 * config/arm/vfp.md: Likewise.
9997 * config/arm/arm-generic.md: Likewise.
9998 * config/arm/arm1020e.md: Likewise.
9999 * config/arm/arm1026ejs.md: Likewise.
10000 * config/arm/arm1136jfs.md: Likewise.
10001 * config/arm/arm926ejs.md: Likewise.
10002 * config/arm/cortex-a15.md: Likewise.
10003 * config/arm/cortex-a17.md: Likewise.
10004 * config/arm/cortex-a5.md: Likewise.
10005 * config/arm/cortex-a53.md: Likewise.
10006 * config/arm/cortex-a57.md: Likewise.
10007 * config/arm/cortex-a7.md: Likewise.
10008 * config/arm/cortex-a8.md: Likewise.
10009 * config/arm/cortex-a9.md: Likewise.
10010 * config/arm/cortex-m4.md: Likewise.
10011 * config/arm/cortex-m7.md: Likewise.
10012 * config/arm/cortex-r4.md: Likewise.
10013 * config/arm/exynos-m1.md: Likewise.
10014 * config/arm/fa526.md: Likewise.
10015 * config/arm/fa606te.md: Likewise.
10016 * config/arm/fa626te.md: Likewise.
10017 * config/arm/fa726te.md: Likewise.
10018 * config/arm/fmp626.md: Likewise.
10019 * config/arm/iwmmxt.md: Likewise.
10020 * config/arm/ldmstm.md: Likewise.
10021 * config/arm/marvell-pj4.md: Likewise.
10022 * config/arm/xgene1.md: Likewise.
10023 * config/aarch64/thunderx.md: Likewise.
10024 * config/aarch64/thunderx2t99.md: Likewise.
10025 * config/aarch64/falkor.md: Likewise.
10027 2017-09-12 Martin Liska <mliska@suse.cz>
10029 * attribs.c (private_lookup_attribute): New function.
10030 * attribs.h (private_lookup_attribute): Declared here.
10031 (lookup_attribute): Called from this place.
10033 2017-09-12 Richard Biener <rguenther@suse.de>
10035 PR tree-optimization/82157
10036 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10037 stmts with side-effects.
10039 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10040 Alan Hayward <alan.hayward@arm.com>
10041 David Sherwood <david.sherwood@arm.com>
10043 * target.def (hard_regno_nregs): New hook.
10044 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10045 * targhooks.h (default_hard_regno_nregs): Declare.
10046 * targhooks.c (default_hard_regno_nregs): New function.
10047 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10048 (TARGET_HARD_REGNO_NREGS): ...this hook.
10049 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10050 (CLASS_MAX_NREGS): Likewise.
10051 * doc/tm.texi: Regenerate.
10052 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10053 instead of HARD_REGNO_NREGS.
10054 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10055 HARD_REGNO_NREGS in the comment.
10056 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10057 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10058 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10059 Return an unsigned int.
10060 (TARGET_HARD_REGNO_NREGS): Redefine.
10061 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10062 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10063 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10064 (arc_hard_regno_nregs): New function.
10065 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10066 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10067 (arm_hard_regno_nregs): New function.
10068 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10069 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10070 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10071 (TARGET_HARD_REGNO_NREGS): Redefine.
10072 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10073 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10074 (HARD_REGNO_NREGS): Delete.
10075 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10076 (cr16_hard_regno_nregs): New function.
10077 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10078 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10079 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10080 (cris_hard_regno_nregs): New function.
10081 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10082 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10083 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10084 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10085 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10086 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10087 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10088 (frv_hard_regno_nregs): Make static. Take and return an
10090 (frv_class_max_nregs): Remove outdated copy of documentation.
10091 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10092 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10093 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10094 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10095 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10096 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10097 (TARGET_HARD_REGNO_NREGS): Redefine.
10098 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10099 (CLASS_MAX_NREGS): Update comment.
10100 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10101 (ia64_hard_regno_nregs): New function.
10102 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10103 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10104 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10105 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10106 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10108 (m32c_hard_regno_nregs): Likewise. Make static.
10109 (TARGET_HARD_REGNO_NREGS): Redefine.
10110 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10111 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10112 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10113 (m68k_hard_regno_nregs): New function.
10114 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10115 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10116 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10117 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10118 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10119 Take and return an unsigned int.
10120 (TARGET_HARD_REGNO_NREGS): Redefine.
10121 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10122 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10123 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10124 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10125 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10126 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10127 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10128 (msp430_hard_regno_nregs): Make static. Take and return an
10130 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10131 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10132 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10133 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10134 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10135 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10136 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10137 (TARGET_HARD_REGNO_NREGS): Redefine.
10138 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10139 (PA_HARD_REGNO_NREGS): ...this.
10140 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10141 (PA_HARD_REGNO_NREGS): ...this.
10142 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10143 (pa_hard_regno_nregs): New function.
10144 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10145 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10146 (pdp11_hard_regno_nregs): New function.
10147 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10148 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10149 (rs6000_hard_regno_nregs_hook): New function.
10150 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10151 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10152 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10153 Take and return an unsigned int. Move earlier in file.
10154 (TARGET_HARD_REGNO_NREGS): Redefine.
10155 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10156 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10157 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10158 (rl78_hard_regno_nregs): Make static. Take and return an
10160 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10161 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10162 (rs6000_hard_regno_nregs_hook): New function.
10163 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10164 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10165 (TARGET_HARD_REGNO_NREGS): Redefine.
10166 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10167 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10168 instead of HARD_REGNO_NREGS.
10169 (s390_hard_regno_nregs): New function.
10170 (s390_hard_regno_mode_ok): Add comment from s390.h.
10171 (TARGET_HARD_REGNO_NREGS): Redefine.
10172 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10173 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10174 (sh_hard_regno_nregs): New function.
10175 (sh_pass_in_reg_p): Use it.
10176 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10177 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10178 (sparc_hard_regno_nregs): New function.
10179 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10180 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10181 (spu_function_arg_advance): Use it, supplying a valid register number.
10182 (TARGET_HARD_REGNO_NREGS): Redefine.
10183 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10184 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10185 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10186 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10187 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10188 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10189 (CLASS_MAX_NREGS): Remove copy of old documentation.
10190 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10191 (visium_hard_regno_nregs): New function.
10192 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10193 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10194 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10195 xtensa_hard_regno_nregs): New function.
10196 * system.h (HARD_REGNO_NREGS): Poison.
10198 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10200 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10201 hard_regno_nregs instead of HARD_REGNO_NREGS.
10202 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10203 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10204 (c6x_expand_epilogue): Likewise.
10205 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10206 (frv_read_iacc_argument): Likewise.
10207 * config/sh/sh.c: Include regs.h.
10208 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10209 (regs_used): Likewise.
10210 (output_stack_adjust): Likewise.
10211 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10212 * expmed.c: Include regs.h.
10213 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10214 * ree.c: Include regs.h.
10215 (combine_reaching_defs): Use hard_regno_nregs instead of
10217 (add_removable_extension): Likewise.
10219 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10221 * regs.h (hard_regno_nregs): Turn into a function.
10222 (end_hard_regno): Update accordingly.
10223 * caller-save.c (setup_save_areas): Likewise.
10224 (save_call_clobbered_regs): Likewise.
10225 (replace_reg_with_saved_mem): Likewise.
10226 (insert_restore): Likewise.
10227 (insert_save): Likewise.
10228 * combine.c (can_change_dest_mode): Likewise.
10229 (move_deaths): Likewise.
10230 (distribute_notes): Likewise.
10231 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10232 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10233 (rs6000_split_multireg_move): Likewise.
10234 (rs6000_register_move_cost): Likewise.
10235 (rs6000_memory_move_cost): Likewise.
10236 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10237 (rs6000_split_multireg_move): Likewise.
10238 (rs6000_register_move_cost): Likewise.
10239 (rs6000_memory_move_cost): Likewise.
10240 * cselib.c (cselib_reset_table): Likewise.
10241 (cselib_lookup_1): Likewise.
10242 * emit-rtl.c (set_mode_and_regno): Likewise.
10243 * function.c (aggregate_value_p): Likewise.
10244 * ira-color.c (setup_profitable_hard_regs): Likewise.
10245 (check_hard_reg_p): Likewise.
10246 (calculate_saved_nregs): Likewise.
10247 (assign_hard_reg): Likewise.
10248 (improve_allocation): Likewise.
10249 (calculate_spill_cost): Likewise.
10250 * ira-emit.c (modify_move_list): Likewise.
10251 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10252 (ira_hard_reg_in_set_p): Likewise.
10253 * ira.c (setup_reg_mode_hard_regset): Likewise.
10254 (clarify_prohibited_class_mode_regs): Likewise.
10255 (check_allocation): Likewise.
10256 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10257 (lra_setup_reg_renumber): Likewise.
10258 (setup_try_hard_regno_pseudos): Likewise.
10259 (spill_for): Likewise.
10260 (assign_hard_regno): Likewise.
10261 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10262 * lra-constraints.c (in_class_p): Likewise.
10263 (lra_constraint_offset): Likewise.
10264 (simplify_operand_subreg): Likewise.
10265 (lra_constraints): Likewise.
10266 (split_reg): Likewise.
10267 (split_if_necessary): Likewise.
10268 (invariant_p): Likewise.
10269 (inherit_in_ebb): Likewise.
10270 * lra-lives.c (process_bb_lives): Likewise.
10271 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
10272 (get_hard_regs): Likewise.
10273 (do_remat): Likewise.
10274 * lra-spills.c (assign_spill_hard_regs): Likewise.
10275 * mode-switching.c (create_pre_exit): Likewise.
10276 * postreload.c (reload_combine_recognize_pattern): Likewise.
10277 * recog.c (peep2_find_free_register): Likewise.
10278 * regcprop.c (kill_value_regno): Likewise.
10279 (set_value_regno): Likewise.
10280 (copy_value): Likewise.
10281 (maybe_mode_change): Likewise.
10282 (find_oldest_value_reg): Likewise.
10283 (copyprop_hardreg_forward_1): Likewise.
10284 * regrename.c (check_new_reg_p): Likewise.
10285 (regrename_do_replace): Likewise.
10286 * reload.c (push_reload): Likewise.
10287 (combine_reloads): Likewise.
10288 (find_dummy_reload): Likewise.
10289 (operands_match_p): Likewise.
10290 (find_reloads): Likewise.
10291 (find_equiv_reg): Likewise.
10292 (reload_adjust_reg_for_mode): Likewise.
10293 * reload1.c (count_pseudo): Likewise.
10294 (count_spilled_pseudo): Likewise.
10295 (find_reg): Likewise.
10296 (clear_reload_reg_in_use): Likewise.
10297 (free_for_value_p): Likewise.
10298 (allocate_reload_reg): Likewise.
10299 (choose_reload_regs): Likewise.
10300 (reload_adjust_reg_for_temp): Likewise.
10301 (emit_reload_insns): Likewise.
10302 (delete_output_reload): Likewise.
10303 * rtlanal.c (subreg_get_info): Likewise.
10304 * sched-deps.c (sched_analyze_reg): Likewise.
10305 * sel-sched.c (init_regs_for_mode): Likewise.
10306 (mark_unavailable_hard_regs): Likewise.
10307 (choose_best_reg_1): Likewise.
10308 (verify_target_availability): Likewise.
10309 * valtrack.c (dead_debug_insert_temp): Likewise.
10310 * var-tracking.c (track_loc_p): Likewise.
10311 (emit_note_insn_var_location): Likewise.
10312 * varasm.c (make_decl_rtl): Likewise.
10313 * reginfo.c (choose_hard_reg_mode): Likewise.
10314 (init_reg_modes_target): Refer directly to
10315 this_target_regs->x_hard_regno_nregs.
10317 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10319 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10320 instead of hard_regno_nregs.
10322 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10324 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10325 end_hard_regno instead of hard_regno_nregs.
10326 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10327 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10328 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10329 * ira-color.c (improve_allocation): Likewise.
10330 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10331 * lra-lives.c (mark_regno_live): Likewise.
10332 (mark_regno_dead): Likewise.
10333 * lra-remat.c (operand_to_remat): Likewise.
10334 * lra.c (collect_non_operand_hard_regs): Likewise.
10335 * postreload.c (reload_combine_note_store): Likewise.
10336 (move2add_valid_value_p): Likewise.
10337 * reload.c (regno_clobbered_p): Likewise.
10339 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10341 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10343 * config/v850/v850.c (v850_reorg): Likewise.
10344 * reload.c (refers_to_regno_for_reload_p): Likewise.
10345 (find_equiv_reg): Likewise.
10346 * reload1.c (reload_reg_reaches_end_p): Likewise.
10348 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10350 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10352 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10353 * config/arm/arm.c (output_move_neon): Likewise.
10354 (arm_attr_length_move_neon): Likewise.
10355 (neon_split_vcombine): Likewise.
10356 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10357 (c6x_mark_reg_written): Likewise.
10358 (c6x_dwarf_register_span): Likewise.
10359 * config/i386/i386.c (ix86_save_reg): Likewise.
10360 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10361 (rws_access_reg): Likewise.
10362 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10363 * mode-switching.c (create_pre_exit): Likewise.
10364 * ree.c (combine_reaching_defs): Likewise.
10365 (add_removable_extension): Likewise.
10366 * regcprop.c (find_oldest_value_reg): Likewise.
10367 (copyprop_hardreg_forward_1): Likewise.
10368 * reload.c (reload_inner_reg_of_subreg): Likewise.
10369 (push_reload): Likewise.
10370 (combine_reloads): Likewise.
10371 (find_dummy_reload): Likewise.
10372 (reload_adjust_reg_for_mode): Likewise.
10373 * reload1.c (find_reload_regs): Likewise.
10374 (forget_old_reloads_1): Likewise.
10375 (reload_reg_free_for_value_p): Likewise.
10376 (reload_adjust_reg_for_temp): Likewise.
10377 (emit_reload_insns): Likewise.
10378 (delete_output_reload): Likewise.
10379 * sel-sched.c (choose_best_reg_1): Likewise.
10380 (choose_best_pseudo_reg): Likewise.
10382 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10383 Alan Hayward <alan.hayward@arm.com>
10384 David Sherwood <david.sherwood@arm.com>
10386 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10387 * target.def (slow_unaligned_access): New hook.
10388 * targhooks.h (default_slow_unaligned_access): Declare.
10389 * targhooks.c (default_slow_unaligned_access): New function.
10390 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10391 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10392 * doc/tm.texi: Regenerate.
10393 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10394 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10395 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10397 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10398 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10400 (rs6000_slow_unaligned_access): New function.
10401 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10402 (expand_block_compare): Likewise.
10403 (expand_strn_compare): Likewise.
10404 (rs6000_rtx_costs): Likewise.
10405 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10406 (riscv_slow_unaligned_access): Likewise.
10407 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10408 (riscv_slow_unaligned_access_p): ...this and make static.
10409 (riscv_option_override): Update accordingly.
10410 (riscv_slow_unaligned_access): New function.
10411 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10412 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10413 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10414 (rs6000_slow_unaligned_access): New function.
10415 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10416 (rs6000_rtx_costs): Likewise.
10417 * config/rs6000/rs6000-string.c (expand_block_compare)
10418 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10419 of SLOW_UNALIGNED_ACCESS.
10420 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10421 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10422 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10423 of SLOW_UNALIGNED_ACCESS.
10424 * expmed.c (simple_mem_bitfield_p): Likewise.
10425 * expr.c (alignment_for_piecewise_move): Likewise.
10426 (emit_group_load_1): Likewise.
10427 (emit_group_store): Likewise.
10428 (copy_blkmode_from_reg): Likewise.
10429 (emit_push_insn): Likewise.
10430 (expand_assignment): Likewise.
10431 (store_field): Likewise.
10432 (expand_expr_real_1): Likewise.
10433 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10434 * lra-constraints.c (simplify_operand_subreg): Likewise.
10435 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10436 * gimple-ssa-store-merging.c: Likewise in block comment at start
10438 * tree-ssa-strlen.c: Include target.h.
10439 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10440 of SLOW_UNALIGNED_ACCESS.
10441 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10443 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10445 PR rtl-optimization/82185
10446 * expmed.c (emit_store_flag_int): Only test tem if it has been
10449 2017-09-12 Richard Biener <rguenther@suse.de>
10451 PR middle-end/82149
10452 * match.pd ((FTYPE) N CMP CST): Fix typo.
10454 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10456 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10458 (mips_near_type_p): Add 'short_call' attribute as a synonym
10460 * doc/extend.texi (short_call): Document new function attribute.
10462 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10465 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10466 assertion check that in the condition.
10467 (get_atomic_generic_size): Likewise. Before testing if parameter
10468 has pointer type, if it has array type, call for C++
10469 default_conversion to perform array-to-pointer conversion.
10471 2017-09-12 Richard Biener <rguenther@suse.de>
10473 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10474 for operations we cannot scalarize.
10476 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10478 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10479 vectors heap vectors. Clean up comments.
10480 Make visited_bbs a reference.
10481 (profitable_jump_thread_path): Make GC
10482 vectors heap vectors. Clean up comments.
10484 (convert_and_register_jump_thread_path): Make GC vectors heap
10486 (check_subpath_and_update_thread_path): Same. Clean up comments.
10487 Make visited_bbs a reference.
10488 (handle_phi): Abstract common code to to
10489 register_jump_thread_path_if_profitable.
10490 Rename VAR_BB to DEF_BB.
10492 Make GC vectors heap vectors.
10493 Make visited_bbs a reference.
10494 (handle_assignment): Same.
10495 (register_jump_thread_path_if_profitable): New.
10496 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10498 Make GC vectors heap vectors. Clean up comments.
10499 Make visited_bbs a reference.
10500 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10501 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10504 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10507 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10508 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10510 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10512 Revert r251800 and r251799.
10514 2017-09-11 Martin Jambor <mjambor@suse.cz>
10517 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10518 arguments in advance.
10519 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10520 use it to find predecessor edges.
10521 (naive_outof_ssa): Collect vector of predecessors.
10523 2017-09-08 Jason Merrill <jason@redhat.com>
10525 PR c++/70029 - ICE with ref-qualifier and -flto
10526 * langhooks.h (struct lang_hooks_for_types): Add
10527 copy_lang_qualifiers.
10528 * attribs.c (build_type_attribute_qual_variant): Use it.
10529 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10531 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10532 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10534 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10537 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10538 (*mulsi3_sp64): New instruction.
10539 (mulsi3): New expander.
10541 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10543 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10545 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10547 * sancov.c: Include memmodel.h.
10549 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10552 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10555 2017-09-07 Carl Love <cel@us.ibm.com>
10557 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10558 the sldi instruction.
10560 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10562 * sancov.c: Include tm_p.h.
10564 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10567 * output.h (switch_to_other_text_partition): New declaration.
10568 * varasm.c (switch_to_other_text_partition): New function.
10569 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10570 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10571 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10572 to the other text partition before emitting LCL label and switch back
10573 after emitting the word after it.
10575 2017-09-07 Richard Biener <rguenther@suse.de>
10577 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10578 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10579 critical edges here, after loop init.
10580 (pass_data_pre): Remove PROP_no_crit_edges flags.
10581 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10582 for valueization of call args to avoid leaking VN_TOP.
10583 (visit_use): Assert we do not visit default defs.
10584 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10585 Use error_mark_node to more easily detect leaking VN_TOP.
10586 All default-defs are varying, not VN_TOP. Mark them visited.
10587 (run_scc_vn): Make code match comment.
10589 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10591 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10592 OPTION_MASK_FLOAT128_KEYWORD.
10593 (POWERPC_MASKS): Likewise.
10594 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10595 support for the -mfloat128-type option, and make -mfloat128
10596 default on PowerPC Linux systems. Define or undefine
10597 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10598 Define __float128 to be __ieee128 if IEEE 128-bit support is
10599 enabled, or undefine it.
10600 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10601 Delete defining __FLOAT128_TYPE__.
10602 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10603 -mfloat128-type option and make -mfloat128 default on PowerPC
10605 (TARGET_FLOAT128_TYPE): Likewise.
10606 (-mfloat128-type): Likewise.
10607 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10608 Delete the -mfloat128-type option and make -mfloat128 default on
10609 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10610 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10611 128-bit floating point is enabled. Change tests from using
10612 -mfloat128-type to -mfloat128.
10613 (rs6000_mangle_type): Use the correct mangling for the __float128
10614 type even if normal long double is restricted to 64-bits.
10615 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10617 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10618 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10619 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10621 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10622 documentation for -mfloat128.
10624 2017-09-06 Olivier Hainque <hainque@adacore.com>
10626 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10628 2017-09-06 Wish Wu <wishwu007@gmail.com>
10629 Jakub Jelinek <jakub@redhat.com>
10631 * asan.c (initialize_sanitizer_builtins): Add
10632 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10633 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10634 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10635 BT_FN_VOID_UINT64_PTR variables.
10636 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10637 (BT_FN_VOID_UINT16_UINT16): Likewise.
10638 (BT_FN_VOID_UINT32_UINT32): Likewise.
10639 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10640 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10641 (BT_FN_VOID_UINT64_PTR): Likewise.
10642 * common.opt (flag_sanitize_coverage): New variable.
10643 (fsanitize-coverage=trace-pc): Remove.
10644 (fsanitize-coverage=): Add.
10645 * flag-types.h (enum sanitize_coverage_code): New enum.
10646 * fold-const.c (fold_range_test): Disable non-short-circuit
10647 optimization if flag_sanitize_coverage.
10648 (fold_truth_andor): Likewise.
10649 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10650 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10651 (coverage_sanitizer_opts): New array.
10652 (get_closest_sanitizer_option): Add OPTS argument, handle also
10653 OPT_fsanitize_coverage_.
10654 (parse_sanitizer_options): Adjusted to also handle
10655 OPT_fsanitize_coverage_.
10656 (common_handle_option): Add OPT_fsanitize_coverage_.
10657 * sancov.c (instrument_comparison, instrument_switch): New function.
10658 (sancov_pass): Add trace-cmp support.
10659 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10660 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10661 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10662 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10663 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10664 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10665 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10666 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10667 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10668 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10670 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10672 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10673 error. Only quit immediately if parsing is complete.
10674 (BEGIN): Initialize fatal_err and parse_done.
10675 (begin fpu, end fpu): Check number of arguments.
10676 (begin arch, end arch): Likewise.
10677 (begin cpu, end cpu): Likewise.
10678 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10679 (optalias): Likewise.
10681 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10683 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10684 * config/arm/arm-isa.h: Delete. Move definitions to ...
10685 * arm-cpus.in: ... here. Use new feature and fgroup values.
10686 * config/arm/arm.c (arm_option_override): Use lower case for feature
10688 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10689 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10690 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10692 (print_isa_bits_for): New function.
10693 (gen_isa): New function.
10694 (gen_comm_data): Use print_isa_bits_for.
10695 (define feature): New keyword.
10696 (define fgroup): New keyword.
10697 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10698 (arm-isa.h): Add rule to generate file.
10699 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10700 case for feature bit names.
10702 2017-09-06 Richard Biener <rguenther@suse.de>
10704 * tree-ssa-pre.c (NECESSARY): Remove.
10705 (create_expression_by_pieces): Do not touch pass-local flags.
10706 (insert_into_preds_of_block): Likewise.
10707 (do_pre_regular_insertion): Likewise.
10708 (eliminate_insert): Likewise.
10709 (eliminate_dom_walker::before_dom_children): Likewise.
10710 (fini_eliminate): Do not look at inserted_exprs.
10711 (mark_operand_necessary): Remove.
10712 (remove_dead_inserted_code): Replace with simple work-list
10713 algorithm based on inserted_exprs and SSA uses.
10714 (pass_pre::execute): Re-order fini_eliminate and
10715 remove_dead_inserted_code.
10717 2017-09-06 Olivier Hainque <hainque@adacore.com>
10719 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10720 for VxWorks 7. Adjust surrounding comments.
10722 2017-09-06 Richard Biener <rguenther@suse.de>
10724 * gimple-ssa-strength-reduction.c
10725 (find_candidates_dom_walker::before_dom_children): Also allow
10728 2017-09-06 Richard Biener <rguenther@suse.de>
10730 PR tree-optimization/82108
10731 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10732 for gap in the non-permutation SLP case.
10734 2017-09-06 Martin Jambor <mjambor@suse.cz>
10736 PR tree-optimization/82078
10737 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10738 add_access_to_work_queue...
10739 (build_accesses_from_assign): ...here.
10740 (propagate_all_subaccesses): Make sure racc is the group
10741 representative, if there is one.
10743 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10745 PR middle-end/82095
10746 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10749 2017-09-06 Richard Biener <rguenther@suse.de>
10751 * gimple-ssa-strength-reduction.c
10752 (find_candidates_dom_walker::before_doom_children): Use a
10753 type and not a mode check.
10755 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10758 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10760 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10762 2017-09-05 Jeff Law <law@redhat.com>
10764 PR tree-optimization/64910
10765 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10766 cases where we have 3 or more operands.
10768 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10770 PR middle-end/81768
10771 * omp-low.c (lower_omp_for): Recompute tree invariant if
10772 gimple_omp_for_initial/final is ADDR_EXPR.
10774 PR middle-end/81768
10775 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10776 into gimple val before gimplification fo the COND_EXPR.
10778 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10780 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10781 REGION_COPY argument.
10782 (thread_through_all_blocks): Remove unused argument to
10783 duplicate_thread_path.
10785 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10786 Alan Hayward <alan.hayward@arm.com>
10787 David Sherwood <david.sherwood@arm.com>
10789 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10790 Take a scalar_mode rather than a machine_mode.
10791 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10792 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10793 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10794 (aarch64_gen_adjusted_ldpstp): Likewise.
10795 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10798 Alan Hayward <alan.hayward@arm.com>
10799 David Sherwood <david.sherwood@arm.com>
10801 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10802 Take a scalar_int_mode instead of a machine_mode.
10803 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10804 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10805 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10806 (aarch64_simd_attr_length_rglist): Delete.
10807 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10808 a scalar_int_mode instead of a machine_mode.
10809 (aarch64_add_offset): Likewise.
10810 (aarch64_internal_mov_immediate): Likewise
10811 (aarch64_add_constant_internal): Likewise.
10812 (aarch64_add_constant): Likewise.
10813 (aarch64_movw_imm): Likewise.
10814 (aarch64_rtx_arith_op_extract_p): Likewise.
10815 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10816 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10817 Remove assert that the mode isn't a vector.
10818 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10819 (aarch64_expand_mov_immediate): Update calls after above changes.
10820 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10821 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10822 (aarch64_move_imm): Likewise.
10823 (aarch64_can_const_movi_rtx_p): Likewise.
10824 (aarch64_strip_extend): Likewise.
10825 (aarch64_extr_rtx_p): Likewise.
10826 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10827 a CONST_INT when the mode parameter is VOIDmode.
10828 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10832 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10833 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10834 (bitwise_type_for_mode): Update accordingly.
10836 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10838 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10839 * stor-layout.c (mode_for_size_tree): Likewise.
10840 (mode_for_array): Update accordingly.
10841 (layout_decl): Likewise.
10842 (compute_record_mode): Likewise. Only set the mode once.
10844 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10846 * target.def (get_mask_mode): Change return type to opt_mode.
10848 * doc/tm.texi: Regenerate.
10849 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10850 * targhooks.c (default_get_mask_mode): Likewise.
10851 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10852 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10853 targetm.get_mask_mode.
10854 * tree.c (build_truth_vector_type): Likewise.
10856 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10858 * machmode.h (mode_for_vector): Return an opt_mode.
10859 * stor-layout.c (mode_for_vector): Likewise.
10860 (mode_for_int_vector): Update accordingly.
10861 (layout_type): Likewise.
10862 * config/i386/i386.c (emit_memmov): Likewise.
10863 (ix86_expand_set_or_movmem): Likewise.
10864 (ix86_expand_vector_init): Likewise.
10865 (ix86_get_mask_mode): Likewise.
10866 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10868 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10869 * expmed.c (extract_bit_field_1): Likewise.
10870 * expr.c (expand_expr_real_2): Likewise.
10871 * optabs-query.c (can_vec_perm_p): Likewise.
10872 (can_vec_mask_load_store_p): Likewise.
10873 * optabs.c (expand_vec_perm): Likewise.
10874 * targhooks.c (default_get_mask_mode): Likewise.
10875 * tree-vect-stmts.c (vectorizable_store): Likewise.
10876 (vectorizable_load): Likewise.
10877 (get_vectype_for_scalar_type_and_size): Likewise.
10879 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10881 * machmode.h (mode_for_int_vector): New function.
10882 * stor-layout.c (mode_for_int_vector): Likewise.
10883 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10884 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10885 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10886 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10887 (s390_expand_vcond): Likewise.
10889 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10891 * machmode.h (opt_machine_mode): New type.
10892 (opt_mode<T>): Allow construction from anything that can be
10894 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10895 (mode_for_size): Return an opt_machine_mode.
10896 * stor-layout.c (mode_for_size): Likewise.
10897 (mode_for_size_tree): Update call accordingly.
10898 (bitwise_mode_for_mode): Likewise.
10899 (make_fract_type): Likewise.
10900 (make_accum_type): Likewise.
10901 * caller-save.c (replace_reg_with_saved_mem): Update call
10903 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10904 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10905 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10906 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10907 * expmed.c (extract_bit_field_1): Likewise.
10908 * reload.c (get_secondary_mem): Likewise.
10909 * varasm.c (assemble_integer): Likewise.
10910 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10913 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10915 * machmode.h (decimal_float_mode_for_size): New function.
10916 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10917 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10918 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10919 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10920 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10921 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10923 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10925 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10926 (get_builtin_sync_mode): Likewise.
10927 (expand_ifn_atomic_compare_exchange): Likewise.
10928 (expand_builtin_atomic_clear): Likewise.
10929 (expand_builtin_atomic_test_and_set): Likewise.
10930 (fold_builtin_atomic_always_lock_free): Likewise.
10931 * calls.c (compute_argument_addresses): Likewise.
10932 (emit_library_call_value_1): Likewise.
10933 (store_one_arg): Likewise.
10934 * combine.c (combine_instructions): Likewise.
10935 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10936 * config/arm/arm.c (arm_function_value): Likewise.
10937 (aapcs_allocate_return_reg): Likewise.
10938 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10939 * config/i386/i386.c (construct_container): Likewise.
10940 (ix86_gimplify_va_arg): Likewise.
10941 (ix86_expand_sse_cmp): Likewise.
10942 (emit_memmov): Likewise.
10943 (emit_memset): Likewise.
10944 (expand_small_movmem_or_setmem): Likewise.
10945 (ix86_expand_pextr): Likewise.
10946 (ix86_expand_pinsr): Likewise.
10947 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10948 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10950 * config/mips/mips.c (mips_function_value_1) Likewise.
10951 (mips_block_move_straight): Likewise.
10952 (mips_expand_ins_as_unaligned_store): Likewise.
10953 * config/powerpcspe/powerpcspe.c
10954 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10955 (rs6000_darwin64_record_arg_flush): Likewise.
10956 * config/rs6000/rs6000.c
10957 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10958 (rs6000_darwin64_record_arg_flush): Likewise.
10959 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10960 (sparc_function_value_1): Likewise.
10961 * config/spu/spu.c (adjust_operand): Likewise.
10962 (spu_emit_branch_or_set): Likewise.
10963 (arith_immediate_p): Likewise.
10964 * emit-rtl.c (gen_lowpart_common): Likewise.
10965 * expr.c (expand_expr_real_1): Likewise.
10966 * function.c (assign_parm_setup_block): Likewise.
10967 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10968 * reload1.c (alter_reg): Likewise.
10969 * stor-layout.c (mode_for_vector): Likewise.
10970 (layout_type): Likewise.
10972 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10974 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10975 (spu_convert_move): Likewise.
10976 * lower-subreg.c (resolve_simple_move): Likewise.
10978 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10981 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10982 define_insn to a define_expand.
10983 (altivec_vsum2sws_direct): New define_insn.
10984 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10986 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10988 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10991 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10996 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10997 realign stack if stack alignment needed is less than incoming
11000 2017-09-05 Marek Polacek <polacek@redhat.com>
11003 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11006 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11008 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11010 2017-09-05 Richard Biener <rguenther@suse.de>
11012 PR tree-optimization/82084
11013 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11015 2017-09-05 Richard Biener <rguenther@suse.de>
11017 PR tree-optimization/82102
11018 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11020 2017-09-05 Martin Liska <mliska@suse.cz>
11022 PR tree-optimization/82032
11023 * tree-cfg.c (generate_range_test): New function.
11024 * tree-cfg.h (generate_range_test): Declared here.
11025 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11026 (cleanup_control_expr_graph): Use it.
11027 * tree-switch-conversion.c (try_switch_expansion): Remove
11029 (emit_case_nodes): Use generate_range_test.
11031 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11034 * config/i386/i386.md (*<btsc><mode>_mask): Add
11035 TARGET_USE_BT to insn constraint.
11036 (*btr<mode>_mask): Ditto.
11038 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11040 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11041 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11043 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11046 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11047 TARGET_NEON and TARGET_IWMMXT.
11048 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11049 TARGET_NEON and TARGET_IWMMXT.
11050 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11052 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11054 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11055 (ix86_rewrite_tls_address): Ditto.
11056 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11057 (ix86_rewrite_tls_address_1): Ditto.
11058 (ix86_rewrite_tls_address): Ditto.
11059 * config/i386/predicates.md (tls_address_pattern): New predicate.
11060 * config/i386/i386.md (TLS address splitter): New splitter.
11062 2017-09-04 Richard Biener <rguenther@suse.de>
11064 PR tree-optimization/82084
11065 * fold-const.h (can_native_encode_string_p): Declare.
11066 * fold-const.c (can_native_encode_string_p): Factor out from ...
11067 (native_encode_string): ... here.
11068 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11069 vectorizing stores from constants we later cannot handle.
11071 2017-09-04 Marek Polacek <polacek@redhat.com>
11074 * doc/invoke.texi: Update -Wtautological-compare documentation.
11076 2017-09-04 Jeff Law <law@redhat.com>
11078 PR tree-optimization/64910
11079 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11080 swap the first and last operand if the last is a constant.
11082 2017-09-04 Marek Polacek <polacek@redhat.com>
11085 * convert.c (do_narrow): When sanitizing signed integer overflows,
11086 bail out for signed types.
11087 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11089 2017-09-04 Richard Biener <rguenther@suse.de>
11091 PR tree-optimization/82060
11092 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11093 Move devirtualization after stmt folding and before EH/AB/noreturn
11094 cleanup to get the stmt refs canonicalized. Use a bool instead
11095 of gimple_modified_p since that doesn't work for NOPs. Schedule
11096 NOPs generated by folding for removal.
11098 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11099 Alan Hayward <alan.hayward@arm.com>
11100 David Sherwood <david.sherwood@arm.com>
11102 * coretypes.h (pad_direction): New enum.
11103 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11104 (FUNCTION_ARG_PADDING): Likewise.
11105 * target.def (function_arg_padding): New hook.
11106 * targhooks.h (default_function_arg_padding): Declare.
11107 * targhooks.c (default_function_arg_padding): New function.
11108 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11109 (TARGET_FUNCTION_ARG_PADDING): ...this.
11110 * doc/tm.texi: Regenerate.
11111 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11112 instead of direction.
11113 (compute_argument_addresses): Likewise.
11114 (load_register_parameters): Likewise.
11115 (emit_library_call_value_1): Likewise.
11116 (store_one_arg): Use targetm.calls.function_arg_padding instead
11117 of FUNCTION_ARG_PADDING.
11118 (must_pass_in_stack_var_size_or_pad): Likewise.
11119 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11120 (emit_group_store): Likewise.
11121 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11122 instead of FUNCTION_ARG_PADDING.
11123 (emit_push_insn): Likewise, and propagate enum change throughout
11125 * function.h (direction): Delete.
11126 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11128 * function.c (assign_parm_find_stack_rtl): Likewise.
11129 (assign_parm_setup_block_p): Likewise.
11130 (assign_parm_setup_block): Likewise.
11131 (gimplify_parameters): Likewise.
11132 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11133 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11135 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11136 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11137 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11138 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11139 (aarch64_function_arg_padding): ...this new function.
11140 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11141 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11142 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11143 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11144 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11145 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11146 (arm_pad_arg_upward): Replace with...
11147 (arm_function_arg_padding): ...this new function.
11148 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11150 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11151 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11152 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11153 (ia64_hpux_function_arg_padding): Replace with...
11154 (ia64_function_arg_padding): ...this new function. Use pad_direction
11155 instead of direction. Check for TARGET_HPUX.
11156 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11157 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11158 (iq2000_function_arg_padding): New function.
11159 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11160 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11161 (mips_function_arg_padding): ...this new function.
11162 (mips_pad_reg_upward): Update accordingly.
11163 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11164 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11165 targetm.calls.function_arg_padding.
11166 (FUNCTION_ARG_PADDING): Delete.
11167 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11168 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11169 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11170 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11171 (nios2_block_reg_padding): Return pad_direction instead of direction.
11172 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11173 instead of direction.
11174 (nios2_function_arg_padding): Likewise. Make static.
11175 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11176 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11177 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11178 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11179 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11180 (pa_function_arg_padding): Make static. Return pad_direction instead
11182 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11183 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11184 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11185 instead of direction. Use targetm.calls.function_arg_padding.
11186 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11187 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11188 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11189 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11190 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11192 (function_arg_padding): Rename to...
11193 (rs6000_function_arg_padding): ...this. Make static. Return
11194 pad_direction instead of direction.
11195 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11196 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11197 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11198 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11199 instead of direction. Use targetm.calls.function_arg_padding.
11200 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11201 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11202 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11203 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11204 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11205 (function_arg_padding): Rename to...
11206 (rs6000_function_arg_padding): ...this. Make static. Return
11207 pad_direction instead of direction.
11208 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11209 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11210 * config/s390/s390.c (s390_function_arg_padding): New function.
11211 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11212 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11213 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11214 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11215 (function_arg_padding): Rename to...
11216 (sparc_function_arg_padding): ...this. Make static. Return
11217 pad_direction instead of direction.
11218 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11219 * config/spu/spu.c (spu_function_arg_padding): New function.
11220 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11221 * system.h (FUNCTION_ARG_PADDING): Poison.
11223 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11224 Alan Hayward <alan.hayward@arm.com>
11225 David Sherwood <david.sherwood@arm.com>
11227 * target.def (modes_tieable_p): New hook.
11228 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11229 (TARGET_MODES_TIEABLE_P): ...this.
11230 * doc/tm.texi.in: Regenerate.
11231 * hooks.h (hook_bool_mode_mode_true): Declare.
11232 * hooks.c (hook_bool_mode_mode_true): New function.
11233 * combine.c (subst): Use targetm.modes_tieable_p instead of
11235 * dse.c (find_shift_sequence): Likewise.
11236 * expmed.c (extract_low_bits): Likewise.
11237 * lower-subreg.c: Include target.h.
11238 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11240 * rtlanal.c (rtx_cost): Likewise.
11241 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11242 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11243 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11244 (TARGET_MODES_TIEABLE_P): Redefine.
11245 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11246 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11247 (TARGET_MODES_TIEABLE_P): Redefine.
11248 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11249 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11250 (arc_modes_tieable_p): New function.
11251 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11252 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11253 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11254 (arm_modes_tieable_p): Make static.
11255 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11256 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11257 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11258 (TARGET_MODES_TIEABLE_P): Redefine.
11259 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11260 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11261 (TARGET_MODES_TIEABLE_P): Redefine.
11262 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11263 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11264 (cr16_modes_tieable_p): New function.
11265 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11266 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11267 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11268 (TRULY_NOOP_TRUNCATION): Update comment.
11269 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
11270 (TRULY_NOOP_TRUNCATION): Update comment.
11271 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
11272 (frv_modes_tieable_p): New function.
11273 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
11274 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
11275 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
11276 (TARGET_MODES_TIEABLE_P): Redefine.
11277 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
11278 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
11279 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
11280 (TARGET_MODES_TIEABLE_P): Redefine.
11281 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
11282 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
11283 (ia64_modes_tieable_p): New function.
11284 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
11285 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
11286 (iq2000_modes_tieable_p): New function.
11287 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
11288 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
11289 (lm32_modes_tieable_p): New function.
11290 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
11291 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
11292 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
11293 (TARGET_MODES_TIEABLE_P): Redefine.
11294 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
11295 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
11296 (m32r_modes_tieable_p): New function.
11297 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
11298 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
11299 (m68k_modes_tieable_p): New function.
11300 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
11301 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
11302 (mcore_modes_tieable_p): New function.
11303 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11304 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11306 (TARGET_MODES_TIEABLE_P): Redefine.
11307 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11308 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11309 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11310 (TARGET_MODES_TIEABLE_P): Redefine.
11311 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11312 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11313 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11314 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11315 (mn10300_modes_tieable_p): ...this and make static.
11316 (TARGET_MODES_TIEABLE_P): Redefine.
11317 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11318 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11319 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11320 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11321 (msp430_modes_tieable_p): Make static.
11322 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11323 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11324 (TARGET_MODES_TIEABLE_P): Redefine.
11325 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11326 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11327 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11328 (TARGET_MODES_TIEABLE_P): Redefine.
11329 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11330 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11331 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11332 (TARGET_MODES_TIEABLE_P): Redefine.
11333 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11334 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11335 (pdp11_modes_tieable_p): New function.
11336 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11337 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11338 (rs6000_modes_tieable_p): New function.
11339 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11340 * config/powerpcspe/powerpcspe.md: Update comment.
11341 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11342 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11343 (TARGET_MODES_TIEABLE_P): Redefine.
11344 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11345 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11346 (rl78_modes_tieable_p): New function.
11347 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11348 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11349 (rs6000_modes_tieable_p): New function.
11350 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11351 * config/rs6000/rs6000.md: Update comment.
11352 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11353 * config/rx/rx.c (rx_modes_tieable_p): New function.
11354 (TARGET_MODES_TIEABLE_P): Redefine.
11355 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11356 * config/s390/s390.c (s390_modes_tieable_p): New function.
11357 (TARGET_MODES_TIEABLE_P): Redefine.
11358 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11359 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11360 (sh_modes_tieable_p): New function.
11361 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11362 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11363 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11364 (sparc_modes_tieable_p): Make static.
11365 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11366 * config/spu/spu.c (spu_modes_tieable_p): New function.
11367 (TARGET_MODES_TIEABLE_P): Redefine.
11368 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11369 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11370 (TARGET_MODES_TIEABLE_P): Redefine.
11371 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11372 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11373 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11374 * config/v850/v850.c (v850_modes_tieable_p): New function.
11375 (TARGET_MODES_TIEABLE_P): Redefine.
11376 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11377 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11378 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11379 (visium_modes_tieable_p): New function.
11380 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11381 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11382 (xtensa_modes_tieable_p): New function.
11383 * system.h (MODES_TIEABLE_P): Poison.
11385 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11386 Alan Hayward <alan.hayward@arm.com>
11387 David Sherwood <david.sherwood@arm.com>
11389 * target.def (hard_regno_mode_ok): New hook.
11390 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11391 (TARGET_HARD_REGNO_MODE_OK): ...this.
11392 * doc/tm.texi.in: Regenerate.
11393 * hooks.h (hook_bool_uint_mode_true): Declare.
11394 * hooks.c (hook_bool_uint_mode_true): New function.
11395 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11396 HARD_REGNO_MODE_OK.
11397 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11398 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11399 instead of HARD_REGNO_MODE_OK.
11400 * caller-save.c: Include target.h.
11401 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11402 HARD_REGNO_MODE_OK.
11403 * combine.c (can_combine_p): Likewise.
11404 (combinable_i3pat): Likewise.
11405 (can_change_dest_mode): Likewise.
11406 * expr.c (init_expr_target): Likewise.
11407 (convert_move): Likewise.
11408 (convert_modes): Likewise.
11409 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11410 (setup_prohibited_mode_move_regs): Likewise.
11411 * ira.h (target_ira): Likewise.
11412 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11413 * lra-constraints.c (process_alt_operands): Likewise.
11414 (split_reg): Likewise.
11415 * recog.c (peep2_find_free_register): Likewise.
11416 * ree.c (combine_reaching_defs): Likewise.
11417 * regcprop.c (maybe_mode_change): Likewise.
11418 * reginfo.c (init_reg_sets_1): Likewise.
11419 (choose_hard_reg_mode): Likewise.
11420 (simplifiable_subregs): Likewise.
11421 * regrename.c (check_new_reg_p): Likewise.
11422 * reload.c (find_valid_class): Likewise.
11423 (find_valid_class_1): Likewise.
11424 (reload_inner_reg_of_subreg): Likewise.
11425 (push_reload): Likewise.
11426 (combine_reloads): Likewise.
11427 (find_dummy_reload): Likewise.
11428 (find_reloads): Likewise.
11429 * reload1.c (find_reg): Likewise.
11430 (set_reload_reg): Likewise.
11431 (allocate_reload_reg): Likewise.
11432 (choose_reload_regs): Likewise.
11433 (reload_adjust_reg_for_temp): Likewise.
11434 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11435 (simplify_subreg_regno): Likewise.
11436 * sel-sched.c (init_regs_for_mode): Likewise.
11437 * varasm.c (make_decl_rtl): Likewise.
11438 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11439 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11440 HARD_REGNO_MODE_OK.
11441 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11442 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11444 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11445 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11447 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11448 (arc_mode_class): Delete.
11449 (HARD_REGNO_MODE_OK): Delete.
11450 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11451 (arc_hard_regno_mode_ok): Rename old array to...
11452 (arc_hard_regno_mode_ok_modes): ...this.
11453 (arc_conditional_register_usage): Update accordingly.
11454 (arc_mode_class): Make static.
11455 (arc_hard_regno_mode_ok): New function.
11456 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11457 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11458 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11459 (arm_hard_regno_mode_ok): Make static.
11460 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11461 HARD_REGNO_MODE_OK.
11462 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11463 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11464 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11466 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11467 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11468 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11469 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11470 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11471 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11472 * config/bfin/predicates.md (valid_reg_operand): Use
11473 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11474 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11475 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11476 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11477 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11478 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11479 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11480 (cr16_hard_regno_mode_ok): Make static and return a bool.
11481 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11482 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11483 (cris_hard_regno_mode_ok): New function.
11484 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11485 (epiphany_mode_class): Delete.
11486 (HARD_REGNO_MODE_OK): Delete.
11487 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11488 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11489 (hard_regno_mode_ok): Rename to...
11490 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11491 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11492 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11493 HARD_REGNO_MODE_OK.
11494 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11495 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11496 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11497 (frv_hard_regno_mode_ok): Make static and return a bool.
11498 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11499 HARD_REGNO_MODE_OK.
11500 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11501 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11502 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11503 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11505 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11506 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11507 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11508 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11511 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11512 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11513 (ia64_hard_regno_mode_ok): New function.
11514 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11515 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11516 (iq2000_hard_regno_mode_ok): New function.
11517 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11518 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11519 (lm32_hard_regno_mode_ok): New function.
11520 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11521 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11522 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11523 instead of HARD_REGNO_MODE_OK.
11524 (m32c_hard_regno_ok): Rename to...
11525 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11526 (m32c_cannot_change_mode_class): Update accordingly.
11527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11528 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11529 (m32r_mode_class): Delete.
11530 (HARD_REGNO_MODE_OK): Delete.
11531 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11532 (m32r_hard_regno_mode_ok): Rename to...
11533 (m32r_hard_regno_modes): ...this.
11534 (m32r_mode_class): Make static.
11535 (m32r_hard_regno_mode_ok): New function.
11536 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11537 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11538 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11539 (m68k_hard_regno_mode_ok): Make static.
11540 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11541 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11542 (mcore_hard_regno_mode_ok): New function.
11543 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11544 (HARD_REGNO_MODE_OK): Delete.
11545 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11547 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11548 (microblaze_hard_regno_mode_ok): New function.
11549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11550 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11551 (mips_hard_regno_mode_ok): Delete.
11552 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11553 (mips_hard_regno_mode_ok_p): ...this and make static.
11554 (mips_hard_regno_mode_ok_p): Rename to...
11555 (mips_hard_regno_mode_ok_uncached): ...this.
11556 (mips_hard_regno_mode_ok): New function.
11557 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11558 of HARD_REGNO_MODE_OK.
11559 (mips_option_override): Update after above name changes.
11560 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11561 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11562 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11563 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11564 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11566 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11567 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11568 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11569 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11570 (msp430_hard_regno_mode_ok): Make static and return a bool.
11571 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11572 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11573 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11576 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11577 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11578 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11579 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11580 (PA_HARD_REGNO_MODE_OK): ...this
11581 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11582 (PA_HARD_REGNO_MODE_OK): ...this.
11583 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11584 (pa_hard_regno_mode_ok): New function.
11585 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11586 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11587 (pdp11_hard_regno_mode_ok): New function.
11588 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11589 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11591 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11594 (rs6000_hard_regno_mode_ok): Rename to...
11595 (rs6000_hard_regno_mode_ok_uncached): ...this.
11596 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11597 (rs6000_hard_regno_mode_ok): New function.
11598 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11599 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11600 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11601 (riscv_hard_regno_mode_ok): ...this and make static.
11602 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11603 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11604 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11605 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11606 (rl78_hard_regno_mode_ok): Make static and return bool.
11607 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11608 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11610 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11612 (rs6000_hard_regno_mode_ok): Rename to...
11613 (rs6000_hard_regno_mode_ok_uncached): ...this.
11614 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11615 (rs6000_hard_regno_mode_ok): New function.
11616 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11617 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11619 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11620 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11621 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11622 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11623 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11624 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11625 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11626 (sh_hard_regno_mode_ok): Make static.
11627 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11628 instead of HARD_REGNO_MODE_OK.
11629 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11630 (sparc_mode_class): Delete.
11631 (HARD_REGNO_MODE_OK): Delete.
11632 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11633 (hard_regno_mode_classes): Make static.
11634 (sparc_mode_class): Likewise.
11635 (sparc_hard_regno_mode_ok): New function.
11636 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11637 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11638 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11641 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11642 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11643 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11644 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11646 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11647 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11648 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11649 (visium_hard_regno_mode_ok): New function.
11650 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11651 instead of HARD_REGNO_MODE_OK.
11652 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11653 (HARD_REGNO_MODE_OK): Delete.
11654 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11655 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11656 (xtensa_option_override): Update accordingly.
11657 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11658 (xtensa_hard_regno_mode_ok): New function.
11659 * system.h (HARD_REGNO_MODE_OK): Poison.
11661 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11662 Alan Hayward <alan.hayward@arm.com>
11663 David Sherwood <david.sherwood@arm.com>
11665 * target.def (hard_regno_call_part_clobbered): New hook.
11666 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11667 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11668 * doc/tm.texi: Regenerate.
11669 * hooks.h (hook_bool_uint_mode_false): Declare.
11670 * hooks.c (hook_bool_uint_mode_false): New function.
11671 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11672 * cselib.c (cselib_process_insn): Use
11673 targetm.hard_regno_call_part_clobbered instead of
11674 HARD_REGNO_CALL_PART_CLOBBERED.
11675 * ira-conflicts.c (ira_build_conflicts): Likewise.
11676 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11677 * lra-constraints.c (need_for_call_save_p): Likewise.
11678 * lra-lives.c: Include target.h.
11679 (check_pseudos_live_through_calls): Use
11680 targetm.hard_regno_call_part_clobbered instead of
11681 HARD_REGNO_CALL_PART_CLOBBERED.
11682 * regcprop.c: Include target.h.
11683 (copyprop_hardreg_forward_1): Use
11684 targetm.hard_regno_call_part_clobbered instead of
11685 HARD_REGNO_CALL_PART_CLOBBERED.
11686 * reginfo.c (choose_hard_reg_mode): Likewise.
11687 * regrename.c (check_new_reg_p): Likewise.
11688 * reload.c (find_equiv_reg): Likewise.
11689 * reload1.c (emit_reload_insns): Likewise.
11690 * sched-deps.c (deps_analyze_insn): Likewise.
11691 * sel-sched.c (init_regs_for_mode): Likewise.
11692 (mark_unavailable_hard_regs): Likewise.
11693 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11694 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11695 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11697 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11698 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11699 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11701 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11703 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11704 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11705 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11707 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11708 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11709 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11711 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11712 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11714 * config/powerpcspe/powerpcspe.c
11715 (rs6000_hard_regno_call_part_clobbered): New function.
11716 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11717 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11718 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11720 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11721 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11722 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11724 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11725 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11726 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11728 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11729 Alan Hayward <alan.hayward@arm.com>
11730 David Sherwood <david.sherwood@arm.com>
11732 * rtl.h (subreg_memory_offset): Declare.
11733 * emit-rtl.c (subreg_memory_offset): New function.
11734 * expmed.c (store_bit_field_1): Use it.
11735 * expr.c (undefined_operand_subword_p): Likewise.
11736 * simplify-rtx.c (simplify_subreg): Likewise.
11738 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11740 PR rtl-optimization/57448
11743 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11744 using atomic_load pattern.
11745 (expand_atomic_store): Likewise.
11747 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11750 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11751 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11752 handling. Use replace_call_with_value with NULL instead of
11753 gsi_replace, unlink_stmt_vdef and release_defs.
11755 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11758 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11760 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11763 * rtl.h (emit_library_call_value_1): Declare.
11764 (emit_library_call): Replace declaration with a series of overloads.
11765 Remove the parameter count argument.
11766 (emit_library_call_value): Likewise.
11767 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11768 with an "rtx_mode_t *".
11769 (emit_library_call_value): Delete.
11770 (emit_library_call): Likewise.
11771 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11772 (asan_emit_allocas_unpoison): Likewise.
11773 * builtins.c (expand_builtin_powi): Likewise.
11774 (expand_asan_emit_allocas_unpoison): Likewise.
11775 * cfgexpand.c (expand_main_function): Likewise.
11776 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11777 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11778 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11779 * config/arm/arm.c (arm_trampoline_init): Likewise.
11780 (arm_call_tls_get_addr): Likewise.
11781 (arm_expand_divmod_libfunc): Likewise.
11782 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11783 (smulsi3_highpart): Likewise.
11784 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11785 (c6x_expand_compare): Likewise.
11786 (c6x_expand_movmem): Likewise.
11787 * config/frv/frv.c (frv_trampoline_init): Likewise.
11788 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11789 (ix86_expand_divmod_libfunc): Likewise.
11790 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11791 (ia64_expand_compare): Likewise.
11792 (ia64_profile_hook): Likewise.
11793 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11794 (nonlocal_goto): Likewise.
11795 (restore_stack_nonlocal): Likewise.
11796 * config/m32r/m32r.c (block_move_call): Likewise.
11797 (m32r_trampoline_init): Likewise.
11798 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11799 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11800 (m68k_call_m68k_read_tp): Likewise.
11801 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11802 (microblaze_expand_divide): Likewise.
11803 * config/mips/mips.h (mips_args): Likewise.
11804 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11805 (MIPS_ICACHE_SYNC): Likewise.
11806 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11807 (nios2_trampoline_init): Likewise.
11808 * config/pa/pa.c (hppa_tls_call): Likewise.
11809 (pa_trampoline_init): Likewise.
11810 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11811 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11812 (expand_strn_compare): Likewise.
11813 (rs6000_generate_compare): Likewise.
11814 (rs6000_expand_float128_convert): Likewise.
11815 (output_profile_hook): Likewise.
11816 (rs6000_trampoline_init): Likewise.
11817 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11818 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11819 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11820 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11821 (rs6000_generate_compare): Likewise.
11822 (rs6000_expand_float128_convert): Likewise.
11823 (output_profile_hook): Likewise.
11824 (rs6000_trampoline_init): Likewise.
11825 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11826 * config/sh/sh.c (sh_trampoline_init): Likewise.
11827 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11828 (sparc_emit_float_lib_cmp): Likewise.
11829 (sparc32_initialize_trampoline): Likewise.
11830 (sparc64_initialize_trampoline): Likewise.
11831 (sparc_profile_hook): Likewise.
11832 * config/spu/spu.c (ea_load_store): Likewise.
11833 * config/spu/spu.md (floatunssidf2): Likewise.
11834 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11835 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11836 * config/visium/visium.c (expand_block_move_4): Likewise.
11837 (expand_block_move_2): Likewise.
11838 (expand_block_move_1): Likewise.
11839 (expand_block_set_4): Likewise.
11840 (expand_block_set_2): Likewise.
11841 (expand_block_set_1): Likewise.
11842 (visium_trampoline_init): Likewise.
11843 (visium_profile_hook): Likewise.
11844 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11845 (xtensa_setup_frame_addresses): Likewise.
11846 (xtensa_trampoline_init): Likewise.
11847 * except.c (sjlj_emit_function_enter): Likewise.
11848 (sjlj_emit_function_exit): Likewise.
11849 * explow.c (allocate_dynamic_stack_space): Likewise.
11850 (probe_stack_range): Likewise.
11851 * expr.c (convert_mode_scalar): Likewise.
11852 * optabs.c (expand_binop): Likewise.
11853 (expand_twoval_binop_libfunc): Likewise.
11854 (expand_unop): Likewise.
11855 (prepare_cmp_insn): Likewise.
11856 (prepare_float_lib_cmp): Likewise.
11857 (expand_float): Likewise.
11858 (expand_fix): Likewise.
11859 (expand_fixed_convert): Likewise.
11860 (maybe_emit_sync_lock_test_and_set): Likewise.
11861 (expand_atomic_compare_and_swap): Likewise.
11862 (expand_mem_thread_fence): Likewise.
11863 (expand_atomic_fetch_op): Likewise.
11865 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11867 * doc/generic.texi (OpenACC): Adjust URL.
11868 * doc/invoke.texi (C Dialect Options): Ditto.
11870 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11872 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11873 predicate for operand 1. Add (m,<S>) constraint.
11874 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11875 Prevent memory operand 1 with register operand 2.
11877 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11879 PR rtl-optimization/82024
11880 * combine.c (try_combine): If the combination result is a PARALLEL,
11881 and we only need to retain the SET in there that would be placed
11882 at I2, check that we can place that at I3 instead, before doing so.
11884 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11887 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11889 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11890 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11893 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11894 Jeff Law <law@redhat.com>
11896 * varasm.c (bss_initializer_p): Do not put constants into .bss
11897 (categorize_decl_for_section): Handle bss_initializer_p returning
11898 false when DECL_INITIAL is NULL.
11900 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11903 * config/s390/s390.c (s390_can_inline_p): New function.
11905 2017-09-01 Jeff Law <law@redhat.com>
11907 PR tree-optimization/82052
11908 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11909 Always initialize the returned slot after a hash table miss
11910 when INSERT is true.
11912 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11914 * config/s390/s390.md (mem_signal_fence): Remove.
11915 * doc/md.texi (mem_signal_fence): Remove.
11916 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11918 * target-insns.def (mem_signal_fence): Remove.
11920 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11923 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11926 * asan.c (create_odr_indicator): Strip name encoding from assembler
11927 name before appending it after __odr_asan_.
11929 2017-09-01 Martin Liska <mliska@suse.cz>
11931 PR tree-optimization/82059
11932 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11933 frequency only when an edge is redirected.
11935 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11937 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11938 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11939 (arc_conditional_register_usage): Remove ARC600 lp_count
11941 (arc_file_start): Emit Tag_ARC_CPU_variation.
11942 (arc_can_use_doloop_p): New conditions to use ZOLs.
11943 (hwloop_fail): New function.
11944 (hwloop_optimize): Likewise.
11945 (hwloop_pattern_reg): Likewise.
11946 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11947 (arc_reorg_loops): New function, calls reorg_loops.
11948 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11949 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11951 (arc_loop_hazard): Remove function, functionality moved into
11953 (arc_hazard): Remove arc_loop_hazard call.
11954 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11955 into hwloop_optimize.
11956 (arc_label_align): Remove ZOL handling.
11957 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11958 * config/arc/arc.md (doloop_begin): Remove pattern.
11959 (doloop_begin_i): Likewise.
11960 (doloop_end_i): Likewise.
11961 (doloop_fallback): Likewise.
11962 (doloop_fallback_m): Likewise.
11963 (doloop_end): Reimplement expand.
11964 (arc_lp): New pattern for LP instruction.
11965 (loop_end): New pattern.
11966 (loop_fail): Likewise.
11967 (decrement_and_branch_until_zero): Likewise.
11968 * config/arc/arc.opt (mlpc-width): New option.
11969 * doc/invoke.texi (mlpc-width): Document option.
11971 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11973 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11974 (arc_ccfsm_advance): Fix checking for delay slots.
11975 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11977 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11979 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11981 (movhi_insn): Update store instruction constraint which are saving
11982 6-bit short immediates.
11983 (movsi_insn): Consider also short scaled load operations.
11984 (zero_extendhisi2_i): Use Usd constraint instead of T.
11985 (extendhisi2_i): Add q constraint.
11986 (arc_clzsi2): Add type and length attributes.
11987 (arc_ctzsi2): Likewise.
11988 * config/arc/constraints.md (Usc): Update constraint, the
11989 assembler can parse two relocations for a single instruction.
11991 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11993 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11994 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11996 2017-08-31 Olivier Hainque <hainque@adacore.com>
11998 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11999 match as powerpc-wrs-vxworks*.
12001 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12003 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12004 register constraint for by-element operand.
12005 (aarch64_mls_elt_merge<mode>): Likewise.
12007 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12009 * config/arc/arc.c (arc_can_follow_jump): Check for short
12012 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12014 * config.gcc: Use g.opt for arc.
12015 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12016 functionality moved to ...
12017 (legitimate_scaled_address_p): New function, ...here.
12018 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12019 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12020 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12022 (arc_override_options): Handle G option.
12023 (arc_output_pic_addr_const): Correct function definition.
12024 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12025 (arc_decl_anon_ns_mem_p): Delete.
12026 (arc_in_small_data_p): Overhaul this function to take into
12027 consideration the value given via G option.
12028 (arc_rewrite_small_data_1): Renamed and corrected old
12029 arc_rewrite_small_data function.
12030 (arc_rewrite_small_data): New function.
12031 (small_data_pattern): Don't use pic_offset_table_rtx.
12032 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12033 * config/arc/simdext.md (movmisalignv2hi): Use
12034 prepare_move_operands function.
12036 (movmisalign*): Likewise.
12037 * doc/invoke.texi (ARC options): Document -G option.
12039 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12041 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12043 * config/arc/arc.c (arc_print_operand): Output scalled address for
12044 sdata whenever is possible.
12045 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12046 load/stores are available.
12047 (compact_sda_memory_operand): Check for the alignment required by
12048 code density instructions.
12049 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12051 * config/arc/constraints.md (Usd): Update constraint.
12052 (Us0): New constraint.
12053 (Usc): Update constraint.
12055 2017-08-31 Richard Biener <rguenther@suse.de>
12057 PR middle-end/82054
12058 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12059 function only once.
12061 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12063 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12064 Resize type_signature.
12066 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12067 Alan Hayward <alan.hayward@arm.com>
12068 David Sherwood <david.sherwood@arm.com>
12070 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12071 subregs whose inner modes can be stored in GPRs.
12072 (aarch64_classify_index): Likewise.
12074 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12075 Alan Hayward <alan.hayward@arm.com>
12076 David Sherwood <david.sherwood@arm.com>
12078 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12079 (V_INT_EQUIV): ...this.
12080 (v_cmp_result): Rename to...
12081 (v_int_equiv): ...this.
12082 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12083 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12084 (copysign<mode>3): Likewise.
12085 (aarch64_simd_bsl<mode>_internal): Likewise.
12086 (aarch64_simd_bsl<mode>): Likewise.
12087 (vec_cmp<mode><mode>): Likewise.
12088 (vcond<mode><mode>): Likewise.
12089 (vcond<v_cmp_mixed><mode>): Likewise.
12090 (vcondu<mode><v_cmp_mixed>): Likewise.
12091 (aarch64_cm<optab><mode>): Likewise.
12092 (aarch64_cmtst<mode>): Likewise.
12093 (aarch64_fac<optab><mode>): Likewise.
12094 (vec_perm_const<mode>): Likewise.
12095 (vcond_mask_<mode><v_cmp_result>): Rename to...
12096 (vcond_mask_<mode><v_int_equiv>): ...this.
12097 (vec_cmp<mode><v_cmp_result>): Rename to...
12098 (vec_cmp<mode><v_int_equiv>): ...this.
12100 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12101 Alan Hayward <alan.hayward@arm.com>
12102 David Sherwood <david.sherwood@arm.com>
12104 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12106 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12107 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12108 (UNSPEC_LD4_DREG): New unspecs.
12109 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12110 (aarch64_ld2<mode>_dreg_be): Replace with...
12111 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12113 (aarch64_ld3<mode>_dreg_le)
12114 (aarch64_ld3<mode>_dreg_be): Replace with...
12115 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12117 (aarch64_ld4<mode>_dreg_le)
12118 (aarch64_ld4<mode>_dreg_be): Replace with...
12119 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12122 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12124 PR tree-optimization/81987
12125 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12126 insert an initializer in a location not dominated by the stride
12129 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12131 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12132 on the entire header of the finally block in the fallthru case.
12134 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12136 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12138 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12140 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12141 rs6000_emit_move_from_cr and call renamed function.
12142 (rs6000_emit_prologue): Call renamed functions.
12143 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12144 movesi_from_cr, remove volatile CRs.
12146 2017-08-30 Jon Beniston <jon@beniston.com>
12147 Richard Biener <rguenther@suse.de>
12149 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12150 of VECTOR_MODE_P check.
12151 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12152 element vector types.
12154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12156 * df.h (df_read_modify_subreg_p): Remove in favor of...
12157 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12158 const_rtx instead of an rtx.
12159 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12160 * df-problems.c (df_word_lr_mark_ref): Likewise.
12161 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12162 (mark_pseudo_reg_dead): Likewise.
12163 (mark_ref_dead): Likewise.
12164 * reginfo.c (init_subregs_of_mode): Likewise.
12165 * sched-deps.c (sched_analyze_1): Likewise.
12166 * df-scan.c (df_def_record_1): Likewise.
12167 (df_uses_record): Likewise.
12168 (df_read_modify_subreg_p): Remove in favor of...
12169 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12170 const_rtx instead of an rtx.
12172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * rtl.h (partial_subreg_p): New function.
12177 * caller-save.c (save_call_clobbered_regs): Use it.
12178 * calls.c (expand_call): Likewise.
12179 * combine.c (combinable_i3pat): Likewise.
12180 (simplify_set): Likewise.
12181 (make_extraction): Likewise.
12182 (make_compound_operation_int): Likewise.
12183 (gen_lowpart_or_truncate): Likewise.
12184 (force_to_mode): Likewise.
12185 (make_field_assignment): Likewise.
12186 (reg_truncated_to_mode): Likewise.
12187 (record_truncated_value): Likewise.
12188 (move_deaths): Likewise.
12189 * cse.c (record_jump_cond): Likewise.
12190 (cse_insn): Likewise.
12191 * cselib.c (cselib_lookup_1): Likewise.
12192 * expmed.c (extract_bit_field_using_extv): Likewise.
12193 * function.c (assign_parm_setup_reg): Likewise.
12194 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12195 * ira-build.c (create_insn_allocnos): Likewise.
12196 * lra-coalesce.c (merge_pseudos): Likewise.
12197 * lra-constraints.c (match_reload): Likewise.
12198 (simplify_operand_subreg): Likewise.
12199 (curr_insn_transform): Likewise.
12200 * lra-lives.c (process_bb_lives): Likewise.
12201 * lra.c (new_insn_reg): Likewise.
12202 (lra_substitute_pseudo): Likewise.
12203 * regcprop.c (mode_change_ok): Likewise.
12204 (maybe_mode_change): Likewise.
12205 (copyprop_hardreg_forward_1): Likewise.
12206 * reload.c (push_reload): Likewise.
12207 (find_reloads): Likewise.
12208 (find_reloads_subreg_address): Likewise.
12209 * reload1.c (alter_reg): Likewise.
12210 (eliminate_regs_1): Likewise.
12211 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12213 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12215 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12216 back to if statements, including unpack.
12218 2017-08-30 Martin Liska <mliska@suse.cz>
12220 PR inline-asm/82001
12221 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12223 (func_checker::compare_asm_inputs_outputs): ... this function.
12224 (func_checker::compare_gimple_asm): Use the function to compare
12225 also ASM constrains.
12226 * ipa-icf-gimple.h: Rename the function.
12228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12229 Alan Hayward <alan.hayward@arm.com>
12230 David Sherwood <david.sherwood@arm.com>
12232 * coretypes.h (complex_mode): New type.
12233 * gdbhooks.py (build_pretty_printer): Handle it.
12234 * machmode.h (complex_mode): New class.
12235 (complex_mode::includes_p): New function.
12236 (is_complex_int_mode): Likewise.
12237 (is_complex_float_mode): Likewise.
12238 * genmodes.c (get_mode_class): Handle complex mode classes.
12239 * function.c (expand_function_end): Use is_complex_int_mode.
12241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12242 Alan Hayward <alan.hayward@arm.com>
12243 David Sherwood <david.sherwood@arm.com>
12245 * coretypes.h (scalar_mode_pod): New typedef.
12246 * gdbhooks.py (build_pretty_printer): Handle it.
12247 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12248 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12249 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12250 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12251 as_a <scalar_mode>.
12253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12254 Alan Hayward <alan.hayward@arm.com>
12255 David Sherwood <david.sherwood@arm.com>
12257 * machmode.h (mode_for_vector): Take a scalar_mode instead
12259 * stor-layout.c (mode_for_vector): Likewise.
12260 * explow.c (promote_mode): Use as_a <scalar_mode>.
12261 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * target.def (preferred_simd_mode): Take a scalar_mode
12268 instead of a machine_mode.
12269 * targhooks.h (default_preferred_simd_mode): Likewise.
12270 * targhooks.c (default_preferred_simd_mode): Likewise.
12271 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
12272 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
12273 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
12274 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
12275 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
12276 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
12277 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
12278 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
12280 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
12281 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
12282 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
12283 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
12284 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
12285 * doc/tm.texi: Regenerate.
12286 * optabs-query.c (can_vec_mask_load_store_p): Return false for
12289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12290 Alan Hayward <alan.hayward@arm.com>
12291 David Sherwood <david.sherwood@arm.com>
12293 * target.def (scalar_mode_supported_p): Take a scalar_mode
12294 instead of a machine_mode.
12295 * targhooks.h (default_scalar_mode_supported_p): Likewise.
12296 * targhooks.c (default_scalar_mode_supported_p): Likewise.
12297 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
12298 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12299 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
12300 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
12301 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
12302 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12303 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12304 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12305 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12306 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12307 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12308 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12310 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12311 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12312 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12313 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12314 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12316 * doc/tm.texi: Regenerate.
12318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12319 Alan Hayward <alan.hayward@arm.com>
12320 David Sherwood <david.sherwood@arm.com>
12322 * coretypes.h (opt_scalar_mode): New typedef.
12323 * gdbhooks.py (build_pretty_printers): Handle it.
12324 * machmode.h (mode_iterator::get_2xwider): Add overload for
12326 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12328 * expr.c (convert_mode_scalar): Likewise.
12329 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12330 * optabs.c (expand_float): Likewise.
12331 (expand_fix): Likewise.
12332 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12335 Alan Hayward <alan.hayward@arm.com>
12336 David Sherwood <david.sherwood@arm.com>
12338 * optabs.c (expand_float): Explicitly check for scalars before
12339 using a branching expansion.
12340 (expand_fix): Likewise.
12342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12343 Alan Hayward <alan.hayward@arm.com>
12344 David Sherwood <david.sherwood@arm.com>
12346 * expr.c (convert_mode): Split scalar handling out into...
12347 (convert_mode_scalar): ...this new function. Treat the modes
12350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12351 Alan Hayward <alan.hayward@arm.com>
12352 David Sherwood <david.sherwood@arm.com>
12354 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12356 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12359 Alan Hayward <alan.hayward@arm.com>
12360 David Sherwood <david.sherwood@arm.com>
12362 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12363 rather than a machine_mode.
12364 (fixed_from_string): Likewise.
12365 (fixed_convert): Likewise.
12366 (fixed_convert_from_int): Likewise.
12367 (fixed_convert_from_real): Likewise.
12368 (real_convert_from_fixed): Likewise.
12369 * fixed-value.c (fixed_from_double_int): Likewise.
12370 (fixed_from_string): Likewise.
12371 (fixed_convert): Likewise.
12372 (fixed_convert_from_int): Likewise.
12373 (fixed_convert_from_real): Likewise.
12374 (real_convert_from_fixed): Likewise.
12375 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12378 Alan Hayward <alan.hayward@arm.com>
12379 David Sherwood <david.sherwood@arm.com>
12381 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12382 of separate mode class checks. Do not allow vector modes here.
12383 (immed_wide_int_const): Use as_a <scalar_mode>.
12384 * explow.c (trunc_int_for_mode): Likewise.
12385 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12386 (wi::shwi): Likewise.
12387 (wi::min_value): Likewise.
12388 (wi::max_value): Likewise.
12389 * dwarf2out.c (loc_descriptor): Likewise.
12390 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12391 for CONST_WIDE_INT.
12393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12394 Alan Hayward <alan.hayward@arm.com>
12395 David Sherwood <david.sherwood@arm.com>
12397 * tree.h (SCALAR_TYPE_MODE): New macro.
12398 * expr.c (expand_expr_addr_expr_1): Use it.
12399 (expand_expr_real_2): Likewise.
12400 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12401 (fold_convert_const_fixed_from_int): Likewise.
12402 (fold_convert_const_fixed_from_real): Likewise.
12403 (native_encode_fixed): Likewise
12404 (native_encode_complex): Likewise
12405 (native_encode_vector): Likewise.
12406 (native_interpret_fixed): Likewise.
12407 (native_interpret_real): Likewise.
12408 (native_interpret_complex): Likewise.
12409 (native_interpret_vector): Likewise.
12410 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12411 (simd_clone_adjust_argument_types): Likewise.
12412 (simd_clone_init_simd_arrays): Likewise.
12413 (simd_clone_adjust): Likewise.
12414 * stor-layout.c (layout_type): Likewise.
12415 * tree.c (build_minus_one_cst): Likewise.
12416 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12417 * tree-inline.c (estimate_move_cost): Likewise.
12418 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12419 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12420 (vectorizable_reduction): Likewise.
12421 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12422 (vect_recog_mixed_size_cond_pattern): Likewise.
12423 (check_bool_pattern): Likewise.
12424 (adjust_bool_pattern): Likewise.
12425 (search_type_for_mask_1): Likewise.
12426 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12427 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12428 (vectorizable_load): Likewise.
12429 (vectorizable_store): Likewise.
12430 * ubsan.c (ubsan_encode_value): Likewise.
12431 * varasm.c (output_constant): Likewise.
12433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12434 Alan Hayward <alan.hayward@arm.com>
12435 David Sherwood <david.sherwood@arm.com>
12437 * coretypes.h (scalar_mode): New class.
12438 * machmode.h (scalar_mode): Likewise.
12439 (scalar_mode::includes_p): New function.
12440 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12441 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12442 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12443 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12444 * expmed.c (store_bit_field_1): Likewise.
12445 (extract_bit_field_1): Likewise.
12446 * expr.c (write_complex_part): Likewise.
12447 (read_complex_part): Likewise.
12448 (emit_move_complex_push): Likewise.
12449 (expand_expr_real_2): Likewise.
12450 * function.c (assign_parm_setup_reg): Likewise.
12451 (assign_parms_unsplit_complex): Likewise.
12452 * optabs.c (expand_binop): Likewise.
12453 * rtlanal.c (subreg_get_info): Likewise.
12454 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12455 * varasm.c (output_constant_pool_2): Likewise.
12457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12458 Alan Hayward <alan.hayward@arm.com>
12459 David Sherwood <david.sherwood@arm.com>
12461 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12463 (expmed_mult_highpart_optab): Likewise.
12464 (expmed_mult_highpart): Likewise.
12466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12467 Alan Hayward <alan.hayward@arm.com>
12468 David Sherwood <david.sherwood@arm.com>
12470 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12471 instead of a machine_mode.
12472 (builtin_memset_read_str): Likewise.
12473 * builtins.c (c_readstr): Likewise.
12474 (builtin_memcpy_read_str): Likewise.
12475 (builtin_strncpy_read_str): Likewise.
12476 (builtin_memset_read_str): Likewise.
12477 (builtin_memset_gen_str): Likewise.
12478 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12479 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12480 instead of a machine_mode.
12481 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12483 (make_extraction): Likewise.
12484 (try_widen_shift_mode): Take and return scalar_int_modes instead
12486 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12487 a scalar_int_mode instead of a machine_mode.
12488 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12489 (avr_addr_space_pointer_mode): Likewise.
12490 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12491 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12492 (msp430_unwind_word_mode): Likewise.
12493 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12494 (spu_addr_space_pointer_mode): Likewise.
12495 (spu_addr_space_address_mode): Likewise.
12496 (spu_libgcc_cmp_return_mode): Likewise.
12497 (spu_libgcc_shift_count_mode): Likewise.
12498 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12499 (rl78_addr_space_pointer_mode): Likewise.
12500 (fl78_unwind_word_mode): Likewise.
12501 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12503 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12504 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12505 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12506 (mips_valid_pointer_mode): Likewise.
12507 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12508 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12509 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12511 (ft32_addr_space_address_mode): Likewise.
12512 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12513 scalar_int_mode instead of a machine_mode.
12514 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12516 (m32c_addr_space_address_mode): Likewise.
12517 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12518 (rs6000_eh_return_filter_mode): Likewise.
12519 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12520 (rs6000_eh_return_filter_mode): Likewise.
12521 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12522 (s390_libgcc_shift_count_mode): Likewise.
12523 (s390_unwind_word_mode): Likewise.
12524 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12526 * target.def (mode_rep_extended): Likewise.
12527 (valid_pointer_mode): Likewise.
12528 (addr_space.valid_pointer_mode): Likewise.
12529 (eh_return_filter_mode): Return a scalar_int_mode rather than
12531 (libgcc_cmp_return_mode): Likewise.
12532 (libgcc_shift_count_mode): Likewise.
12533 (unwind_word_mode): Likewise.
12534 (addr_space.pointer_mode): Likewise.
12535 (addr_space.address_mode): Likewise.
12536 * doc/tm.texi: Regenerate.
12537 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12539 (do_jump): Use scalar_int_mode for local variables.
12540 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12541 rather than a machine_mode.
12542 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12543 (scompare_loc_descriptor_wide): Likewise.
12544 (scompare_loc_descriptor_narrow): Likewise.
12545 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12547 * except.c (sjlj_emit_dispatch_table): Likewise.
12548 (expand_builtin_eh_copy_values): Likewise.
12549 * explow.c (convert_memory_address_addr_space_1): Likewise.
12550 Take a scalar_int_mode rather than a machine_mode.
12551 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12552 than a machine_mode.
12553 (memory_address_addr_space): Use scalar_int_mode for local variables.
12554 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12555 rather than a machine_mode.
12556 * expmed.c (mask_rtx): Likewise.
12557 (init_expmed_one_conv): Likewise.
12558 (expand_mult_highpart_adjust): Likewise.
12559 (extract_high_half): Likewise.
12560 (expmed_mult_highpart_optab): Likewise.
12561 (expmed_mult_highpart): Likewise.
12562 (expand_smod_pow2): Likewise.
12563 (expand_sdiv_pow2): Likewise.
12564 (emit_store_flag_int): Likewise.
12565 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12567 (extract_low_bits): Likewise.
12568 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12570 * expr.c (pieces_addr::adjust): Likewise.
12571 (can_store_by_pieces): Likewise.
12572 (store_by_pieces): Likewise.
12573 (clear_by_pieces_1): Likewise.
12574 (expand_expr_addr_expr_1): Likewise.
12575 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12576 (expand_expr_real_1): Likewise.
12577 (try_casesi): Likewise.
12578 * final.c (shorten_branches): Likewise.
12579 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12580 type of "mode" to machine_mode.
12581 * internal-fn.c (expand_arith_overflow_result_store): Take a
12582 scalar_int_mode rather than a machine_mode.
12583 (expand_mul_overflow): Use scalar_int_mode for local variables.
12584 * loop-doloop.c (doloop_modify): Likewise.
12585 (doloop_optimize): Likewise.
12586 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12587 than a machine_mode.
12588 (expand_doubleword_shift_condmove): Likewise.
12589 (expand_doubleword_shift): Likewise.
12590 (expand_doubleword_clz): Likewise.
12591 (expand_doubleword_popcount): Likewise.
12592 (expand_doubleword_parity): Likewise.
12593 (expand_absneg_bit): Use scalar_int_mode for local variables.
12594 (prepare_float_lib_cmp): Likewise.
12595 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12596 rather than a machine_mode.
12597 (convert_memory_address_addr_space): Likewise.
12598 (get_mode_bounds): Likewise.
12599 (get_address_mode): Return a scalar_int_mode rather than a
12601 * rtlanal.c (get_address_mode): Likewise.
12602 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12603 than a machine_mode.
12604 * targhooks.c (default_mode_rep_extended): Likewise.
12605 (default_valid_pointer_mode): Likewise.
12606 (default_addr_space_valid_pointer_mode): Likewise.
12607 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12608 than a machine_mode.
12609 (default_libgcc_cmp_return_mode): Likewise.
12610 (default_libgcc_shift_count_mode): Likewise.
12611 (default_unwind_word_mode): Likewise.
12612 (default_addr_space_pointer_mode): Likewise.
12613 (default_addr_space_address_mode): Likewise.
12614 * targhooks.h (default_eh_return_filter_mode): Likewise.
12615 (default_libgcc_cmp_return_mode): Likewise.
12616 (default_libgcc_shift_count_mode): Likewise.
12617 (default_unwind_word_mode): Likewise.
12618 (default_addr_space_pointer_mode): Likewise.
12619 (default_addr_space_address_mode): Likewise.
12620 (default_mode_rep_extended): Take a scalar_int_mode rather than
12622 (default_valid_pointer_mode): Likewise.
12623 (default_addr_space_valid_pointer_mode): Likewise.
12624 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12626 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12627 rather than a machine_mode.
12628 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12629 for local variables.
12630 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12631 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12632 than a machine_mode.
12634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12635 Alan Hayward <alan.hayward@arm.com>
12636 David Sherwood <david.sherwood@arm.com>
12638 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12639 the mode argument to scalar_int_mode.
12640 (do_jump_by_parts_zero_rtx): Likewise.
12641 (do_jump_by_parts_equality_rtx): Likewise.
12642 (do_jump_by_parts_greater): Take a mode argument.
12643 (do_jump_by_parts_equality): Likewise.
12644 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
12651 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12652 (jump_table_for_label): Likewise.
12653 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12654 instead of an rtx_insn *.
12655 (shorten_branches): Use dyn_cast instead of LABEL_P and
12656 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12657 rtx_jump_table_data::get_data_mode.
12658 (final_scan_insn): Likewise.
12660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12661 Alan Hayward <alan.hayward@arm.com>
12662 David Sherwood <david.sherwood@arm.com>
12664 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12665 trying to combine a full-register integer set with a subreg
12668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12669 Alan Hayward <alan.hayward@arm.com>
12670 David Sherwood <david.sherwood@arm.com>
12672 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12673 that is always either address_mode or pointer_mode.
12675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12676 Alan Hayward <alan.hayward@arm.com>
12677 David Sherwood <david.sherwood@arm.com>
12679 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12680 when the two are known to be equal.
12682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12683 Alan Hayward <alan.hayward@arm.com>
12684 David Sherwood <david.sherwood@arm.com>
12686 * simplify-rtx.c (simplify_const_unary_operation): Use
12687 is_a <scalar_int_mode> instead of checking for a nonzero
12688 precision. Forcibly convert op_mode to a scalar_int_mode
12689 in that case. More clearly differentiate the operand and
12690 result modes and use the former when deciding what the value
12691 of a count-bits operation should be. Use is_int_mode instead
12692 of checking for a MODE_INT. Remove redundant check for whether
12693 this mode has a zero precision.
12695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12696 Alan Hayward <alan.hayward@arm.com>
12697 David Sherwood <david.sherwood@arm.com>
12699 * optabs.c (widen_leading): Change the type of the mode argument
12700 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12701 (widen_bswap): Likewise.
12702 (expand_parity): Likewise.
12703 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12704 (expand_ffs): Likewise.
12705 (epand_unop): Check for scalar integer modes before calling the
12708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12709 Alan Hayward <alan.hayward@arm.com>
12710 David Sherwood <david.sherwood@arm.com>
12712 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12714 (expand_expr_real_1): Update call accordingly.
12716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12717 Alan Hayward <alan.hayward@arm.com>
12718 David Sherwood <david.sherwood@arm.com>
12720 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12721 value_mode arguments. Use scalar_int_mode internally.
12722 (store_bit_field_1): Rename the new integer mode from imode
12723 to op0_mode and use it instead of GET_MODE (op0). Update calls
12724 to store_split_bit_field, store_bit_field_using_insv and
12725 store_fixed_bit_field.
12726 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12727 Use scalar_int_mode internally. Use a bit count rather than a mode
12728 when calculating the largest bit size for get_best_mode.
12729 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12730 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12731 Remove assertion that OP0 has a scalar integer mode.
12732 (store_split_bit_field): Add op0_mode and value_mode arguments.
12733 Update calls to extract_fixed_bit_field.
12734 (extract_bit_field_using_extv): Add an op0_mode argument.
12735 Use scalar_int_mode internally.
12736 (extract_bit_field_1): Rename the new integer mode from imode to
12737 op0_mode and use it instead of GET_MODE (op0). Update calls to
12738 extract_split_bit_field, extract_bit_field_using_extv and
12739 extract_fixed_bit_field.
12740 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12741 to extract_split_bit_field and extract_fixed_bit_field_1.
12742 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12743 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12744 on the target mode.
12745 (extract_split_bit_field): Add an op0_mode argument. Update call
12746 to extract_fixed_bit_field.
12748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12749 Alan Hayward <alan.hayward@arm.com>
12750 David Sherwood <david.sherwood@arm.com>
12752 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12753 * explow.c (hard_function_value): Likewise.
12754 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12755 convert_to_mode call outside the loop.
12756 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12757 for the mode iterator. Require the mode specified by max_pieces
12759 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12761 (copy_blkmode_to_reg): Likewise.
12762 (set_storage_via_setmem): Likewise.
12763 * optabs.c (prepare_cmp_insn): Likewise.
12764 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12765 * stor-layout.c (finish_bitfield_representative): Likewise.
12767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12772 * expr.c (convert_move): Use them.
12773 (convert_modes): Likewise.
12774 (store_expr_with_bounds): Likewise.
12776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12777 Alan Hayward <alan.hayward@arm.com>
12778 David Sherwood <david.sherwood@arm.com>
12780 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12781 parameter for the mode of "x". Remove the "known_x", "known_mode"
12782 and "known_ret" arguments. Change the type of the mode argument
12783 to scalar_int_mode.
12784 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12785 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12786 (reg_num_sign_bit_copies_for_combine): Likewise.
12787 * rtlanal.c (nonzero_bits1): Likewise.
12788 (num_sign_bit_copies1): Likewise.
12789 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12790 (reg_num_sign_bit_copies_general): Likewise.
12791 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12792 (reg_nonzero_bits_general): Likewise.
12794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12795 Alan Hayward <alan.hayward@arm.com>
12796 David Sherwood <david.sherwood@arm.com>
12798 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12799 than in subroutines. Return 1 for non-integer modes.
12800 (cached_num_sign_bit_copies): Change the type of the mode parameter
12801 to scalar_int_mode.
12802 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12803 classes. Handle CONST_INT_P first and then check whether X also
12804 has a scalar integer mode. Check the same thing for inner registers
12805 of a SUBREG and for values that are being extended or truncated.
12807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12808 Alan Hayward <alan.hayward@arm.com>
12809 David Sherwood <david.sherwood@arm.com>
12811 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12812 in subroutines. Return the mode mask for non-integer modes.
12813 (cached_nonzero_bits): Change the type of the mode parameter
12814 to scalar_int_mode.
12815 (nonzero_bits1): Likewise. Remove early exit for other mode
12816 classes. Handle CONST_INT_P first and then check whether X
12817 also has a scalar integer mode.
12819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12820 Alan Hayward <alan.hayward@arm.com>
12821 David Sherwood <david.sherwood@arm.com>
12823 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12824 Return a scalar_int_mode and assert that the size is greater than
12826 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12827 (op_by_pieces_d::op_by_pieces_d): Likewise.
12828 (op_by_pieces_d::run): Likewise.
12829 (can_store_by_pieces): Likewise.
12831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12832 Alan Hayward <alan.hayward@arm.com>
12833 David Sherwood <david.sherwood@arm.com>
12835 * combine.c (extract_left_shift): Add a mode argument and update
12837 (make_compound_operation_int): Change the type of the mode parameter
12838 to scalar_int_mode and update the call to extract_left_shift.
12840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12841 Alan Hayward <alan.hayward@arm.com>
12842 David Sherwood <david.sherwood@arm.com>
12844 * combine.c (simplify_and_const_int): Change the type of the mode
12845 parameter to scalar_int_mode.
12846 (simplify_and_const_int_1): Likewise. Update recursive call.
12848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12849 Alan Hayward <alan.hayward@arm.com>
12850 David Sherwood <david.sherwood@arm.com>
12852 * combine.c (simplify_compare_const): Check that the mode is a
12853 scalar_int_mode (rather than VOIDmode) before testing its
12855 (simplify_comparison): Move COMPARISON_P handling out of the
12856 loop and restrict the latter part of the loop to scalar_int_modes.
12857 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12858 and when considering SUBREG_REGs. Use is_int_mode instead of
12859 checking GET_MODE_CLASS against MODE_INT.
12861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12862 Alan Hayward <alan.hayward@arm.com>
12863 David Sherwood <david.sherwood@arm.com>
12865 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12866 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12867 shift_unit_mode and for modes involved in scalar shifts.
12869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12870 Alan Hayward <alan.hayward@arm.com>
12871 David Sherwood <david.sherwood@arm.com>
12873 * combine.c (force_int_to_mode): New function, split out from...
12874 (force_to_mode): ...here. Keep xmode up-to-date and use it
12875 instead of GET_MODE (x).
12877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12878 Alan Hayward <alan.hayward@arm.com>
12879 David Sherwood <david.sherwood@arm.com>
12881 * optabs-query.h (extraction_insn::struct_mode): Change type to
12882 opt_scalar_int_mode and update comment.
12883 (extraction_insn::field_mode): Change type to scalar_int_mode.
12884 (extraction_insn::pos_mode): Likewise.
12885 * combine.c (make_extraction): Update accordingly.
12886 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12887 (get_optab_extraction_insn): Likewise.
12888 * recog.c (simplify_while_replacing): Likewise.
12889 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12890 parameter to opt_scalar_int_mode.
12892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12893 Alan Hayward <alan.hayward@arm.com>
12894 David Sherwood <david.sherwood@arm.com>
12896 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12897 to a scalar_int_mode instead of a machine_mode.
12898 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12899 (get_best_mode): Return a boolean and use a pointer argument to store
12900 the selected mode. Replace the limit mode parameter with a bit limit.
12901 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12902 for the values returned by bit_field_mode_iterator::next_mode.
12903 (store_bit_field): Update call to get_best_mode.
12904 (store_fixed_bit_field): Likewise.
12905 (extract_fixed_bit_field): Likewise.
12906 * expr.c (optimize_bitfield_assignment_op): Likewise.
12907 * fold-const.c (optimize_bit_field_compare): Likewise.
12908 (fold_truth_andor_1): Likewise.
12909 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12910 Update for new type of m_mode.
12911 (get_best_mode): As above.
12913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12914 Alan Hayward <alan.hayward@arm.com>
12915 David Sherwood <david.sherwood@arm.com>
12917 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12918 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12919 (store_bit_field): Check is_a <scalar_int_mode> before calling
12920 strict_volatile_bitfield_p.
12921 (extract_bit_field): Likewise.
12923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12924 Alan Hayward <alan.hayward@arm.com>
12925 David Sherwood <david.sherwood@arm.com>
12927 * target.def (cstore_mode): Return a scalar_int_mode.
12928 * doc/tm.texi: Regenerate.
12929 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12930 * targhooks.h (default_cstore_mode): Likewise.
12931 * targhooks.c (default_cstore_mode): Likewise, using a forced
12933 * expmed.c (emit_cstore): Expect the target of the cstore to be
12936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12937 Alan Hayward <alan.hayward@arm.com>
12938 David Sherwood <david.sherwood@arm.com>
12940 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12942 (niter_desc): Likewise mode.
12943 (iv_analyze): Add a mode parameter.
12945 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12946 and change its type to scalar_int_mode.
12947 * loop-iv.c: Update commentary at head of file.
12948 (iv_constant): Pass the mode paraeter before the rtx it describes
12949 and change its type to scalar_int_mode. Remove VOIDmode handling.
12950 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12951 (iv_extend): Likewise.
12952 (shorten_into_mode): Likewise.
12953 (iv_add): Use scalar_int_mode.
12954 (iv_mult): Likewise.
12955 (iv_shift): Likewise.
12956 (canonicalize_iv_subregs): Likewise.
12957 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12958 it describes and change its mode to scalar_int_mode. Also change
12959 the type of the returned inner_mode to scalar_int_mode.
12960 (get_biv_step): Likewise, turning outer_mode from a pointer
12961 into a direct parameter. Update call to get_biv_step_1.
12962 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12963 iv_constant and get_biv_step.
12964 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12965 and change its type to scalar_int_mode. Don't initialise iv->mode
12966 to VOIDmode and remove later checks for its still being VOIDmode.
12967 Update calls to iv_analyze_op and iv_analyze_expr. Check
12968 is_a <scalar_int_mode> when changing the mode under consideration.
12969 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12970 Update call to iv_analyze_expr.
12971 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12972 inner register is not also a scalar_int_mode. Update call to
12974 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12975 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12976 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12977 separate mode class checks. Update calls to iv_analyze. Remove
12978 fix-up of VOIDmodes after iv_analyze_biv.
12979 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12980 don't have a scalar_int_mode. Update call to biv_p.
12982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12983 Alan Hayward <alan.hayward@arm.com>
12984 David Sherwood <david.sherwood@arm.com>
12986 * cfgexpand.c (convert_debug_memory_address): Use
12987 as_a <scalar_int_mode>.
12988 * combine.c (expand_compound_operation): Likewise.
12989 (make_extraction): Likewise.
12990 (change_zero_ext): Likewise.
12991 (simplify_comparison): Likewise.
12992 * cse.c (cse_insn): Likewise.
12993 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12994 (mem_loc_descriptor): Likewise.
12995 (loc_descriptor): Likewise.
12996 * expmed.c (init_expmed_one_mode): Likewise.
12997 (synth_mult): Likewise.
12998 (emit_store_flag_1): Likewise.
12999 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13000 of a comparison with size.
13001 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13002 (reduce_to_bit_field_precision): Likewise.
13003 * function.c (expand_function_end): Likewise.
13004 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13005 * loop-doloop.c (doloop_modify): Likewise.
13006 * optabs.c (expand_binop): Likewise.
13007 (expand_unop): Likewise.
13008 (expand_copysign_absneg): Likewise.
13009 (prepare_cmp_insn): Likewise.
13010 (maybe_legitimize_operand): Likewise.
13011 * recog.c (const_scalar_int_operand): Likewise.
13012 * rtlanal.c (get_address_mode): Likewise.
13013 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13014 (simplify_cond_clz_ctz): Likewise.
13015 * tree-nested.c (get_nl_goto_field): Likewise.
13016 * tree.c (build_vector_type_for_mode): Likewise.
13017 * var-tracking.c (use_narrower_mode): Likewise.
13019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13020 Alan Hayward <alan.hayward@arm.com>
13021 David Sherwood <david.sherwood@arm.com>
13023 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13024 * builtins.c (expand_builtin_signbit): Use it.
13025 * cfgexpand.c (expand_debug_expr): Likewise.
13026 * dojump.c (do_jump): Likewise.
13027 (do_compare_and_jump): Likewise.
13028 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13029 * expmed.c (make_tree): Likewise.
13030 * expr.c (expand_expr_real_2): Likewise.
13031 (expand_expr_real_1): Likewise.
13032 (try_casesi): Likewise.
13033 * fold-const-call.c (fold_const_call_ss): Likewise.
13034 * fold-const.c (unextend): Likewise.
13035 (extract_muldiv_1): Likewise.
13036 (fold_single_bit_test): Likewise.
13037 (native_encode_int): Likewise.
13038 (native_encode_string): Likewise.
13039 (native_interpret_int): Likewise.
13040 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13041 * internal-fn.c (expand_addsub_overflow): Likewise.
13042 (expand_neg_overflow): Likewise.
13043 (expand_mul_overflow): Likewise.
13044 (expand_arith_overflow): Likewise.
13045 * match.pd: Likewise.
13046 * stor-layout.c (layout_type): Likewise.
13047 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13048 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13049 * tree-ssanames.c (get_range_info): Likewise.
13050 * tree-switch-conversion.c (array_value_type) Likewise.
13051 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13052 (vect_recog_divmod_pattern): Likewise.
13053 (vect_recog_mixed_size_cond_pattern): Likewise.
13054 * tree-vrp.c (extract_range_basic): Likewise.
13055 (simplify_float_conversion_using_ranges): Likewise.
13056 * tree.c (int_fits_type_p): Likewise.
13057 * ubsan.c (instrument_bool_enum_load): Likewise.
13058 * varasm.c (mergeable_string_section): Likewise.
13059 (narrowing_initializer_constant_valid_p): Likewise.
13060 (output_constant): Likewise.
13062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13063 Alan Hayward <alan.hayward@arm.com>
13064 David Sherwood <david.sherwood@arm.com>
13066 * machmode.h (NARROWEST_INT_MODE): New macro.
13067 * expr.c (alignment_for_piecewise_move): Use it instead of
13068 GET_CLASS_NARROWEST_MODE (MODE_INT).
13069 (push_block): Likewise.
13070 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13072 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13075 Alan Hayward <alan.hayward@arm.com>
13076 David Sherwood <david.sherwood@arm.com>
13078 * postreload.c (move2add_valid_value_p): Change the type of the
13079 mode parameter to scalar_int_mode.
13080 (move2add_use_add2_insn): Add a mode parameter and use it instead
13082 (move2add_use_add3_insn): Likewise.
13083 (reload_cse_move2add): Update accordingly.
13085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13086 Alan Hayward <alan.hayward@arm.com>
13087 David Sherwood <david.sherwood@arm.com>
13089 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13091 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13092 * optabs.c (expand_unop): Likewise.
13094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13095 Alan Hayward <alan.hayward@arm.com>
13096 David Sherwood <david.sherwood@arm.com>
13098 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13099 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13100 (popcount_loc_descriptor): Likewise.
13101 (bswap_loc_descriptor): Likewise.
13102 (rotate_loc_descriptor): Likewise.
13103 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13104 calling the functions above.
13106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13107 Alan Hayward <alan.hayward@arm.com>
13108 David Sherwood <david.sherwood@arm.com>
13110 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13112 (try_combine): Likewise.
13113 (simplify_if_then_else): Likewise.
13114 * cse.c (cse_insn): Likewise.
13115 * dwarf2out.c (mem_loc_descriptor): Likewise.
13116 * emit-rtl.c (gen_lowpart_common): Likewise.
13117 * simplify-rtx.c (simplify_truncation): Likewise.
13118 (simplify_binary_operation_1): Likewise.
13119 (simplify_const_relational_operation): Likewise.
13120 (simplify_ternary_operation): Likewise.
13121 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13124 Alan Hayward <alan.hayward@arm.com>
13125 David Sherwood <david.sherwood@arm.com>
13127 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13128 * reload.c (push_reload): Likewise.
13129 (find_reloads): Likewise.
13131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13132 Alan Hayward <alan.hayward@arm.com>
13133 David Sherwood <david.sherwood@arm.com>
13135 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13136 (make_compound_operation_int): Likewise.
13137 (change_zero_ext): Likewise.
13138 * expr.c (convert_move): Likewise.
13139 (convert_modes): Likewise.
13140 * fwprop.c (forward_propagate_subreg): Likewise.
13141 * loop-iv.c (get_biv_step_1): Likewise.
13142 * optabs.c (widen_operand): Likewise.
13143 * postreload.c (move2add_valid_value_p): Likewise.
13144 * recog.c (simplify_while_replacing): Likewise.
13145 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13146 (simplify_binary_operation_1): Likewise. Remove redundant
13147 mode equality check.
13149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13150 Alan Hayward <alan.hayward@arm.com>
13151 David Sherwood <david.sherwood@arm.com>
13153 * combine.c (combine_simplify_rtx): Add checks for
13154 is_a <scalar_int_mode>.
13155 (simplify_if_then_else): Likewise.
13156 (make_field_assignment): Likewise.
13157 (simplify_comparison): Likewise.
13158 * ifcvt.c (noce_try_bitop): Likewise.
13159 * loop-invariant.c (canonicalize_address_mult): Likewise.
13160 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13163 Alan Hayward <alan.hayward@arm.com>
13164 David Sherwood <david.sherwood@arm.com>
13166 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13167 is_a <scalar_int_mode> instead of != BLKmode.
13169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13170 Alan Hayward <alan.hayward@arm.com>
13171 David Sherwood <david.sherwood@arm.com>
13173 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13174 instead of != VOIDmode.
13175 * combine.c (if_then_else_cond): Likewise.
13176 (change_zero_ext): Likewise.
13177 * dwarf2out.c (mem_loc_descriptor): Likewise.
13178 (loc_descriptor): Likewise.
13179 * rtlanal.c (canonicalize_condition): Likewise.
13180 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13183 Alan Hayward <alan.hayward@arm.com>
13184 David Sherwood <david.sherwood@arm.com>
13186 * simplify-rtx.c (simplify_binary_operation_1): Use
13187 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13190 Alan Hayward <alan.hayward@arm.com>
13191 David Sherwood <david.sherwood@arm.com>
13193 * wide-int.h (int_traits<unsigned char>) New class.
13194 (int_traits<unsigned short>) Likewise.
13195 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13196 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13198 * combine.c (set_nonzero_bits_and_sign_copies): Use
13199 is_a <scalar_int_mode>.
13200 (find_split_point): Likewise.
13201 (combine_simplify_rtx): Likewise.
13202 (simplify_logical): Likewise.
13203 (expand_compound_operation): Likewise.
13204 (expand_field_assignment): Likewise.
13205 (make_compound_operation): Likewise.
13206 (extended_count): Likewise.
13207 (change_zero_ext): Likewise.
13208 (simplify_comparison): Likewise.
13209 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13210 (ucompare_loc_descriptor): Likewise.
13211 (minmax_loc_descriptor): Likewise.
13212 (mem_loc_descriptor): Likewise.
13213 (loc_descriptor): Likewise.
13214 * expmed.c (init_expmed_one_mode): Likewise.
13215 * lra-constraints.c (lra_constraint_offset): Likewise.
13216 * optabs.c (prepare_libcall_arg): Likewise.
13217 * postreload.c (move2add_note_store): Likewise.
13218 * reload.c (operands_match_p): Likewise.
13219 * rtl.h (load_extend_op): Likewise.
13220 * rtlhooks.c (gen_lowpart_general): Likewise.
13221 * simplify-rtx.c (simplify_truncation): Likewise.
13222 (simplify_unary_operation_1): Likewise.
13223 (simplify_binary_operation_1): Likewise.
13224 (simplify_const_binary_operation): Likewise.
13225 (simplify_const_relational_operation): Likewise.
13226 (simplify_subreg): Likewise.
13227 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13228 * var-tracking.c (adjust_mems): Likewise.
13229 (prepare_call_arguments): Likewise.
13231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13232 Alan Hayward <alan.hayward@arm.com>
13233 David Sherwood <david.sherwood@arm.com>
13235 * machmode.h (is_int_mode): New fuction.
13236 * combine.c (find_split_point): Use it.
13237 (combine_simplify_rtx): Likewise.
13238 (simplify_if_then_else): Likewise.
13239 (simplify_set): Likewise.
13240 (simplify_shift_const_1): Likewise.
13241 (simplify_comparison): Likewise.
13242 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13243 * cse.c (notreg_cost): Likewise.
13244 (cse_insn): Likewise.
13245 * cselib.c (cselib_lookup_1): Likewise.
13246 * dojump.c (do_jump_1): Likewise.
13247 (do_compare_rtx_and_jump): Likewise.
13248 * dse.c (get_call_args): Likewise.
13249 * dwarf2out.c (rtl_for_decl_init): Likewise.
13250 (native_encode_initializer): Likewise.
13251 * expmed.c (emit_store_flag_1): Likewise.
13252 (emit_store_flag): Likewise.
13253 * expr.c (convert_modes): Likewise.
13254 (store_field): Likewise.
13255 (expand_expr_real_1): Likewise.
13256 * fold-const.c (fold_read_from_constant_string): Likewise.
13257 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13258 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13259 * optabs.c (expand_binop): Likewise.
13260 (expand_unop): Likewise.
13261 (expand_abs_nojump): Likewise.
13262 (expand_one_cmpl_abs_nojump): Likewise.
13263 * simplify-rtx.c (mode_signbit_p): Likewise.
13264 (val_signbit_p): Likewise.
13265 (val_signbit_known_set_p): Likewise.
13266 (val_signbit_known_clear_p): Likewise.
13267 (simplify_relational_operation_1): Likewise.
13268 * tree.c (vector_type_mode): Likewise.
13270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13271 Alan Hayward <alan.hayward@arm.com>
13272 David Sherwood <david.sherwood@arm.com>
13274 * machmode.h (smallest_mode_for_size): Fix formatting.
13275 (smallest_int_mode_for_size): New function.
13276 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
13277 instead of smallest_mode_for_size.
13278 * combine.c (make_extraction): Likewise.
13279 * config/arc/arc.c (arc_expand_movmem): Likewise.
13280 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
13281 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13282 * config/s390/s390.c (s390_expand_insv): Likewise.
13283 * config/sparc/sparc.c (assign_int_registers): Likewise.
13284 * config/spu/spu.c (spu_function_value): Likewise.
13285 (spu_function_arg): Likewise.
13286 * coverage.c (get_gcov_type): Likewise.
13287 (get_gcov_unsigned_t): Likewise.
13288 * dse.c (find_shift_sequence): Likewise.
13289 * expmed.c (store_bit_field_1): Likewise.
13290 * expr.c (convert_move): Likewise.
13291 (store_field): Likewise.
13292 * internal-fn.c (expand_arith_overflow): Likewise.
13293 * optabs-query.c (get_best_extraction_insn): Likewise.
13294 * optabs.c (expand_twoval_binop_libfunc): Likewise.
13295 * stor-layout.c (layout_type): Likewise.
13296 (initialize_sizetypes): Likewise.
13297 * targhooks.c (default_get_mask_mode): Likewise.
13298 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
13300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13301 Alan Hayward <alan.hayward@arm.com>
13302 David Sherwood <david.sherwood@arm.com>
13304 * machmode.h (opt_mode::else_blk): New function.
13305 (int_mode_for_mode): Declare.
13306 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13307 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13309 * cfgexpand.c (expand_debug_expr): Likewise.
13310 * combine.c (gen_lowpart_or_truncate): Likewise.
13311 (gen_lowpart_for_combine): Likewise.
13312 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13313 * config/avr/avr.c (avr_to_int_mode): Likewise.
13314 (avr_out_plus_1): Likewise.
13315 (avr_out_plus): Likewise.
13316 (avr_out_round): Likewise.
13317 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13318 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13319 (s390_expand_vcond): Likewise.
13320 * config/spu/spu.c (spu_split_immediate): Likewise.
13321 (spu_expand_mov): Likewise.
13322 * dse.c (get_stored_val): Likewise.
13323 * expmed.c (store_bit_field_1): Likewise.
13324 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13326 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13327 (extract_low_bits): Likewise.
13328 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13329 handling rather than repeating the check.
13330 (emit_group_store): Likewise.
13331 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13332 * optabs.c (expand_absneg_bit): Likewise.
13333 (expand_copysign_absneg): Likewise.
13334 (expand_copysign_bit): Likewise.
13335 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13336 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13337 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13338 * var-tracking.c (prepare_call_arguments): Likewise.
13339 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13340 int_mode_for_mode instead of mode_for_size.
13341 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13344 Alan Hayward <alan.hayward@arm.com>
13345 David Sherwood <david.sherwood@arm.com>
13347 * machmode.h (int_mode_for_size): New function.
13348 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13349 instead of mode_for_size.
13350 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13352 * combine.c (expand_field_assignment): Use int_mode_for_size
13353 instead of mode_for_size.
13354 (make_extraction): Likewise.
13355 (simplify_shift_const_1): Likewise.
13356 (simplify_comparison): Likewise.
13357 * dojump.c (do_jump): Likewise.
13358 * dwarf2out.c (mem_loc_descriptor): Likewise.
13359 * emit-rtl.c (init_derived_machine_modes): Likewise.
13360 * expmed.c (flip_storage_order): Likewise.
13361 (convert_extracted_bit_field): Likewise.
13362 * expr.c (copy_blkmode_from_reg): Likewise.
13363 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13364 * internal-fn.c (expand_mul_overflow): Likewise.
13365 * lower-subreg.c (simple_move): Likewise.
13366 * optabs-libfuncs.c (init_optabs): Likewise.
13367 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13368 * tree.c (vector_type_mode): Likewise.
13369 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13370 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13371 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13372 * tree-vect-stmts.c (vectorizable_load): Likewise.
13373 (vectorizable_store): Likewise.
13375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13376 Alan Hayward <alan.hayward@arm.com>
13377 David Sherwood <david.sherwood@arm.com>
13379 * coretypes.h (pod_mode): New type.
13380 (scalar_int_mode_pod): New typedef.
13381 * machmode.h (pod_mode): New class.
13382 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13383 * genmodes.c (emit_mode_int_n): Update accordingly.
13384 * lower-subreg.h (target_lower_subreg): Change type to
13385 scalar_int_mode_pod.
13386 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13387 scalar_int_mode_pod.
13389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13390 Alan Hayward <alan.hayward@arm.com>
13391 David Sherwood <david.sherwood@arm.com>
13393 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13394 machine_mode to scalar_int_mode.
13395 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13396 (rs6000_option_override_internal): Remove cast to int.
13397 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13398 machine_mode to scalar_int_mode.
13399 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13400 (rs6000_option_override_internal): Remove cast to int.
13401 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13402 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13404 * config/s390/s390.c (s390_expand_builtin): Likewise.
13405 * coretypes.h (scalar_int_mode): New type.
13406 (opt_scalar_int_mode): New typedef.
13407 * machmode.h (scalar_int_mode): New class.
13408 (scalar_int_mode::includes_p): New function.
13409 (byte_mode): Change type to scalar_int_mode.
13410 (word_mode): Likewise.
13411 (ptr_mode): Likewise.
13412 * emit-rtl.c (byte_mode): Likewise.
13413 (word_mode): Likewise.
13414 (ptr_mode): Likewise.
13415 (init_derived_machine_modes): Update accordingly.
13416 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13417 and MODE_PARTIAL_INT.
13418 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13419 opt_scalar_int_mode.
13421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13422 Alan Hayward <alan.hayward@arm.com>
13423 David Sherwood <david.sherwood@arm.com>
13425 * target.def (libgcc_floating_mode_supported_p): Take a
13427 * doc/tm.texi: Regenerate.
13428 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13430 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13431 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13435 Alan Hayward <alan.hayward@arm.com>
13436 David Sherwood <david.sherwood@arm.com>
13438 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13439 * doc/tm.texi: Regenerate.
13440 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13441 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13442 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13443 * targhooks.h (default_floatn_mode): Likewise.
13444 * targhooks.c (default_floatn_mode): Likewise.
13445 * tree.c (build_common_tree_nodes): Update accordingly.
13447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13448 Alan Hayward <alan.hayward@arm.com>
13449 David Sherwood <david.sherwood@arm.com>
13451 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13452 (mode_iterator::iterate_p): Likewise.
13453 (mode_iterator::get_wider): Likewise.
13454 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13457 Alan Hayward <alan.hayward@arm.com>
13458 David Sherwood <david.sherwood@arm.com>
13460 * coretypes.h (opt_scalar_float_mode): New typedef.
13461 * machmode.h (float_mode_for_size): New function.
13462 * emit-rtl.c (double_mode): Delete.
13463 (init_emit_once): Use float_mode_for_size.
13464 * stor-layout.c (layout_type): Likewise.
13465 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13468 Alan Hayward <alan.hayward@arm.com>
13469 David Sherwood <david.sherwood@arm.com>
13471 * output.h (assemble_real): Take a scalar_float_mode.
13472 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13473 * config/arm/arm.md (consttable_4): Likewise.
13474 (consttable_8): Likewise.
13475 (consttable_16): Likewise.
13476 * config/mips/mips.md (consttable_float): Likewise.
13477 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13478 * varasm.c (assemble_real): Take a scalar_float_mode.
13479 (output_constant_pool_2): Update accordingly.
13480 (output_constant): Likewise.
13482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13483 Alan Hayward <alan.hayward@arm.com>
13484 David Sherwood <david.sherwood@arm.com>
13486 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13487 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13488 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13489 (native_encode_real): Likewise.
13490 (native_interpret_real): Likewise.
13491 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13492 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13495 Alan Hayward <alan.hayward@arm.com>
13496 David Sherwood <david.sherwood@arm.com>
13498 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13499 <scalar_float_mode>. Simplify.
13500 (gen_extend_conv_libfunc): Likewise.
13502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13503 Alan Hayward <alan.hayward@arm.com>
13504 David Sherwood <david.sherwood@arm.com>
13506 * coretypes.h (scalar_float_mode): New type.
13507 * machmode.h (mode_traits::from_int): Use machine_mode if
13508 USE_ENUM_MODES is defined.
13509 (is_a): New function.
13511 (dyn_cast): Likewise.
13512 (scalar_float_mode): New class.
13513 (scalar_float_mode::includes_p): New function.
13514 (is_float_mode): Likewise.
13515 * gdbhooks.py (MachineModePrinter): New class.
13516 (build_pretty_printer): Use it for scalar_float_mode.
13517 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13518 (format_helper::format_helper): Turn into a template.
13519 * genmodes.c (get_mode_class): New function.
13520 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13521 or machine_mode if none.
13522 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13523 as_a <scalar_float_mode>.
13524 * dwarf2out.c (mem_loc_descriptor): Likewise.
13525 (insert_float): Likewise.
13526 (add_const_value_attribute): Likewise.
13527 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13528 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13529 (expand_unop): Update accordingly.
13530 (expand_abs_nojump): Likewise.
13531 (expand_copysign_absneg): Take a scalar_float_mode.
13532 (expand_copysign_bit): Likewise.
13533 (expand_copysign): Update accordingly.
13535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13536 Alan Hayward <alan.hayward@arm.com>
13537 David Sherwood <david.sherwood@arm.com>
13539 * coretypes.h (opt_mode): New class.
13540 * machmode.h (opt_mode): Likewise.
13541 (opt_mode::else_void): New function.
13542 (opt_mode::require): Likewise.
13543 (opt_mode::exists): Likewise.
13544 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13545 (GET_MODE_2XWIDER_MODE): Likewise.
13546 (mode_iterator::get_wider): Update accordingly.
13547 (mode_iterator::get_2xwider): Likewise.
13548 (mode_iterator::get_known_wider): Likewise, turning into a template.
13549 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13550 forcing a wider mode to exist.
13551 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13552 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13553 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13554 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13555 * lower-subreg.c (init_lower_subreg): Likewise.
13556 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13557 on the final iteration.
13558 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13559 a wider mode exists before asking for a move pattern.
13560 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13561 forcing a wider mode to exist.
13562 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13563 returning false if no such mode exists.
13564 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13565 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13566 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13567 Avoid checking for a MODE_INT if we already know the mode is not a
13569 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13570 forcing a wider mode to exist.
13571 (expmed_mult_highpart_optab): Likewise.
13572 (expmed_mult_highpart): Likewise.
13573 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13575 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13576 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13577 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13578 * internal-fn.c (expand_mul_overflow): Update use of
13579 GET_MODE_2XWIDER_MODE.
13580 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13581 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13582 GET_MODE_WIDER_MODE.
13583 (convert_plusminus_to_widen): Likewise.
13584 * tree-switch-conversion.c (array_value_type): Likewise.
13585 * var-tracking.c (emit_note_insn_var_location): Likewise.
13586 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13587 Return false inside rather than outside the loop if no wider mode
13589 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13590 and GET_MODE_2XWIDER_MODE
13591 (can_compare_p): Use else_void.
13592 * gdbhooks.py (OptMachineModePrinter): New class.
13593 (build_pretty_printer): Use it for opt_mode.
13595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13596 Alan Hayward <alan.hayward@arm.com>
13597 David Sherwood <david.sherwood@arm.com>
13599 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13600 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13603 Alan Hayward <alan.hayward@arm.com>
13604 David Sherwood <david.sherwood@arm.com>
13606 * machmode.h (mode_traits): New structure.
13607 (get_narrowest_mode): New function.
13608 (mode_iterator::start): Likewise.
13609 (mode_iterator::iterate_p): Likewise.
13610 (mode_iterator::get_wider): Likewise.
13611 (mode_iterator::get_known_wider): Likewise.
13612 (mode_iterator::get_2xwider): Likewise.
13613 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13614 (FOR_EACH_MODE): Likewise.
13615 (FOR_EACH_MODE_FROM): Likewise.
13616 (FOR_EACH_MODE_UNTIL): Likewise.
13617 (FOR_EACH_WIDER_MODE): Likewise.
13618 (FOR_EACH_2XWIDER_MODE): Likewise.
13619 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13620 * combine.c (simplify_comparison): Likewise
13621 * config/i386/i386.c (type_natural_mode): Likewise.
13622 * cse.c (cse_insn): Likewise.
13623 * dse.c (find_shift_sequence): Likewise.
13624 * emit-rtl.c (init_derived_machine_modes): Likewise.
13625 (init_emit_once): Likewise.
13626 * explow.c (hard_function_value): Likewise.
13627 * expmed.c (extract_fixed_bit_field_1): Likewise.
13628 (extract_bit_field_1): Likewise.
13629 (expand_divmod): Likewise.
13630 (emit_store_flag_1): Likewise.
13631 * expr.c (init_expr_target): Likewise.
13632 (convert_move): Likewise.
13633 (alignment_for_piecewise_move): Likewise.
13634 (widest_int_mode_for_size): Likewise.
13635 (emit_block_move_via_movmem): Likewise.
13636 (copy_blkmode_to_reg): Likewise.
13637 (set_storage_via_setmem): Likewise.
13638 (compress_float_constant): Likewise.
13639 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13640 * optabs-query.c (get_best_extraction_insn): Likewise.
13641 * optabs.c (expand_binop): Likewise.
13642 (expand_twoval_unop): Likewise.
13643 (expand_twoval_binop): Likewise.
13644 (widen_leading): Likewise.
13645 (widen_bswap): Likewise.
13646 (expand_parity): Likewise.
13647 (expand_unop): Likewise.
13648 (prepare_cmp_insn): Likewise.
13649 (prepare_float_lib_cmp): Likewise.
13650 (expand_float): Likewise.
13651 (expand_fix): Likewise.
13652 (expand_sfix_optab): Likewise.
13653 * postreload.c (move2add_use_add2_insn): Likewise.
13654 * reg-stack.c (reg_to_stack): Likewise.
13655 * reginfo.c (choose_hard_reg_mode): Likewise.
13656 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13657 * stor-layout.c (mode_for_size): Likewise.
13658 (smallest_mode_for_size): Likewise.
13659 (mode_for_vector): Likewise.
13660 (finish_bitfield_representative): Likewise.
13661 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13662 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13663 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13664 * var-tracking.c (prepare_call_arguments): Likewise.
13666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13667 Alan Hayward <alan.hayward@arm.com>
13668 David Sherwood <david.sherwood@arm.com>
13670 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13671 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13672 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13673 * machmode.h (mode_size): Move earlier in file.
13674 (mode_precision): Likewise.
13675 (mode_inner): Likewise.
13676 (mode_nunits): Likewise.
13677 (mode_unit_size): Likewise.
13678 (unit_unit_precision): Likewise.
13679 (mode_wider): Likewise.
13680 (mode_2xwider): Likewise.
13681 (machine_mode): New class.
13682 (mode_to_bytes): New function.
13683 (mode_to_bits): Likewise.
13684 (mode_to_precision): Likewise.
13685 (mode_to_inner): Likewise.
13686 (mode_to_unit_size): Likewise.
13687 (mode_to_unit_precision): Likewise.
13688 (mode_to_nunits): Likewise.
13689 (GET_MODE_SIZE): Use mode_to_bytes.
13690 (GET_MODE_BITSIZE): Use mode_to_bits.
13691 (GET_MODE_PRECISION): Use mode_to_precision.
13692 (GET_MODE_INNER): Use mode_to_inner.
13693 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13694 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13695 (GET_MODE_NUNITS): Use mode_to_nunits.
13696 * system.h (ALWAYS_INLINE): New macro.
13697 * config/powerpcspe/powerpcspe-c.c
13698 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13699 int for arg1_mode and arg2_mode.
13701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13702 Alan Hayward <alan.hayward@arm.com>
13703 David Sherwood <david.sherwood@arm.com>
13705 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13706 Prefix mode names with E_ in case statements.
13707 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13708 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13709 (aarch64_split_simd_move): Likewise.
13710 (aarch64_gen_storewb_pair): Likewise.
13711 (aarch64_gen_loadwb_pair): Likewise.
13712 (aarch64_gen_store_pair): Likewise.
13713 (aarch64_gen_load_pair): Likewise.
13714 (aarch64_get_condition_code_1): Likewise.
13715 (aarch64_constant_pool_reload_icode): Likewise.
13716 (get_rsqrte_type): Likewise.
13717 (get_rsqrts_type): Likewise.
13718 (get_recpe_type): Likewise.
13719 (get_recps_type): Likewise.
13720 (aarch64_gimplify_va_arg_expr): Likewise.
13721 (aarch64_simd_container_mode): Likewise.
13722 (aarch64_emit_load_exclusive): Likewise.
13723 (aarch64_emit_store_exclusive): Likewise.
13724 (aarch64_expand_compare_and_swap): Likewise.
13725 (aarch64_gen_atomic_cas): Likewise.
13726 (aarch64_emit_bic): Likewise.
13727 (aarch64_emit_atomic_swap): Likewise.
13728 (aarch64_emit_atomic_load_op): Likewise.
13729 (aarch64_evpc_trn): Likewise.
13730 (aarch64_evpc_uzp): Likewise.
13731 (aarch64_evpc_zip): Likewise.
13732 (aarch64_evpc_ext): Likewise.
13733 (aarch64_evpc_rev): Likewise.
13734 (aarch64_evpc_dup): Likewise.
13735 (aarch64_gen_ccmp_first): Likewise.
13736 (aarch64_gen_ccmp_next): Likewise.
13737 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13738 (alpha_emit_xfloating_libcall): Likewise.
13739 (emit_insxl): Likewise.
13740 (alpha_arg_type): Likewise.
13741 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13742 (arc_preferred_simd_mode): Likewise.
13743 (arc_secondary_reload): Likewise.
13744 (get_arc_condition_code): Likewise.
13745 (arc_print_operand): Likewise.
13746 (arc_legitimate_constant_p): Likewise.
13747 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13748 * config/arc/arc.md (casesi_load): Likewise.
13749 (casesi_compact_jump): Likewise.
13750 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13751 (cc_use_register): Likewise.
13752 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13753 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13754 (arm_init_iwmmxt_builtins): Likewise.
13755 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13756 (neon_expand_vector_init): Likewise.
13757 (arm_attr_length_move_neon): Likewise.
13758 (maybe_get_arm_condition_code): Likewise.
13759 (arm_emit_vector_const): Likewise.
13760 (arm_preferred_simd_mode): Likewise.
13761 (arm_output_iwmmxt_tinsr): Likewise.
13762 (thumb1_output_casesi): Likewise.
13763 (thumb2_output_casesi): Likewise.
13764 (arm_emit_load_exclusive): Likewise.
13765 (arm_emit_store_exclusive): Likewise.
13766 (arm_expand_compare_and_swap): Likewise.
13767 (arm_evpc_neon_vuzp): Likewise.
13768 (arm_evpc_neon_vzip): Likewise.
13769 (arm_evpc_neon_vrev): Likewise.
13770 (arm_evpc_neon_vtrn): Likewise.
13771 (arm_evpc_neon_vext): Likewise.
13772 (arm_validize_comparison): Likewise.
13773 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13774 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13775 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13776 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13777 (c6x_preferred_simd_mode): Likewise.
13778 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13779 (epiphany_rtx_costs): Likewise.
13780 * config/epiphany/predicates.md (proper_comparison_operator):
13782 * config/frv/frv.c (condexec_memory_operand): Likewise.
13783 (frv_emit_move): Likewise.
13784 (output_move_single): Likewise.
13785 (output_condmove_single): Likewise.
13786 (frv_hard_regno_mode_ok): Likewise.
13787 (frv_matching_accg_mode): Likewise.
13788 * config/h8300/h8300.c (split_adds_subs): Likewise.
13789 (h8300_rtx_costs): Likewise.
13790 (h8300_print_operand): Likewise.
13791 (compute_mov_length): Likewise.
13792 (output_logical_op): Likewise.
13793 (compute_logical_op_length): Likewise.
13794 (compute_logical_op_cc): Likewise.
13795 (h8300_shift_needs_scratch_p): Likewise.
13796 (output_a_shift): Likewise.
13797 (compute_a_shift_length): Likewise.
13798 (compute_a_shift_cc): Likewise.
13799 (expand_a_rotate): Likewise.
13800 (output_a_rotate): Likewise.
13801 * config/i386/i386.c (classify_argument): Likewise.
13802 (function_arg_advance_32): Likewise.
13803 (function_arg_32): Likewise.
13804 (function_arg_64): Likewise.
13805 (function_value_64): Likewise.
13806 (ix86_gimplify_va_arg): Likewise.
13807 (ix86_legitimate_constant_p): Likewise.
13808 (put_condition_code): Likewise.
13809 (split_double_mode): Likewise.
13810 (ix86_avx256_split_vector_move_misalign): Likewise.
13811 (ix86_expand_vector_logical_operator): Likewise.
13812 (ix86_split_idivmod): Likewise.
13813 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13814 (ix86_build_const_vector): Likewise.
13815 (ix86_build_signbit_mask): Likewise.
13816 (ix86_match_ccmode): Likewise.
13817 (ix86_cc_modes_compatible): Likewise.
13818 (ix86_expand_branch): Likewise.
13819 (ix86_expand_sse_cmp): Likewise.
13820 (ix86_expand_sse_movcc): Likewise.
13821 (ix86_expand_int_sse_cmp): Likewise.
13822 (ix86_expand_vec_perm_vpermi2): Likewise.
13823 (ix86_expand_vec_perm): Likewise.
13824 (ix86_expand_sse_unpack): Likewise.
13825 (ix86_expand_int_addcc): Likewise.
13826 (ix86_split_to_parts): Likewise.
13827 (ix86_vectorize_builtin_gather): Likewise.
13828 (ix86_vectorize_builtin_scatter): Likewise.
13829 (avx_vpermilp_parallel): Likewise.
13830 (inline_memory_move_cost): Likewise.
13831 (ix86_tieable_integer_mode_p): Likewise.
13832 (x86_maybe_negate_const_int): Likewise.
13833 (ix86_expand_vector_init_duplicate): Likewise.
13834 (ix86_expand_vector_init_one_nonzero): Likewise.
13835 (ix86_expand_vector_init_one_var): Likewise.
13836 (ix86_expand_vector_init_concat): Likewise.
13837 (ix86_expand_vector_init_interleave): Likewise.
13838 (ix86_expand_vector_init_general): Likewise.
13839 (ix86_expand_vector_set): Likewise.
13840 (ix86_expand_vector_extract): Likewise.
13841 (emit_reduc_half): Likewise.
13842 (ix86_emit_i387_round): Likewise.
13843 (ix86_mangle_type): Likewise.
13844 (ix86_expand_round_sse4): Likewise.
13845 (expand_vec_perm_blend): Likewise.
13846 (canonicalize_vector_int_perm): Likewise.
13847 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13848 (expand_vec_perm_1): Likewise.
13849 (expand_vec_perm_interleave3): Likewise.
13850 (expand_vec_perm_even_odd_pack): Likewise.
13851 (expand_vec_perm_even_odd_1): Likewise.
13852 (expand_vec_perm_broadcast_1): Likewise.
13853 (ix86_vectorize_vec_perm_const_ok): Likewise.
13854 (ix86_expand_vecop_qihi): Likewise.
13855 (ix86_expand_mul_widen_hilo): Likewise.
13856 (ix86_expand_sse2_abs): Likewise.
13857 (ix86_expand_pextr): Likewise.
13858 (ix86_expand_pinsr): Likewise.
13859 (ix86_preferred_simd_mode): Likewise.
13860 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13861 * config/i386/sse.md (*andnot<mode>3): Likewise.
13862 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13863 (*<code><mode>3): Likewise.
13864 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13865 (ia64_expand_atomic_op): Likewise.
13866 (ia64_arg_type): Likewise.
13867 (ia64_mode_to_int): Likewise.
13868 (ia64_scalar_mode_supported_p): Likewise.
13869 (ia64_vector_mode_supported_p): Likewise.
13870 (expand_vec_perm_broadcast): Likewise.
13871 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13872 (iq2000_function_arg_advance): Likewise.
13873 (iq2000_function_arg): Likewise.
13874 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13875 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13876 (m68k_libcall_value): Likewise.
13877 (m68k_function_value): Likewise.
13878 (sched_attr_op_type): Likewise.
13879 * config/mcore/mcore.c (mcore_output_move): Likewise.
13880 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13882 (microblaze_function_arg): Likewise.
13883 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13884 (mips_print_operand): Likewise.
13885 (mips_mode_ok_for_mov_fmt_p): Likewise.
13886 (mips_vector_mode_supported_p): Likewise.
13887 (mips_preferred_simd_mode): Likewise.
13888 (mips_expand_vpc_loongson_even_odd): Likewise.
13889 (mips_expand_vec_unpack): Likewise.
13890 (mips_expand_vi_broadcast): Likewise.
13891 (mips_expand_vector_init): Likewise.
13892 (mips_expand_vec_reduc): Likewise.
13893 (mips_expand_msa_cmp): Likewise.
13894 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13895 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13896 (cc_flags_for_mode): Likewise.
13897 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13898 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13899 (nds32_output_casesi_pc_relative): Likewise.
13900 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13901 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13902 (nvptx_gen_unpack): Likewise.
13903 (nvptx_gen_pack): Likewise.
13904 (nvptx_gen_shuffle): Likewise.
13905 (nvptx_gen_wcast): Likewise.
13906 (nvptx_preferred_simd_mode): Likewise.
13907 * config/pa/pa.c (pa_secondary_reload): Likewise.
13908 * config/pa/predicates.md (base14_operand): Likewise.
13909 * config/powerpcspe/powerpcspe-c.c
13910 (altivec_resolve_overloaded_builtin): Likewise.
13911 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13913 (rs6000_preferred_simd_mode): Likewise.
13914 (output_vec_const_move): Likewise.
13915 (rs6000_expand_vector_extract): Likewise.
13916 (rs6000_split_vec_extract_var): Likewise.
13917 (reg_offset_addressing_ok_p): Likewise.
13918 (rs6000_legitimate_offset_address_p): Likewise.
13919 (rs6000_legitimize_address): Likewise.
13920 (rs6000_emit_set_const): Likewise.
13921 (rs6000_const_vec): Likewise.
13922 (rs6000_emit_move): Likewise.
13923 (spe_build_register_parallel): Likewise.
13924 (rs6000_darwin64_record_arg_recurse): Likewise.
13925 (swap_selector_for_mode): Likewise.
13926 (spe_init_builtins): Likewise.
13927 (paired_init_builtins): Likewise.
13928 (altivec_init_builtins): Likewise.
13929 (do_load_for_compare): Likewise.
13930 (rs6000_generate_compare): Likewise.
13931 (rs6000_expand_float128_convert): Likewise.
13932 (emit_load_locked): Likewise.
13933 (emit_store_conditional): Likewise.
13934 (rs6000_output_function_epilogue): Likewise.
13935 (rs6000_handle_altivec_attribute): Likewise.
13936 (rs6000_function_value): Likewise.
13937 (emit_fusion_gpr_load): Likewise.
13938 (emit_fusion_p9_load): Likewise.
13939 (emit_fusion_p9_store): Likewise.
13940 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13941 (fusion_gpr_mem_load): Likewise.
13942 (fusion_addis_mem_combo_load): Likewise.
13943 (fusion_addis_mem_combo_store): Likewise.
13944 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13945 (fusion_gpr_mem_load): Likewise.
13946 (fusion_addis_mem_combo_load): Likewise.
13947 (fusion_addis_mem_combo_store): Likewise.
13948 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13950 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13951 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13952 (rs6000_preferred_simd_mode): Likewise.
13953 (output_vec_const_move): Likewise.
13954 (rs6000_expand_vector_extract): Likewise.
13955 (rs6000_split_vec_extract_var): Likewise.
13956 (reg_offset_addressing_ok_p): Likewise.
13957 (rs6000_legitimate_offset_address_p): Likewise.
13958 (rs6000_legitimize_address): Likewise.
13959 (rs6000_emit_set_const): Likewise.
13960 (rs6000_const_vec): Likewise.
13961 (rs6000_emit_move): Likewise.
13962 (rs6000_darwin64_record_arg_recurse): Likewise.
13963 (swap_selector_for_mode): Likewise.
13964 (paired_init_builtins): Likewise.
13965 (altivec_init_builtins): Likewise.
13966 (rs6000_expand_float128_convert): Likewise.
13967 (emit_load_locked): Likewise.
13968 (emit_store_conditional): Likewise.
13969 (rs6000_output_function_epilogue): Likewise.
13970 (rs6000_handle_altivec_attribute): Likewise.
13971 (rs6000_function_value): Likewise.
13972 (emit_fusion_gpr_load): Likewise.
13973 (emit_fusion_p9_load): Likewise.
13974 (emit_fusion_p9_store): Likewise.
13975 * config/rx/rx.c (rx_gen_move_template): Likewise.
13976 (flags_from_mode): Likewise.
13977 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13978 (s390_slb_comparison): Likewise.
13979 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13980 (s390_vector_mode_supported_p): Likewise.
13981 (s390_cc_modes_compatible): Likewise.
13982 (s390_match_ccmode_set): Likewise.
13983 (s390_canonicalize_comparison): Likewise.
13984 (s390_emit_compare_and_swap): Likewise.
13985 (s390_branch_condition_mask): Likewise.
13986 (s390_rtx_costs): Likewise.
13987 (s390_secondary_reload): Likewise.
13988 (__SECONDARY_RELOAD_CASE): Likewise.
13989 (s390_expand_cs): Likewise.
13990 (s390_preferred_simd_mode): Likewise.
13991 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13992 * config/sh/sh.c (sh_print_operand): Likewise.
13993 (dump_table): Likewise.
13994 (sh_secondary_reload): Likewise.
13995 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13996 * config/sh/sh.md (casesi_worker_1): Likewise.
13997 (casesi_worker_2): Likewise.
13998 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13999 (fcc_comparison_operator): Likewise.
14000 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14001 (emit_soft_tfmode_cvt): Likewise.
14002 (sparc_preferred_simd_mode): Likewise.
14003 (output_cbranch): Likewise.
14004 (sparc_print_operand): Likewise.
14005 (sparc_expand_vec_perm_bmask): Likewise.
14006 (vector_init_bshuffle): Likewise.
14007 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14008 (spu_vector_mode_supported_p): Likewise.
14009 (spu_expand_insv): Likewise.
14010 (spu_emit_branch_or_set): Likewise.
14011 (spu_handle_vector_attribute): Likewise.
14012 (spu_builtin_splats): Likewise.
14013 (spu_builtin_extract): Likewise.
14014 (spu_builtin_promote): Likewise.
14015 (spu_expand_sign_extend): Likewise.
14016 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14017 (tilegx_simd_int): Likewise.
14018 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14019 (tilepro_simd_int): Likewise.
14020 * config/v850/v850.c (const_double_split): Likewise.
14021 (v850_print_operand): Likewise.
14022 (ep_memory_offset): Likewise.
14023 * config/vax/vax.c (vax_rtx_costs): Likewise.
14024 (vax_output_int_move): Likewise.
14025 (vax_output_int_add): Likewise.
14026 (vax_output_int_subtract): Likewise.
14027 * config/visium/predicates.md (visium_branch_operator): Likewise.
14028 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14029 (visium_print_operand_address): Likewise.
14030 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14031 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14032 (xtensa_expand_conditional_branch): Likewise.
14033 (xtensa_copy_incoming_a7): Likewise.
14034 (xtensa_output_literal): Likewise.
14035 * dfp.c (decimal_real_maxval): Likewise.
14036 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14039 Alan Hayward <alan.hayward@arm.com>
14040 David Sherwood <david.sherwood@arm.com>
14042 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14043 (mode_nunits_inline): Likewise.
14044 (mode_inner_inline): Likewise.
14045 (mode_unit_size_inline): Likewise.
14046 (mode_unit_precision_inline): Likewise.
14047 (emit_insn_modes_h): Likewise. Also emit a #define of the
14049 (emit_mode_wider): Add an E_ prefix to mode names.
14050 (emit_mode_complex): Likewise.
14051 (emit_mode_inner): Likewise.
14052 (emit_mode_adjustments): Likewise.
14053 (emit_mode_int_n): Likewise.
14054 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14055 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14056 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14057 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14058 (CRC32_BUILTIN, ENTRY): Likewise.
14059 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14060 (aarch64_pop_regs): Likewise.
14061 (aarch64_process_components): Likewise.
14062 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14063 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14064 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14065 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14066 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14067 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14069 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14070 (pcmpistr): Likewise.
14071 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14072 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14073 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14075 * config/rl78/rl78.c (mduc_regs): Likewise.
14076 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14077 (htm_expand_builtin): Likewise.
14078 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14079 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14080 * config/xtensa/xtensa.c (print_operand): Likewise.
14081 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14082 (NUM_MODE_VECTOR_INT): Likewise.
14083 * genoutput.c (null_operand): Likewise.
14084 (output_operand_data): Likewise.
14085 * genrecog.c (print_parameter_value): Likewise.
14086 * lra.c (debug_operand_data): Likewise.
14088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14090 * dwarf2out.c (scompare_loc_descriptor_wide)
14091 (scompare_loc_descriptor_narrow): New functions, split out from...
14092 (scompare_loc_descriptor): ...here.
14093 * expmed.c (emit_store_flag_int): New function, split out from...
14094 (emit_store_flag): ...here.
14096 2017-08-30 Richard Biener <rguenther@suse.de>
14098 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14099 (dwarf2out_early_finish): Move setting of AT_pubnames from
14100 early debug output to early finish.
14102 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14104 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14105 and -mdata-region to the assembler.
14107 2017-08-30 Richard Biener <rguenther@suse.de>
14109 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14111 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14112 (dwarf2out_early_global_decl): Only generate a DIE for the
14113 abstract origin if it doesn't already exist or is a declaration DIE.
14114 (resolve_addr): Do not add the linkage name twice when
14115 generating a stub DIE for the DW_TAG_GNU_call_site target.
14117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14119 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14120 Use machine_mode rather than int for arg1_mode.
14122 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14125 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14126 that the second argument of the built-in functions to unpack
14127 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14128 switch statement instead a lot of if statements.
14129 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14130 Allow 64-bit values to be in Altivec registers as well as
14131 traditional floating point registers.
14132 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14134 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14136 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14137 MAX_REGS_PER_ADDRESS == 1.
14139 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14141 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14142 * config/i386/i386.c (ix86_option_override_internal): Simplify
14143 setting of opts->x_flag_entry.
14145 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14146 Jakub Jelinek <jakub@redhat.com>
14147 Richard Biener <rguenther@suse.de>
14149 PR tree-optimization/81503
14150 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14151 folded constant fits in the target type; reorder tests for clarity.
14153 2017-08-29 Martin Liska <mliska@suse.cz>
14155 * passes.def: Include pass_lower_switch.
14156 * stmt.c (dump_case_nodes): Remove and move to
14157 tree-switch-conversion.
14158 (case_values_threshold): Likewise.
14159 (expand_switch_as_decision_tree_p): Likewise.
14160 (emit_case_decision_tree): Likewise.
14161 (expand_case): Likewise.
14162 (balance_case_nodes): Likewise.
14163 (node_has_low_bound): Likewise.
14164 (node_has_high_bound): Likewise.
14165 (node_is_bounded): Likewise.
14166 (emit_case_nodes): Likewise.
14167 (struct simple_case_node): New struct.
14168 (add_case_node): Remove.
14169 (emit_case_dispatch_table): Use vector instead of case_list.
14170 (reset_out_edges_aux): Remove.
14171 (compute_cases_per_edge): Likewise.
14172 (expand_case): Build list of simple_case_node.
14173 (expand_sjlj_dispatch_table): Use it.
14174 * tree-switch-conversion.c (struct case_node): Moved from
14175 stmt.c and adjusted.
14176 (emit_case_nodes): Likewise.
14177 (node_has_low_bound): Likewise.
14178 (node_has_high_bound): Likewise.
14179 (node_is_bounded): Likewise.
14180 (case_values_threshold): Likewise.
14181 (reset_out_edges_aux): Likewise.
14182 (compute_cases_per_edge): Likewise.
14183 (add_case_node): Likewise.
14184 (dump_case_nodes): Likewise.
14185 (balance_case_nodes): Likewise.
14186 (expand_switch_as_decision_tree_p): Likewise.
14187 (emit_jump): Likewise.
14188 (emit_case_decision_tree): Likewise.
14189 (try_switch_expansion): Likewise.
14190 (do_jump_if_equal): Likewise.
14191 (emit_cmp_and_jump_insns): Likewise.
14192 (fix_phi_operands_for_edge): New function.
14193 (record_phi_operand_mapping): Likewise.
14194 (class pass_lower_switch): New pass.
14195 (pass_lower_switch::execute): New function.
14196 (make_pass_lower_switch): Likewise.
14197 (conditional_probability):
14198 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14199 * tree-pass.h: Add make_pass_lower_switch.
14201 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14204 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14207 2017-08-29 Richard Biener <rguenther@suse.de>
14209 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14210 we do not add a DW_AT_inline attribute twice.
14211 (gen_subprogram_die): Remove code setting DW_AT_inline on
14212 DECL_ABSTRACT_P nodes.
14214 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14216 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14217 calls to internal functions.
14218 (gimplify_modify_expr): Likewise.
14219 * tree-call-cdce.c (use_internal_fn): Likewise.
14220 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14221 (convert_to_divmod): Set the nothrow flag.
14222 * tree-if-conv.c (predicate_mem_writes): Likewise.
14223 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14224 (vectorizable_call): Likewise.
14225 (vectorizable_store): Likewise.
14226 (vectorizable_load): Likewise.
14227 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14228 (vect_recog_mask_conversion_pattern): Likewise.
14230 2017-08-29 Martin Liska <mliska@suse.cz>
14233 * gcc.c (driver_handle_option): Add new argument.
14234 * opts-common.c (handle_option): Pass
14235 target_option_override_hook.
14236 * opts-global.c (lang_handle_option): Add new option.
14237 (set_default_handlers): Add new argument.
14238 (decode_options): Likewise.
14239 * opts.c (target_handle_option): Likewise.
14240 (common_handle_option): Call target_option_override_hook.
14241 * opts.h (struct cl_option_handler_func): Add hook for
14242 target option override.
14243 (struct cl_option_handlers): Likewise.
14244 (set_default_handlers): Add new argument.
14245 (decode_options): Likewise.
14246 (common_handle_option): Likewise.
14247 (target_handle_option): Likewise.
14248 * toplev.c (toplev::main): Pass targetm.target_option.override
14251 2017-08-29 Richard Biener <rguenther@suse.de>
14252 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14254 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14255 life to the active subtree.
14257 2017-08-28 Jeff Law <law@redhat.com>
14259 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14260 derive_equivalences.
14261 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14262 Code moved into....
14263 (edge_info::derive_equivalences): New private member function
14265 * tree-ssa-dom.c (class edge_info): Changed from a struct
14266 to a class. Add ctor/dtor, methods and data members.
14267 (edge_info::edge_info): Renamed from allocate_edge_info.
14268 Initialize additional members.
14269 (edge_info::~edge_info): New.
14270 (free_dom_edge_info): Delete the edge info.
14271 (record_edge_info): Use new class & associated member functions.
14272 Tighten forms for testing for edge equivalences.
14273 (record_temporary_equivalences): Iterate over the simple
14274 equivalences rather than assuming there's only one per edge.
14275 (cprop_into_successor_phis): Iterate over the simple
14276 equivalences rather than assuming there's only one per edge.
14277 (optimize_stmt): Use operand_equal_p rather than pointer
14278 equality for mini-DSE code.
14280 2017-08-28 Nathan Sidwell <nathan@acm.org>
14282 * gcc.c (execute): Fold SIGPIPE handling into switch
14283 statement. Adjust internal error message.
14285 2017-08-28 Richard Biener <rguenther@suse.de>
14288 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
14289 Do nothing for removed DIEs.
14291 2017-08-28 Richard Biener <rguenther@suse.de>
14293 PR tree-optimization/81977
14294 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
14297 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
14300 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
14301 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
14302 using targetm.gen_mem_thread_fence.
14304 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14307 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14308 predicate to register_operand. Reorder operands.
14309 (*btr<mode>): Ditto.
14310 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14311 (*btr<mode>_mask): Ditto.
14313 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14315 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14316 * config/rs6000/xmmintrin.h: New file.
14317 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14319 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14322 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14323 parameter and_insn and return it.
14324 (recombine_lvx_pattern): Insert a copy to ensure availability of
14325 the base register of the copied masking operation at the point of
14326 the instruction replacement.
14327 (recombine_stvx_pattern): Likewise.
14329 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14331 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14332 undocumented switches.
14333 (-mpower9-dform-vector): Likewise.
14334 (-mpower9-dform): Likewise.
14335 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14336 comments to delete references to -mpower9-dform* switches.
14337 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14338 Delete reference to -mpower9-dform* switches, test for
14339 -mpower9-vector instead.
14340 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14341 (OTHER_P9_VECTOR_MASKS): Likewise.
14342 (POWERPC_MASKS): Likewise.
14343 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14344 tests against -mpower9-dform* to -mpower9-vector. Delete code
14345 that checked for -mpower9-dform* consistancy with other options.
14346 Add test for -mpower9-misc to enable other power9 switches.
14347 (rs6000_init_hard_regno_mode_ok): Likewise.
14348 (rs6000_option_override_internal): Likewise.
14349 (rs6000_emit_prologue): Likewise.
14350 (rs6000_emit_epilogue): Likewise.
14351 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14352 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14353 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14355 (emit_fusion_p9_store): Likewise.
14356 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14357 resetting these macros if the assembler does not support ISA 3.0
14359 (TARGET_P9_DFORM_VECTOR): Likewise.
14360 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14361 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14363 2017-08-25 Alan Modra <amodra@gmail.com>
14365 PR rtl-optimization/81747
14366 * cse.c (cse_extended_basic_block): Don't attempt to record
14367 equivalences for degenerate conditional branches that branch
14368 to their fall-through.
14370 2017-08-24 Martin Sebor <msebor@redhat.com>
14372 PR middle-end/81908
14373 * gimple-fold.c (size_must_be_zero_p): New function.
14374 (gimple_fold_builtin_memory_op): Call it.
14376 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14378 * config/rs6000/mm_malloc.h: New file.
14380 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14382 PR tree-optimization/81913
14383 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14384 analysis when either IVs in condition can wrap.
14386 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14388 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14389 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14391 2017-08-24 Richard Biener <rguenther@suse.de>
14394 * targhooks.c (default_target_can_inline_p): Properly
14395 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14396 is present and always compare.
14397 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14398 infer -mfpmath=sse from TARGET_SSE_P.
14399 (ix86_can_inline_p): Properly use target_option_default_node when
14400 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14402 2017-08-24 Richard Biener <rguenther@suse.de>
14405 * dwarf2out.c (output_die): Handle flag_generate_offload like
14407 (output_comp_unit): Likewise.
14408 (gen_array_type_die): Likewise.
14409 (dwarf2out_early_finish): Likewise.
14410 (note_variable_value_in_expr): Likewise.
14411 (dwarf2out_finish): Likewise. Adjust assert.
14412 * cgraphunit.c (symbol_table::compile): Move setting of
14413 flag_generate_offload earlier ...
14414 (symbol_table::finalize_compilation_unit): ... here, before
14415 early debug finalization.
14417 2017-08-24 Richard Biener <rguenther@suse.de>
14419 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14420 and ipa-fnsummary.h.
14421 (ix86_can_inline_p): When ix86_fpmath flags do not match
14422 check whether the callee uses FP math at all.
14424 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14426 PR middle-end/81931
14427 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14428 instead of TYPE_PRECISION.
14430 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14431 Alan Hayward <alan.hayward@arm.com>
14432 David Sherwood <david.sherwood@arm.com>
14434 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14436 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14438 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14440 (ix86_frame::stack_realign_allocate): New field.
14441 (struct machine_frame_state): Modify comments.
14442 (machine_frame_state::sp_realigned_fp_end): New field.
14443 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14444 layout calculation.
14445 (sp_valid_at): Add assertion to assure no attempt to access invalid
14446 offset of a realigned stack.
14447 (fp_valid_at): Likewise.
14448 (choose_baseaddr): Modify comments.
14449 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14450 ix86_expand_prologue.
14451 (ix86_expand_prologue): Modify stack realignment and allocation.
14452 (ix86_expand_epilogue): Modify comments.
14453 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14454 avx2_runtime, avx512f, and avx512f_runtime.
14456 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14458 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14459 (mstackrealign): Do not init to -1.
14460 * config/i386/i386.c (ix86_option_override_internal):
14461 Check opts_set, not opts when setting default value of
14462 opts->x_ix86_force_align_arg_pointer.
14464 2017-08-23 Richard Biener <rguenther@suse.de>
14466 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14467 lang_hooks.decl_printable_name.
14468 * print-rtl-function.c (print_rtx_function): Likewise.
14469 * tree-pretty-print.c (dump_function_header): Likewise.
14471 2017-08-23 Richard Biener <rguenther@suse.de>
14474 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14475 -g0 at compile-time.
14477 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14479 PR middle-end/19706
14480 * doc/sourcebuild.texi (Other hardware attributes):
14483 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14485 PR middle-end/19706
14486 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14487 Fix single-use check.
14489 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14491 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14493 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14495 * doc/install.texi: Modify to add more details on running selected
14498 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14500 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14501 is combined with -mabi=ms.
14502 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14505 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14507 PR tree-optimization/81488
14508 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14509 and cached_basis fields.
14510 (MAX_SPREAD): New constant.
14511 (alloc_cand_and_find_basis): Initialize new fields.
14512 (clear_visited): New function.
14513 (create_phi_basis_1): Rename from create_phi_basis, set visited
14514 and cached_basis fields.
14515 (create_phi_basis): New wrapper function.
14516 (phi_add_costs_1): Rename from phi_add_costs, add spread
14517 parameter, set visited field, short-circuit when limits reached.
14518 (phi_add_costs): New wrapper function.
14519 (record_phi_increments_1): Rename from record_phi_increments, set
14521 (record_phi_increments): New wrapper function.
14522 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14523 (phi_incr_cost): New wrapper function.
14524 (all_phi_incrs_profitable_1): Rename from
14525 all_phi_incrs_profitable, set visited field.
14526 (all_phi_incrs_profitable): New wrapper function.
14528 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14529 Alan Hayward <alan.hayward@arm.com>
14530 David Sherwood <david.sherwood@arm.com>
14532 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14533 that takes the outer and inner modes.
14534 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14535 comparison as the canonical test for a paradoxical subreg.
14536 * combine.c (simplify_set): Use paradoxical_subreg_p.
14537 (make_extraction): Likewise.
14538 (force_to_mode): Likewise.
14539 (rtx_equal_for_field_assignment_p): Likewise.
14540 (gen_lowpart_for_combine): Likewise.
14541 (simplify_comparison): Likewise.
14542 * cse.c (equiv_constant): Likewise.
14543 * expmed.c (store_bit_field_1): Likewise.
14544 * final.c (alter_subreg): Likewise.
14545 * fwprop.c (propagate_rtx): Likewise.
14546 (forward_propagate_subreg): Likewise.
14547 * ira-conflicts.c (ira_build_conflicts): Likewise.
14548 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14549 * lra-constraints.c (curr_insn_transform): Likewise.
14550 (split_reg): Likewise.
14551 * lra-eliminations.c (move_plus_up): Likewise.
14552 (lra_eliminate_regs_1): Likewise.
14553 * recog.c (general_operand): Likewise.
14554 * ree.c (combine_reaching_defs): Likewise.
14555 * reload.c (push_reload): Likewise.
14556 (find_reloads): Likewise.
14557 * reload1.c (elimination_effects): Likewise.
14558 (compute_reload_subreg_offset): Likewise.
14559 (choose_reload_regs): Likewise.
14560 * rtlanal.c (subreg_lsb_1): Likewise.
14561 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14562 (simplify_subreg): Likewise.
14563 * var-tracking.c (track_loc_p): Likewise.
14564 * emit-rtl.c (byte_lowpart_offset): Likewise.
14565 (paradoxical_subreg_p): Delete out-of-line definition.
14567 2017-08-22 Jeff Law <law@redhat.com>
14569 PR tree-optimization/81741
14570 PR tree-optimization/71947
14571 * tree-ssa-dom.c: Include tree-inline.h.
14572 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14573 equivalences if one is more expensive to compute than the other.
14574 * tree-ssa-scopedtables.h (class const_or_copies): Make
14575 record_const_or_copy_raw method private.
14576 (class avail_exprs_stack): New method simplify_binary_operation.
14577 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14578 avail_exprs_stack::simplify_binary_operation as needed.
14579 (avail_exprs_stack::simplify_binary_operation): New function.
14581 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14583 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14584 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14585 (DOT_SYMBOLS): Likewise.
14586 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14587 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14588 (RS6000_ABI_NAME): Likewise.
14589 (TARGET_CMODEL): Likewise.
14590 (TOC_SECTION_ASM_OP): Likewise.
14591 (SET_CMODEL): New macro.
14592 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14594 2017-08-22 Richard Biener <rguenther@suse.de>
14596 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14597 to allow for free-lang-data replacements similar to verify_type_variant.
14599 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14600 Alan Hayward <alan.hayward@arm.com>
14601 David Sherwood <david.sherwood@arm.com>
14603 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14604 VOIDmode for the LABEL_REF.
14606 2017-08-22 Richard Biener <rguenther@suse.de>
14608 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14610 2017-08-22 Richard Biener <rguenther@suse.de>
14612 * common.opt (feliminate-dwarf2-dups): Ignore.
14613 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14614 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14615 same_die_p_wrap, compute_section_prefix,
14616 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14617 (comdat_symbol_id, comdat_symbol_number): Likewise.
14618 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14620 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14621 (output_die): Mark unreachable path unreachable.
14622 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14623 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14624 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14625 (dwarf2out_early_finish): Likewise.
14627 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14629 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14631 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14634 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14635 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14636 (avr_attribute_table) <io, io_low, address>: Initialize
14637 .decl_required with true.
14639 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14641 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14642 undocumented debugging options.
14643 (-mvsx-scalar-double): Likewise.
14644 (-mallow-df-permute): Likewise.
14645 (-mvectorize-builtins): Likewise.
14646 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14647 (rs6000_builtin_vectorized_function): Likewise.
14648 (rs6000_builtin_md_vectorized_function): Likewise.
14649 (rs6000_opt_vars): Likewise.
14651 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14654 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14655 (*btrq_imm): Rename from *btrq.
14656 (*btcq_imm): Rename from *btcq.
14657 (btsc): New code attribute.
14658 (*<btsc><mode>): New insn pattern.
14659 (*btr<mode>): Ditto.
14660 (*<btsc><mode>_mask): New insn_and_split pattern.
14661 (*btr<mode>_mask): Ditto.
14663 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14664 Alan Hayward <alan.hayward@arm.com>
14665 David Sherwood <david.sherwood@arm.com>
14667 * function.c (pad_below): Simplify padding calculation.
14669 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14670 Alan Hayward <alan.hayward@arm.com>
14671 David Sherwood <david.sherwood@arm.com>
14673 * target.def (function_prologue): Remove frame size argument.
14674 (function_epilogue): Likewise.
14675 * doc/tm.texi: Regenerate.
14676 * final.c (final_start_function): Update call to function_prologue.
14677 (final_end_function): Update call to function_epilogue.
14678 (default_function_pro_epilogue): Remove frame size argument.
14679 * output.h (default_function_pro_epilogue): Likewise.
14680 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14681 (arm_output_function_prologue): Likewise.
14682 * config/frv/frv.c (frv_function_prologue): Likewise.
14683 (frv_function_epilogue): Likewise.
14684 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14685 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14686 (ia64_output_function_epilogue): Likewise.
14687 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14688 (m32r_output_function_epilogue): Likewise.
14689 * config/microblaze/microblaze.c (microblaze_function_prologue)
14690 (microblaze_function_epilogue): Likewise.
14691 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14692 (mips_output_function_epilogue): Likewise.
14693 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14694 (mmix_target_asm_function_epilogue): Likewise.
14695 * config/msp430/msp430.c (msp430_start_function): Likewise.
14696 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14697 (nds32_asm_function_epilogue): Likewise.
14698 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14699 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14700 (pa_output_function_epilogue): Likewise.
14701 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14702 (rs6000_output_function_epilogue): Likewise.
14703 * config/rl78/rl78.c (rl78_start_function): Likewise.
14704 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14705 (rs6000_output_function_epilogue): Likewise.
14706 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14707 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14708 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14709 (sparc_asm_function_epilogue): Likewise.
14711 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14713 * tree.h (type_has_mode_precision_p): New function.
14714 * convert.c (convert_to_integer_1): Use it.
14715 * expr.c (expand_expr_real_2): Likewise.
14716 (expand_expr_real_1): Likewise.
14717 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14718 * match.pd: Likewise.
14719 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14720 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14721 * tree-tailcall.c (process_assignment): Likewise.
14722 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14723 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14724 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14725 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14726 (vectorizable_assignment): Likewise.
14727 (vectorizable_shift): Likewise.
14728 (vectorizable_operation): Likewise.
14729 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14731 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14733 * match.pd: Add pow (C, x) simplification.
14735 2017-08-21 Richard Biener <rguenther@suse.de>
14737 PR tree-optimization/81900
14738 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14739 for blocks with abnormal predecessors.
14740 (compute_antic): Do not set visited flag prematurely.
14742 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14745 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14747 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14749 * stor-layout.h (vector_type_mode): Move to...
14750 * tree.h (vector_type_mode): ...here.
14751 * stor-layout.c (vector_type_mode): Move to...
14752 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14754 2017-08-21 Richard Biener <rguenther@suse.de>
14756 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14757 register_external_die hooks.
14758 (debug_false_tree_charstarstar_uhwistar): Declare.
14759 (debug_nothing_tree_charstar_uhwi): Likewise.
14760 * debug.c (do_nothing_debug_hooks): Adjust.
14761 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14762 (debug_nothing_tree_charstar_uhwi): Likewise.
14763 * dbxout.c (dbx_debug_hooks): Adjust.
14764 (xcoff_debug_hooks): Likewise.
14765 * sdbout.c (sdb_debug_hooks): Likewise.
14766 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14767 * dwarf2out.c (macinfo_label_base): New global.
14768 (dwarf2out_register_external_die): New function for the
14769 register_external_die hook.
14770 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14771 (dwarf2_debug_hooks): Use them.
14772 (dwarf2_lineno_debug_hooks): Adjust.
14773 (struct die_struct): Add with_offset flag.
14774 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14775 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14776 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14777 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14778 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14779 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14780 defining section names for the early LTO debug variants.
14781 (reset_indirect_string): New helper.
14782 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14783 (print_dw_val): Add support for offsetted symbol references.
14784 (get_ultimate_context): Split out from is_cxx.
14785 (is_cxx): Use get_ultimate_context.
14786 (is_fortran): Add decl overload.
14787 (compute_comp_unit_symbol): Split out worker from
14788 compute_section_prefix.
14789 (compute_section_prefix): Call compute_comp_unit_symbol and
14790 set comdat_type_p here.
14791 (output_die): Skip DIE symbol output for the LTO added one.
14792 Handle DIE symbol references with offset.
14793 (output_comp_unit): Guard section name mangling properly.
14794 For LTO debug sections emit a symbol at the section beginning
14795 which we use to refer to its DIEs.
14796 (add_abstract_origin_attribute): For DIEs registered via
14797 dwarf2out_register_external_die directly refer to the early
14798 DIE rather than indirectly through the shadow one we created.
14799 Remove obsolete call to dwarf2out_abstract_function for
14800 non-function/block origins.
14801 (gen_array_type_die): When generating early LTO debug do
14802 not emit DW_AT_string_length.
14803 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14804 late when in LTO. As suggested place a gcc_unreachable for
14805 the DECL_ABSTRACT_P case.
14806 (gen_subprogram_die): Avoid another specification DIE
14807 for early built declarations/definitions for the late LTO case.
14808 (gen_variable_die): Add type references for late duplicated VLA dies
14810 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14811 we have the abstract instance already.
14812 (process_scope_var): Adjust decl DIE contexts in LTO which
14813 first puts them in limbo.
14814 (gen_decl_die): Do not generate type DIEs late apart from
14815 types for VLAs or for decls we do not yet have a DIE. Do not
14816 call dwarf2out_abstract_function late.
14817 (dwarf2out_early_global_decl): Make sure to create DIEs
14818 for abstract instances of a decl first.
14819 (dwarf2out_late_global_decl): Adjust comment.
14820 (output_macinfo_op): With multiple macro sections use
14821 macinfo_label_base to distinguish labels.
14822 (output_macinfo): Likewise. Update macinfo_label_base.
14823 Pass in the line info label.
14824 (note_variable_value_in_expr): When generating LTO resolve
14825 all variable values here by generating DIEs as needed.
14826 (init_sections_and_labels): Add early LTO debug flag parameter
14827 and generate different sections and names if set. Add generation
14828 counter for the labels so we can have multiple of them.
14829 (reset_dies): Helper to allow DIEs to be output multiple times.
14830 (dwarf2out_finish): When outputting DIEs to the fat part of an
14831 LTO object first reset DIEs.
14832 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14833 (modified_type_die): Check for decl_ultimate_origin being self
14835 (gen_type_die_with_usage): Likewise.
14836 (gen_typedef_die): Allow decl_ultimate_origin being self.
14837 (set_decl_abstract_flags): Remove.
14838 (set_block_abstract_flags): Likewise.
14839 (dwarf2out_abstract_function): Treat the early generated DIEs
14840 as the abstract copy and only add DW_AT_inline and
14841 DW_AT_artificial here and call set_decl_origin_self.
14842 If the DIE has an abstract origin don't do anything.
14843 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14844 if we have none yet (Go fails to build one, PR78628).
14845 (variably_modified_type_p): Prevent endless recursion for Ada
14846 cyclic pointer types.
14847 * lto-streamer-in.c: Include debug.h.
14848 (dref_queue): New global.
14849 (lto_read_tree_1): Stream in DIE references.
14850 (lto_input_tree): Register DIE references.
14851 (input_function): Stream DECL_DEBUG_ARGS.
14852 * lto-streamer-out.c: Include debug.h.
14853 (lto_write_tree_1): Output DIE references.
14854 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14855 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14856 (output_function): Stream DECL_DEBUG_ARGS.
14857 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14858 Stream DECL_ABSTRACT_ORIGIN.
14859 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14860 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14861 DECL_CONTEXT for file-scope decls.
14862 * lto-streamer.h (struct dref_entry): Declare.
14863 (dref_queue): Likewise.
14864 * cfgexpand.c (pass_expand::execute): Do not call the
14865 outlining_inline_function hook here.
14866 * lto-wrapper.c (debug_obj): New global.
14867 (tool_cleanup): Unlink it if required.
14868 (debug_objcopy): New function.
14869 (run_gcc): Handle early debug sections in the IL files by
14870 extracting them to separate files, partially linkin them and
14871 feeding the result back as result to the linker.
14872 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14873 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14874 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14875 sections into a separate segment.
14876 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14878 (darwin_asm_dwarf_section): Likewise.
14879 (darwin_asm_output_dwarf_offset): Likewise.
14880 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14882 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14883 Alan Hayward <alan.hayward@arm.com>
14884 David Sherwood <david.sherwood@arm.com>
14886 * read-md.h (md_reader::record_potential_iterator_use): Replace
14887 pointer argument with an rtx and an index.
14888 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14889 (apply_mode_iterator): Likewise.
14890 (apply_code_iterator): Likewise.
14891 (apply_int_iterator): Likewise.
14892 (apply_subst_iterator): Likewise.
14893 (record_iterator_use): Likewise.
14894 (record_attribute_use): Likewise.
14895 (md_reader::record_potential_iterator_use): Likewise. Update calls
14896 to record_iterator_use and apply_iterator.
14897 (iterator_use): Replace ptr with x and index.
14898 (attribute_use): Likewise.
14899 (apply_attribute_uses): Update calls to apply_iterator.
14900 (apply_iterators): Likewise. Update initialization of iterator_use.
14901 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14902 and record_potential_iterator_use.
14903 (rtx_reader::read_rtx_operand): Likewise.
14905 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14906 Alan Hayward <alan.hayward@arm.com>
14907 David Sherwood <david.sherwood@arm.com>
14909 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14912 2017-08-21 Richard Biener <rguenther@suse.de>
14914 PR middle-end/81884
14915 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14916 at struct end conservatively when comparing common bases.
14918 2017-08-21 Richard Biener <rguenther@suse.de>
14920 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14921 (mem_ref_in_stmt): Remove.
14922 (determine_max_movement): Use ref index to get at the reference.
14923 (invariantness_dom_walker::before_dom_children): Deal with
14924 lim data already initialized.
14925 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14927 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14929 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14930 (TARGET_ISA_ROUND): Ditto.
14931 (TARGET_ROUND): Ditto.
14932 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14933 * config/i386/i386.md: Ditto.
14934 * config/i386/sse.md: Ditto.
14935 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14936 with OPTION_MASK_ISA_SSE4_1.
14938 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14941 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14942 __builtin_ia32_lzcnt_u16.
14944 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14947 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14948 (rs6000_set_current_function): Rewrite function to use it.
14950 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14953 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14954 and TYPE_WARN_IF_NOT_ALIGN.
14955 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14956 (handle_warn_if_not_align): New.
14957 (place_union_field): Call handle_warn_if_not_align.
14958 (place_field): Call handle_warn_if_not_align.
14959 Copy TYPE_WARN_IF_NOT_ALIGN.
14960 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14961 (layout_type): Likewise.
14962 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14964 (tree_decl_common): Add warn_if_not_align.
14965 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14966 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14967 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14968 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14969 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14970 * doc/extend.texi: Document warn_if_not_aligned attribute.
14971 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14973 2017-08-17 Martin Liska <mliska@suse.cz>
14976 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14977 (get_data_dependence): Use it as pointer type.
14978 (distribute_loop): Likewise.
14980 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14982 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14983 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14984 (p8_vmrgow_<mode>_direct): New define_insn.
14985 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14986 handle endianness for vmrgew and vmrgow permute patterns.
14988 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14990 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14991 * config/rs6000/rs6000-cpus.def: Remove comment.
14992 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14993 (POWERPC_MASKS): Likewise.
14994 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14995 use of TARGET_VSX_TIMODE.
14996 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14997 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14998 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14999 (rs6000_option_override_internal): Remove dead code.
15000 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15001 (rs6000_legitimize_reload_address): Likewise.
15002 (rs6000_legitimate_address_p): Likewise.
15003 (rs6000_opt_masks): Delete "vsx-timode".
15004 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15005 from function comment.
15006 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15007 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15008 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15010 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15011 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15012 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15013 (VSX_TI): Likewise.
15015 (define_peephole2): Likewise.
15017 2017-08-17 Martin Sebor <msebor@redhat.com>
15020 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15021 past the end of an array.
15022 (test_pp_format): Add test cases.
15024 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15026 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15027 missing ECF_NOTHROW flags.
15029 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15032 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15033 operands residing in integer registers.
15034 (*vsx_le_perm_load_<mode>): Likewise.
15035 (*vsx_le_perm_store_<mode>): Likewise.
15036 (define_peephole2): Add peepholes to optimize the above.
15038 2017-08-17 Marek Polacek <polacek@redhat.com>
15040 PR middle-end/81814
15041 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15042 to mimic what shorten_compare did. Change the return type to bool.
15043 (fold_cond_expr_with_comparison): Update call to
15044 operand_equal_for_comparison_p.
15045 (fold_ternary_loc): Likewise.
15047 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15049 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15051 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15053 2017-08-17 Richard Biener <rguenther@suse.de>
15055 * tree-ssa-structalias.c (solve_graph): When propagating
15056 to successors update the graphs succ edges and avoid duplicate work.
15058 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15061 * config/i386/i386.c (ix86_option_override_internal): Save target
15062 specific options after ix86_stack_protector_guard_reg was changed.
15064 2017-08-17 Richard Biener <rguenther@suse.de>
15066 PR tree-optimization/81827
15067 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15068 (new_var_info): Initialize it conservatively.
15069 (get_call_vi): Mark register vars.
15070 (new_scalar_tmp_constraint_exp): Likewise.
15071 (handle_rhs_call): Likewise.
15072 (handle_const_call): Likewise.
15073 (create_function_info_for): Likewise.
15074 (solve_constraints): Sort varinfos to separate register from
15075 non-register vars to pack points-to solution bitmaps during
15078 2017-08-17 Marek Polacek <polacek@redhat.com>
15080 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15082 2017-08-17 Richard Biener <rguenther@suse.de>
15084 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15085 to true when overflow is undefined and we saturated the result.
15087 2017-08-17 Alan Modra <amodra@gmail.com>
15090 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15091 Don't use store multiple if only one reg needs saving.
15092 (interesting_frame_related_regno): New function.
15093 (rs6000_frame_related): Don't emit frame info for regs that
15095 (rs6000_emit_epilogue): Likewise.
15097 2017-08-16 Nathan Sidwell <nathan@acm.org>
15099 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15100 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15101 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15102 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15103 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15104 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15105 (verify_type): Adjust for TYPE_BINFO move.
15106 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15107 process TYPE_BINFO directly.
15108 (hash_tree): Likewise.
15109 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15111 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15114 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15116 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15118 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15121 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15122 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15123 Add ix86_binary_operator_ok to insn constraint.
15124 (*iordi_1_bts): Ditto.
15125 (*xordi_1_btc): Ditto.
15126 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15127 Update corresponding peephole2 pattern.
15131 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15133 PR tree-optimization/81832
15134 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15135 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15137 2017-08-16 Marek Polacek <polacek@redhat.com>
15140 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15141 regarding size_zero_node.
15143 2017-08-16 Martin Liska <mliska@suse.cz>
15146 * config.gcc: Respect previously set extra_objs in case
15149 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15151 PR tree-optimization/81835
15152 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15153 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15154 not depend on the phi.
15156 2017-08-16 Alan Modra <amodra@gmail.com>
15158 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15161 2017-08-16 Alan Modra <amodra@gmail.com>
15163 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15164 (save_reg_p): ..into this. Update all callers.
15165 (first_reg_to_save): Simplify.
15167 2017-08-16 Alan Modra <amodra@gmail.com>
15169 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15172 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15176 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15177 constant count if that count is less than 32.
15179 2017-08-15 Nathan Sidwell <nathan@acm.org>
15181 * gcc.c (execute): Emit friendlier message if inferior is killed
15182 by an external cause.
15184 2017-08-15 Richard Biener <rguenther@suse.de>
15186 PR tree-optimization/81790
15187 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15188 CONSTRUCTORs from simplifying and VN.
15190 2017-08-14 Martin Sebor <msebor@redhat.com>
15192 * builtin-attrs.def: Add comments.
15194 2017-08-14 Martin Sebor <msebor@redhat.com>
15197 * doc/extend.texi (attribute nonstring): Document new attribute.
15199 2017-08-14 Martin Sebor <msebor@redhat.com>
15202 * tree-diagnostic.c (default_tree_printer): Handle %G.
15203 * gimple-pretty-print.h (percent_G_format): Declare new function.
15204 * gimple-pretty-print.c (percent_G_format): Define.
15205 * tree-pretty-print.c (percent_K_format): Add argument.
15207 2017-08-14 Martin Sebor <msebor@redhat.com>
15209 PR translation/79998
15210 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15211 Remove a stray space.
15213 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15216 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15217 (*iordi_1_bts): Ditto.
15218 (*xordi_1_btc): Ditto.
15220 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15223 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15224 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15226 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15227 quoted strings, and make more translator-friendly.
15228 (darwin_rs6000_override_options): Likewise.
15229 (rs6000_option_override_internal): Likewise.
15230 (rs6000_return_in_memory): Fix overlong line.
15231 (init_cmulative_args): Use quoted strings, and make more
15232 translator-friendly.
15233 (rs6000_pass_by_reference): Fix overlong line.
15234 (def_builtin): Use quoted strings.
15235 (altivec_expand_predicate_builtin): Use quoted strings, and make
15236 more translator-friendly.
15237 (htm_expand_builtin): Use quoted strings.
15238 (cpu_expand_builtin): Use quoted strings, and make more
15239 translator-friendly.
15240 (altivec_expand_builtin): Likewise.
15241 (paired_expand_predicate_builtin): Likewise.
15242 (rs6000_invalid_builtin): Likewise.
15243 (builtin_function_type): Use quoted strings.
15244 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15245 more translator-friendly.
15246 (rs6000_trampoline_init): Likewise.
15247 (rs6000_handle_altivec_attribute): Likewise.
15248 (rs6000_inner_target_options): Use quoted strings.
15249 (rs6000_disable_incompatible_switches): Likewise.
15250 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15251 strings, and make more translator-friendly.
15252 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15254 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15256 PR tree-optimization/81799
15257 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15258 cond_expr to simple gimple operand.
15260 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15262 PR middle-end/46932
15263 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15265 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
15269 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
15270 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
15271 TARGET_GASISR_PROLOGUES.
15272 * config/avr/avr.c (avr_option_override): Same.
15273 (avr_pass_pre_proep::execute): Same.
15275 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
15278 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
15279 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
15281 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
15283 * config/i386/i386.md (*load_tp_<mode>): Redefine as
15284 define_insn_and_split. Split to a memory load from 0 in
15285 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
15286 using PTR mode iterator.
15287 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
15288 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
15289 (*add_tp_<mode>): Redefine as define_insn_and_split.
15290 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
15291 address space. Merge with *add_tp_x32 using PTR mode iterator.
15292 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
15293 Split to an add with a memory load from 0 in
15294 DEFAULT_TLS_SEG_REG address space.
15296 2017-08-12 Andrew Pinski <apinski@cavium.com>
15298 * config/aarch64/aarch64-option-extensions.def (rdma):
15299 Fix feature string to what Linux prints out in /proc/cpuinfo.
15301 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15304 * dwarf2out.c (modified_type_die): For C typedef types that have
15305 an ultimate origin, process the ultimate origin instead of the
15307 (gen_typedef_die): Assert that input DECLs have no ultimate
15309 (gen_type_die_with_usage): For typedef variants that have an
15310 ultimate origin, just call gen_decl_die on the original DECL.
15311 (process_scope_var): Avoid creating DIEs for local typedefs and
15312 concrete static variables.
15314 2017-08-12 Alan Modra <amodra@gmail.com>
15318 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15319 match gnu-user.h startfile.
15320 (ENDFILE_LINUX_SPEC): Similarly.
15322 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15325 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15327 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15329 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15330 * config/aarch64/aarch64.md (mov<mode>): Change.
15331 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15332 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15333 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15335 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15337 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15338 for storage order barriers.
15340 2017-08-11 Martin Liska <mliska@suse.cz>
15343 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15344 variables of void type.
15346 2017-08-11 Martin Liska <mliska@suse.cz>
15348 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15349 TARGET_SUPPORTS_ALIASES.
15350 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15351 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15352 (optimize_weakref): Likewise.
15353 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15354 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15355 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15357 2017-08-11 Martin Liska <mliska@suse.cz>
15360 * config/i386/i386.c (make_resolver_func): Do complete
15361 refactoring of the function.
15363 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15366 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15367 * config/i386/i386.c (ix86_stack_protect_guard): Use
15368 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15369 * doc/invoke.texi (x86 Options): Document
15370 -mstack-protector-guard-symbol= option.
15372 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15374 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15375 * config/i386/i386.c (ix86_split_stack_guard): New function.
15376 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15377 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15378 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15379 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15380 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15381 (split_stack_space_check): Call ix86_split_stack_guard.
15383 2017-08-10 Martin Sebor <msebor@redhat.com>
15385 * print-tree.c (print_node): Print location using the established
15387 Replace spaces with colons.
15388 (debug_raw, debug): Ditto.
15390 2017-08-10 Martin Sebor <msebor@redhat.com>
15393 * pretty-print.c (pp_format): Correct the handling of %s precision.
15395 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15398 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15400 (ix86_finalize_stack_frame_flags): This. Also clear
15401 frame_pointer_needed if -fno-omit-frame-pointer is used without
15403 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15404 with ix86_finalize_stack_frame_flags.
15405 (ix86_expand_epilogue): Likewise.
15406 (ix86_expand_split_stack_prologue): Likewise.
15407 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15409 2017-08-10 Martin Liska <mliska@suse.cz>
15412 * c-attribs.c (handle_target_attribute):
15413 Report warning for an empty string argument of target attribute.
15415 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15418 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15420 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15421 or DECL_NONLOCAL labels.
15422 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15423 or DECL_NONLOCAL labels here.
15425 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15427 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15428 to indicate when early gimple folding has been disabled.
15429 (rs6000_gimple_fold_builtin): Add debug content.
15430 (rs6000_invalid_builtin): Fix whitespace.
15431 (rs6000_expand_builtin): Fix whitespace.
15432 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15434 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15437 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15438 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15440 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15442 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15443 (qdf24xx): Likewise.
15444 * config/aarch64/aarch64.md: Include falkor.md.
15445 * config/aarch64/falkor.md: New.
15447 2017-08-09 Marek Polacek <polacek@redhat.com>
15450 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15451 * diagnostic.c (emit_diagnostic): Add a comment.
15452 (emit_diagnostic_valist): New function.
15454 2017-08-09 Marek Polacek <polacek@redhat.com>
15457 * input.c (make_location): New overload.
15458 * input.h (make_location): Declare.
15460 2017-08-08 Alan Modra <amodra@gmail.com>
15461 H.J. Lu <hongjiu.lu@intel.com>
15464 * gcc.c (NO_PIE_SPEC): Delete.
15465 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15467 (LINK_PIE_SPEC): ..to here.
15468 (LINK_COMMAND_SPEC): Support -no-pie.
15469 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15470 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15472 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15473 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15474 (ENDFILE_CRTEND_SPEC): Similarly.
15476 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15479 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15480 (mstack-protector-guard-offset=): Ditto.
15481 * config/i386/i386.c (ix86_option_override): Handle
15482 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15484 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15485 ix86_stack_protect_guard_offset variables.
15486 (TARGET_STACK_PROTECT_GUARD): Always define.
15487 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15488 and -mstack-protector-guard-offset= options.
15490 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15492 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15493 boundary case for the last candidate.
15495 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15497 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15498 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15500 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15502 PR middle-end/19706
15503 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15504 * config/aarch64/aarch64-builtins.c
15505 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15506 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15507 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15509 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15510 Andrew Pinski <pinskia@gmail.com>
15512 PR middle-end/19706
15513 * internal-fn.def (XORSIGN): New.
15514 * optabs.def (xorsign_optab): New.
15515 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15516 (convert_expand_mult_copysign): New.
15517 (pass_optimize_widening_mul::execute): Call
15518 convert_expand_mult_copysign.
15520 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15522 PR tree-optimization/81354
15523 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15524 Insert on edges rather than explicitly creating landing pads.
15525 (analyze_candidates_and_replace): Commit edge inserts.
15527 2017-08-08 Richard Biener <rguenther@suse.de>
15529 PR middle-end/81719
15530 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15531 (expand_simple_operations): Also look through ADDR_EXPRs with
15532 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15534 2017-08-08 Richard Biener <rguenther@suse.de>
15536 PR tree-optimization/81723
15537 * tree-vect-slp.c (struct bst_traits): New hash traits.
15538 (bst_fail): New global.
15539 (vect_build_slp_tree_2): New worker, split out from ...
15540 (vect_build_slp_tree): ... this now wrapping it with using
15541 bst_fail set to cache SLP tree build fails. Properly handle
15543 (vect_analyze_slp_instance): Allocate and free bst_fail.
15545 2017-08-08 Martin Liska <mliska@suse.cz>
15548 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15549 LABEL_DECLs that can be from a different function.
15551 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15553 PR tree-optimization/81744
15554 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15555 loop's number of iterations.
15557 2017-08-08 Martin Liska <mliska@suse.cz>
15559 * asan.c: Include header files.
15560 * attribs.c (build_decl_attribute_variant): New function moved
15562 (build_type_attribute_qual_variant): Likewise.
15563 (cmp_attrib_identifiers): Likewise.
15564 (simple_cst_list_equal): Likewise.
15565 (omp_declare_simd_clauses_equal): Likewise.
15566 (attribute_value_equal): Likewise.
15567 (comp_type_attributes): Likewise.
15568 (build_type_attribute_variant): Likewise.
15569 (lookup_ident_attribute): Likewise.
15570 (remove_attribute): Likewise.
15571 (merge_attributes): Likewise.
15572 (merge_type_attributes): Likewise.
15573 (merge_decl_attributes): Likewise.
15574 (merge_dllimport_decl_attributes): Likewise.
15575 (handle_dll_attribute): Likewise.
15576 (attribute_list_equal): Likewise.
15577 (attribute_list_contained): Likewise.
15578 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15579 (lookup_attribute_by_prefix): Likewise.
15580 * bb-reorder.c: Include header files.
15581 * builtins.c: Likewise.
15582 * calls.c: Likewise.
15583 * cfgexpand.c: Likewise.
15584 * cgraph.c: Likewise.
15585 * cgraphunit.c: Likewise.
15586 * convert.c: Likewise.
15587 * dwarf2out.c: Likewise.
15588 * final.c: Likewise.
15589 * fold-const.c: Likewise.
15590 * function.c: Likewise.
15591 * gimple-expr.c: Likewise.
15592 * gimple-fold.c: Likewise.
15593 * gimple-pretty-print.c: Likewise.
15594 * gimple.c: Likewise.
15595 * gimplify.c: Likewise.
15596 * hsa-common.c: Likewise.
15597 * hsa-gen.c: Likewise.
15598 * internal-fn.c: Likewise.
15599 * ipa-chkp.c: Likewise.
15600 * ipa-cp.c: Likewise.
15601 * ipa-devirt.c: Likewise.
15602 * ipa-fnsummary.c: Likewise.
15603 * ipa-inline.c: Likewise.
15604 * ipa-visibility.c: Likewise.
15606 * lto-cgraph.c: Likewise.
15607 * omp-expand.c: Likewise.
15608 * omp-general.c: Likewise.
15609 * omp-low.c: Likewise.
15610 * omp-offload.c: Likewise.
15611 * omp-simd-clone.c: Likewise.
15612 * opts-global.c: Likewise.
15613 * passes.c: Likewise.
15614 * predict.c: Likewise.
15615 * sancov.c: Likewise.
15616 * sanopt.c: Likewise.
15617 * symtab.c: Likewise.
15618 * toplev.c: Likewise.
15619 * trans-mem.c: Likewise.
15620 * tree-chkp.c: Likewise.
15621 * tree-eh.c: Likewise.
15622 * tree-into-ssa.c: Likewise.
15623 * tree-object-size.c: Likewise.
15624 * tree-parloops.c: Likewise.
15625 * tree-profile.c: Likewise.
15626 * tree-ssa-ccp.c: Likewise.
15627 * tree-ssa-live.c: Likewise.
15628 * tree-ssa-loop.c: Likewise.
15629 * tree-ssa-sccvn.c: Likewise.
15630 * tree-ssa-structalias.c: Likewise.
15631 * tree-ssa.c: Likewise.
15632 * tree-streamer-in.c: Likewise.
15633 * tree-vectorizer.c: Likewise.
15634 * tree-vrp.c: Likewise.
15635 * tsan.c: Likewise.
15636 * ubsan.c: Likewise.
15637 * varasm.c: Likewise.
15638 * varpool.c: Likewise.
15639 * tree.c: Remove functions moved to attribs.[ch].
15640 * tree.h: Likewise.
15641 * config/aarch64/aarch64.c: Add attrs.h header file.
15642 * config/alpha/alpha.c: Likewise.
15643 * config/arc/arc.c: Likewise.
15644 * config/arm/arm.c: Likewise.
15645 * config/avr/avr.c: Likewise.
15646 * config/bfin/bfin.c: Likewise.
15647 * config/c6x/c6x.c: Likewise.
15648 * config/cr16/cr16.c: Likewise.
15649 * config/cris/cris.c: Likewise.
15650 * config/darwin.c: Likewise.
15651 * config/epiphany/epiphany.c: Likewise.
15652 * config/fr30/fr30.c: Likewise.
15653 * config/frv/frv.c: Likewise.
15654 * config/ft32/ft32.c: Likewise.
15655 * config/h8300/h8300.c: Likewise.
15656 * config/i386/winnt.c: Likewise.
15657 * config/ia64/ia64.c: Likewise.
15658 * config/iq2000/iq2000.c: Likewise.
15659 * config/lm32/lm32.c: Likewise.
15660 * config/m32c/m32c.c: Likewise.
15661 * config/m32r/m32r.c: Likewise.
15662 * config/m68k/m68k.c: Likewise.
15663 * config/mcore/mcore.c: Likewise.
15664 * config/microblaze/microblaze.c: Likewise.
15665 * config/mips/mips.c: Likewise.
15666 * config/mmix/mmix.c: Likewise.
15667 * config/mn10300/mn10300.c: Likewise.
15668 * config/moxie/moxie.c: Likewise.
15669 * config/msp430/msp430.c: Likewise.
15670 * config/nds32/nds32-isr.c: Likewise.
15671 * config/nds32/nds32.c: Likewise.
15672 * config/nios2/nios2.c: Likewise.
15673 * config/nvptx/nvptx.c: Likewise.
15674 * config/pa/pa.c: Likewise.
15675 * config/pdp11/pdp11.c: Likewise.
15676 * config/powerpcspe/powerpcspe.c: Likewise.
15677 * config/riscv/riscv.c: Likewise.
15678 * config/rl78/rl78.c: Likewise.
15679 * config/rx/rx.c: Likewise.
15680 * config/s390/s390.c: Likewise.
15681 * config/sh/sh.c: Likewise.
15682 * config/sol2.c: Likewise.
15683 * config/sparc/sparc.c: Likewise.
15684 * config/spu/spu.c: Likewise.
15685 * config/stormy16/stormy16.c: Likewise.
15686 * config/tilegx/tilegx.c: Likewise.
15687 * config/tilepro/tilepro.c: Likewise.
15688 * config/v850/v850.c: Likewise.
15689 * config/vax/vax.c: Likewise.
15690 * config/visium/visium.c: Likewise.
15691 * config/xtensa/xtensa.c: Likewise.
15693 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15696 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15697 constraints since the -mupper-regs-* switches have been
15699 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15700 into a vector from a double word element that was extracted from
15701 another vector, and eliminate extra XXPERMDI instructions.
15702 (vsx_concat_<mode>_2): Likewise.
15703 (vsx_concat_<mode>_3): Likewise.
15704 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15705 concat to allow optimizing inserts from previous extracts.
15707 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15709 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15710 memory reference to a SSP offset in TLS address space.
15711 (ix86_print_operand) <case '@'>: Remove.
15712 (ix86_print_operand_punct_valid_p): Remove '@' code.
15713 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15714 UNSPEC_SP_TLS_TEST.
15715 (stack_tls_protect_set_<mode>): Remove.
15716 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15717 (stack_tls_protect_test_<mode>): Remove.
15718 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15720 2017-08-07 Olivier Hainque <hainque@adacore.com>
15723 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15725 2017-08-07 Douglas Rupp <rupp@adacore.com>
15727 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15728 variable was referenced as multidir in command.
15730 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15733 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15736 2017-08-07 Martin Liska <mliska@suse.cz>
15738 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15739 * config/rl78/rl78.c: Add include of attribs.h.
15740 * config/sh/sh.c: Likewise.
15741 * config/v850/v850.c: Likewise.
15743 2017-08-07 Tom de Vries <tom@codesourcery.com>
15745 PR middle-end/78266
15746 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15748 2017-08-07 Martin Liska <mliska@suse.cz>
15750 * config/mips/mips.c: Include attribs.h.
15752 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15755 * doc/invoke.texi: Document change in behvaior for -Ofast for
15758 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15760 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15762 (aarch64_pop_regs): Likewise.
15763 (aarch64_gen_load_pair): Likewise.
15764 (aarch64_save_callee_saves): Likewise.
15765 (aarch64_restore_callee_saves): Likewise.
15767 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15769 * config/i386/i386.c: Revert the last change.
15771 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15774 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15776 (ix86_finalize_stack_frame_flags): This. Also clear
15777 frame_pointer_needed if -fno-omit-frame-pointer is used without
15779 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15780 with ix86_finalize_stack_frame_flags.
15781 (ix86_expand_epilogue): Likewise.
15782 (ix86_expand_split_stack_prologue): Likewise.
15784 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15787 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15788 to P_AES for Westmere.
15790 2017-08-07 Jonathan Yong <10walls@gmail.com>
15792 * config/i386/mingw.opt (fset-stack-executable): Removed.
15793 * config/i386/cygming.opt (fset-stack-executable): Moved
15795 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15797 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15799 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15801 2017-08-07 Marek Polacek <polacek@redhat.com>
15803 PR middle-end/81737
15804 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15806 2017-08-07 Martin Liska <mliska@suse.cz>
15808 * attribs.h (canonicalize_attr_name): New function.
15809 (cmp_attribs): Move from c-format.c and adjusted.
15810 (is_attribute_p): Moved from tree.h.
15811 * tree-inline.c: Add new includes.
15812 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15813 (private_is_attribute_p): Remove.
15814 (private_lookup_attribute): Likewise.
15815 (private_lookup_attribute_by_prefix): Simplify.
15816 (remove_attribute): Use is_attribute_p.
15817 * tree.h: Remove removed declarations.
15819 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15821 PR middle-end/81698
15822 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15823 instead of computing it in the function. Formatting fix.
15824 (expand_case): Don't rely on default_edge being the first edge,
15825 clear it if removing it, pass default_edge to
15826 emit_case_dispatch_table.
15827 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15830 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15832 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15833 insn in the function, emit NOP after the insn.
15835 2017-08-06 Tom de Vries <tom@codesourcery.com>
15837 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15840 2017-08-06 Tom de Vries <tom@codesourcery.com>
15842 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15845 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15847 PR tree-optimization/57371
15848 * match.pd: New pattern.
15850 2017-08-04 Marek Polacek <polacek@redhat.com>
15852 PR middle-end/81695
15853 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15854 perform the computation in offset_int.
15856 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15858 PR tree-optimization/81136
15859 * tree-vectorizer.h: Include tree-hash-traits.h.
15860 (vec_base_alignments): New typedef.
15861 (vec_info): Add a base_alignments field.
15862 (vect_record_base_alignments): Declare.
15863 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15865 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15866 (create_data_ref): Add an is_conditional_in_stmt argument.
15867 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15868 the is_conditional_in_stmt field.
15869 (data_ref_loc): Add an is_conditional_in_stmt field.
15870 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15871 (find_data_references_in_stmt): Update call to create_data_ref.
15872 (graphite_find_data_references_in_stmt): Likewise.
15873 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15874 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15875 (vect_record_base_alignment): New function.
15876 (vect_record_base_alignments): Likewise.
15877 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15878 for nested statements even if we fail to compute a misalignment.
15879 Use pooled base alignments for unconditional references.
15880 (vect_find_same_alignment_drs): Compare base addresses instead
15882 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15883 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15885 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15887 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15888 Add an explicit name for the enum. Use auto_vec for slp_instances
15889 and grouped_stores.
15890 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15892 (_bb_vec_info): Add a constructor and destructor.
15893 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15894 (destroy_loop_vec_info): Delete.
15895 (vect_destroy_datarefs): Likewise.
15896 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15897 (vec_info::vec_info): New function.
15898 (vec_info::~vec_info): Likewise.
15899 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15900 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15901 destroy_loop_vec_info.
15902 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15903 (_loop_vec_info::_loop_vec_info): ...this.
15904 (destroy_loop_vec_info): Replace with...
15905 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15906 the stmt_vec_infos. Leave handling of vec_info information to its
15907 destructor. Remove explicit vector releases.
15908 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15909 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15910 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15911 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15912 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15913 (destroy_bb_vec_info): Replace with...
15914 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15915 information to its destructor.
15916 (vect_slp_analyze_bb_1): Use new and delete instead of
15917 new_bb_vec_info and destroy_bb_vec_info.
15918 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15921 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15923 * tree-data-ref.h (subscript): Add access_fn field.
15924 (data_dependence_relation): Add could_be_independent_p.
15925 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15926 (same_access_functions): Move to tree-data-ref.c.
15927 * tree-data-ref.c (ref_contains_union_access_p): New function.
15928 (access_fn_component_p): Likewise.
15929 (access_fn_components_comparable_p): Likewise.
15930 (dr_analyze_indices): Add a reference to access_fn_component_p.
15931 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15933 (constant_access_functions): Likewise.
15934 (add_other_self_distances): Likewise.
15935 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15936 (initialize_data_dependence_relation): Use XCNEW and remove
15937 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15938 of access functions that have the same type. Allow the
15939 subsequence to end with different bases in some circumstances.
15940 Record the chosen access functions in SUB_ACCESS_FN.
15941 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15942 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15943 (subscript_dependence_tester_1): Likewise dra and drb.
15944 (build_classic_dist_vector): Update calls accordingly.
15945 (subscript_dependence_tester): Likewise.
15946 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15947 DDR_COULD_BE_INDEPENDENT_P.
15948 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15949 comp_alias_ddrs instead of may_alias_ddrs.
15950 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15952 (vect_analyze_data_ref_dependence): Use it if
15953 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15954 distance vectors if that fails.
15955 (dependence_distance_ge_vf): New function.
15956 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15957 LOOP_VINFO_MAY_ALIAS_DDRS.
15959 2017-08-04 Richard Biener <rguenther@suse.de>
15961 PR middle-end/81705
15962 * fold-const.c (fold_binary_loc): Properly restrict
15963 minus_var0 && minus_var1 case when associating undefined overflow
15966 2017-08-04 Tom de Vries <tom@codesourcery.com>
15968 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15970 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15972 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15973 Don't start diagnostic messages with a capital letter.
15974 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15976 (rs6000_invalid_builtin): Likewise.
15977 (rs6000_trampoline_init): Likewise.
15979 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15982 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15983 after setting changeable df flags.
15985 2017-08-03 Richard Biener <rguenther@suse.de>
15987 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15988 up if the use is in USE - X.
15990 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15992 * toplev.c (dumpfile.h): New include.
15993 (internal_error_reentered): New static function. Use it...
15994 (internal_error_function): ...here to handle reentered internal_error.
15996 2017-08-03 Richard Biener <rguenther@suse.de>
15998 PR middle-end/81148
15999 * fold-const.c (split_tree): Add minus_var and minus_con
16000 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16001 here but always use minus_*.
16002 (associate_trees): Assert we never associate with MINUS_EXPR
16003 and NULL first operand. Do not recurse for PLUS_EXPR operands
16004 when associating as MINUS_EXPR either.
16005 (fold_binary_loc): Track minus_var and minus_con.
16007 2017-08-03 Tom de Vries <tom@codesourcery.com>
16010 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16011 ACCEL_COMPILER, apply finish_options on
16012 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16014 2017-08-03 Tom de Vries <tom@codesourcery.com>
16017 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16018 function_entry_patch_area_size > 0.
16020 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16023 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16024 instead of 10??LU, perform unit multiplication in wide_int,
16025 don't change alloc_object_size_limit if the limit is larger
16028 PR tree-optimization/81655
16029 PR tree-optimization/81588
16030 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16031 the case when ranges[i].low and high are 1 for unsigned type with
16034 PR middle-end/81052
16035 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16036 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16038 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16040 * tree-vrp.h: Add include guard.
16042 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16045 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16046 (ud2): New insn pattern.
16047 * config/i386/i386.c (ix86_expand_epilogue):
16048 For naked functions, generate ud2 instead of trap insn.
16050 2017-08-02 Marek Polacek <polacek@redhat.com>
16053 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16055 2017-08-02 Tom de Vries <tom@codesourcery.com>
16056 Cesar Philippidis <cesar@codesourcery.com>
16058 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16059 Add missing edge probabilities.
16061 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16063 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16064 Correct endianness.
16066 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16068 PR middle-end/79499
16069 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16070 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16071 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16072 prologue_seq sequences - if any.
16074 2017-08-02 Richard Biener <rguenther@suse.de>
16076 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16077 via vectors if supported, integer extracts via punning if supported
16078 or otherwise vector extracts.
16080 2017-08-02 Richard Biener <rguenther@suse.de>
16082 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16084 (bitmap_insert_into_set): ... this.
16086 2017-08-02 Richard Biener <rguenther@suse.de>
16088 PR tree-optimization/81633
16090 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16092 PR tree-optimization/71752
16093 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16095 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16097 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16098 (machine_function::call_ms2sysv_pad_out): Remove field.
16099 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16100 (ix86_compute_frame_layout): Likewise.
16102 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16105 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16106 attribute with interrupt attribute.
16108 2017-08-01 Andrew Pinski <apinski@cavium.com>
16110 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16111 BIT_INSERT_EXPR's operand 1
16112 to see if the types precision matches.
16114 2017-08-01 Martin Liska <mliska@suse.cz>
16116 PR middle-end/70140
16117 * builtins.c (expand_builtin_memcpy_args): Remove.
16118 (expand_builtin_memcpy): Call newly added function
16119 expand_builtin_memory_copy_args.
16120 (expand_builtin_memcpy_with_bounds): Likewise.
16121 (expand_builtin_mempcpy): Remove last argument.
16122 (expand_builtin_mempcpy_with_bounds): Likewise.
16123 (expand_builtin_memory_copy_args): New function created from
16124 expand_builtin_mempcpy_args with small modifications.
16125 (expand_builtin_mempcpy_args): Remove.
16126 (expand_builtin_stpcpy): Remove unused argument.
16127 (expand_builtin): Likewise.
16128 (expand_builtin_with_bounds): Likewise.
16130 2017-08-01 Martin Liska <mliska@suse.cz>
16133 Make mempcpy more optimal (PR middle-end/70140).
16135 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16139 __builtin_vec_cmpne verify both arguments are compatible vectors
16140 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16141 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16142 move computation of aligned to after checking the argument types.
16146 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16147 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16150 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16151 Doug Rupp <rupp@adacore.com>
16152 Olivier Hainque <hainque@adacore.com>
16154 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16155 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16157 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16158 for TARGET_OS_CPP_BUILTIN.
16159 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16160 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16162 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16163 passing required abi options to the assembler for EABI configurations.
16164 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16165 of .text.hot and .text.unlikely sections for kernel modules when
16166 using ARM style exceptions.
16167 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16168 options. Add EXTRA_CC1_SPEC.
16169 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16171 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16173 (ARM_TARGET2_DWARF_FORMAT): Define.
16174 * config/arm/t-vxworks: Adjust multilib control to removal of the
16175 Diab command line options.
16177 2017-08-01 Martin Liska <mliska@suse.cz>
16179 PR gcov-profile/81561
16180 * gcov.c (unblock): Make unblocking safe as we need to preserve
16181 index correspondence of blocks and block_lists.
16183 2017-08-01 Richard Biener <rguenther@suse.de>
16185 PR tree-optimization/81181
16186 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16187 (compute_antic): ... end of iteration here.
16189 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16191 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16192 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16193 (ftree-slp-vectorize): Likewise.
16194 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16195 can no longer be set independent of flag_tree_loop_vectorize.
16196 * omp-general.c (emp_max_vf): Likewise.
16197 * opts.c (enable_fdo_optimizations): Remove references to
16198 flag_tree_vectorize, these are now implicit.
16199 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16200 and leave it for the options machinery.
16202 2017-08-01 Martin Liska <mliska@suse.cz>
16204 PR middle-end/70140
16205 * builtins.c (expand_builtin_memcpy_args): Remove.
16206 (expand_builtin_memcpy): Call newly added function
16207 expand_builtin_memory_copy_args.
16208 (expand_builtin_memcpy_with_bounds): Likewise.
16209 (expand_builtin_mempcpy): Remove last argument.
16210 (expand_builtin_mempcpy_with_bounds): Likewise.
16211 (expand_builtin_memory_copy_args): New function created from
16212 expand_builtin_mempcpy_args with small modifications.
16213 (expand_builtin_mempcpy_args): Remove.
16214 (expand_builtin_stpcpy): Remove unused argument.
16215 (expand_builtin): Likewise.
16216 (expand_builtin_with_bounds): Likewise.
16218 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16221 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16222 print "ds:" only for immediates in generic address space.
16224 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16227 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16228 (ix86_function_ok_for_sibcall): Return false for naked functions.
16230 2017-08-01 Richard Biener <rguenther@suse.de>
16232 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16233 (compute_antic): Seed worklist with exit block predecessors.
16234 * cfganal.c (dfs_find_deadend): For a cycle return the source
16235 of the edge closing it.
16237 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16239 * config/aarch64/aarch64.c
16240 (aarch64_can_const_movi_rtx_p): Move 0 check.
16242 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16244 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16245 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16247 * match.pd: Ditto in address comparison pattern.
16249 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16251 PR tree-optimization/81627
16252 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16253 closed ssa form for store-store chain.
16255 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16257 PR tree-optimization/81620
16258 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16259 for store-store chain.
16261 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16263 PR tree-optimization/81588
16264 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16265 ranges[i].in_p, invert comparison code ccode. For >/>=,
16266 swap rhs1 and rhs2 and comparison code unconditionally,
16267 for </<= don't do that. Don't swap rhs1/rhs2 again if
16268 ranges[i].in_p, instead invert comparison code ccode if
16269 opcode or oe->rank is BIT_IOR_EXPR.
16272 * optabs.def (vec_extract_optab, vec_init_optab): Change from
16273 a direct optab to conversion optab.
16274 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
16275 with GET_MODE_INNER as last argument instead of optab_handler.
16276 * expmed.c (extract_bit_field_1): Likewise. Use vector from
16277 vector extraction if possible and optab is available.
16278 * expr.c (store_constructor): Use convert_optab_handler instead
16279 of optab_handler. Use vector initialization from smaller
16280 vectors if possible and optab is available.
16281 * tree-vect-stmts.c (vectorizable_load): Likewise.
16282 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16283 now have two modes.
16284 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
16285 of vec_init from half-sized vectors with the same element mode.
16286 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
16287 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
16288 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
16289 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
16290 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
16291 after mode in gen_vec_extract* calls.
16292 (vec_extract<mode>): Renamed to ...
16293 (vec_extract<mode><ssescalarmodelower>): ... this.
16294 (vec_extract<mode><ssehalfvecmodelower>): New expander.
16295 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
16296 element mode after mode in gen_vec_init* calls.
16297 (VEC_INIT_HALF_MODE): New mode iterator.
16298 (vec_init<mode>): Renamed to ...
16299 (vec_init<mode><ssescalarmodelower>): ... this.
16300 (vec_init<mode><ssehalfvecmodelower>): New expander.
16301 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
16302 (vec_extractv2sfsf): ... this.
16303 (vec_initv2sf): Renamed to ...
16304 (vec_initv2sfsf): ... this.
16305 (vec_extractv2si): Renamed to ...
16306 (vec_extractv2sisi): ... this.
16307 (vec_initv2si): Renamed to ...
16308 (vec_initv2sisi): ... this.
16309 (vec_extractv4hi): Renamed to ...
16310 (vec_extractv4hihi): ... this.
16311 (vec_initv4hi): Renamed to ...
16312 (vec_initv4hihi): ... this.
16313 (vec_extractv8qi): Renamed to ...
16314 (vec_extractv8qiqi): ... this.
16315 (vec_initv8qi): Renamed to ...
16316 (vec_initv8qiqi): ... this.
16317 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16318 (vec_init<mode>): Renamed to ...
16319 (vec_init<mode><VEC_base_l>): ... this.
16320 (vec_extract<mode>): Renamed to ...
16321 (vec_extract<mode><VEC_base_l>): ... this.
16322 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16323 (vec_initv2sfsf): ... this.
16324 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16325 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16326 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16327 element mode after mode in gen_vec_init* calls.
16328 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16329 (vec_init<mode><Vel>): ... this.
16330 (vec_extract<mode>): Renamed to ...
16331 (vec_extract<mode><Vel>): ... this.
16332 * config/aarch64/iterators.md (Vel): New mode attribute.
16333 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16334 Add element mode after mode in gen_vec_extract* calls.
16335 * config/s390/vector.md (non_vec_l): New mode attribute.
16336 (vec_extract<mode>): Renamed to ...
16337 (vec_extract<mode><non_vec_l>): ... this.
16338 (vec_init<mode>): Renamed to ...
16339 (vec_init<mode><non_vec_l>): ... this.
16340 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16341 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16343 * config/arm/iterators.md (V_elem_l): New mode attribute.
16344 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16345 (vec_extract<mode><V_elem_l>): ... this.
16346 (vec_extractv2di): Renamed to ...
16347 (vec_extractv2didi): ... this.
16348 (vec_init<mode>): Renamed to ...
16349 (vec_init<mode><V_elem_l>): ... this.
16350 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16351 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16352 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16353 Add element mode after gen_vec_extract* calls.
16354 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16355 (vec_init<mode><unitmode>): ... this.
16356 (vec_extract<mode>): Renamed to ...
16357 (vec_extract<mode><unitmode>): ... this.
16358 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16359 (vec_init<mode><unitmode>): ... this.
16360 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16361 (vec_initv2sfsf): ... this.
16362 (vec_extractv2sf): Renamed to ...
16363 (vec_extractv2sfsf): ... this.
16364 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16365 Add element mode after gen_vec_extract* calls.
16366 * config/mips/mips.md (unitmode): New mode iterator.
16367 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16368 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16369 * config/spu/spu.md (inner_l): New mode attribute.
16370 (vec_init<mode>): Renamed to ...
16371 (vec_init<mode><inner_l>): ... this.
16372 (vec_extract<mode>): Renamed to ...
16373 (vec_extract<mode><inner_l>): ... this.
16374 * config/sparc/sparc.md (veltmode): New mode iterator.
16375 (vec_init<VMALL:mode>): Renamed to ...
16376 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16377 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16378 (vec_initv2sisi): ... this.
16379 (vec_initv2sf): Renamed to ...
16380 (vec_initv2sfsf): ... this.
16381 (vec_extractv2sf): Renamed to ...
16382 (vec_extractv2sfsf): ... this.
16383 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16384 (vec_init<mode>): Renamed to ...
16385 (vec_init<mode><VEC_base_l>): ... this.
16386 (vec_extract<mode>): Renamed to ...
16387 (vec_extract<mode><VEC_base_l>): ... this.
16388 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16389 (vec_initv2sfsf): ... this.
16390 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16391 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16392 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16393 gen_vec_init* calls.
16395 2017-08-01 Richard Biener <rguenther@suse.de>
16397 PR tree-optimization/81297
16398 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16399 TREE_OVERFLOW from INTEGER_CSTs.
16401 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16403 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16405 2017-07-31 Carl Love <cel@us.ibm.com>
16407 * config/rs6000/rs6000-c: Add support for built-in functions
16408 vector signed char vec_xl_be (signed long long, signed char *);
16409 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16410 vector signed int vec_xl_be (signed long long, signed int *);
16411 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16412 vector signed long long vec_xl_be (signed long long, signed long long *);
16413 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16414 vector signed short vec_xl_be (signed long long, signed short *);
16415 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16416 vector double vec_xl_be (signed long long, double *);
16417 vector float vec_xl_be (signed long long, float *);
16418 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16419 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16420 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16422 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16423 (altivec_expand_builtin): Add switch statement to call
16424 altivec_expand_xl_be for each builtin.
16425 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16426 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16427 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16428 __builtin_vsx_le_be_v16qi.
16429 * doc/extend.texi: Update the built-in documentation file for the
16430 new built-in functions.
16432 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16435 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16437 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16439 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16441 * config.gcc: Add z14.
16442 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16443 CPU model numbers for z13s and z14.
16444 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16446 * config/s390/s390-opts.h (enum processor_type): Rename
16447 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16448 * config/s390/s390.c (processor_table): Add field for CPU name to
16449 be passed to Binutils.
16450 (s390_asm_output_machine_for_arch): Use the new field in
16451 processor_table for Binutils.
16452 (s390_expand_builtin): Replace arch12 with z14.
16453 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16454 (s390_get_sched_attrmask): Likewise.
16455 (s390_get_unit_mask): Likewise.
16456 * config/s390/s390.opt: Add z14 to processor_type enum.
16458 2017-07-31 Martin Jambor <mjambor@suse.cz>
16461 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16462 regardless of optimization level.
16464 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16465 Martin Liska <mliska@suse.cz>
16467 * predict.def: Remove old comment and adjust probability.
16468 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16469 PREDICT statements.
16471 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16474 * config/i386/i386.c (ix86_function_naked): New function.
16475 (ix86_can_use_return_insn_p): Return false for naked functions.
16476 (ix86_expand_prologue): Skip prologue for naked functions.
16477 (ix86_expand_epilogue): Skip epilogue for naked functions
16478 and emit trap instruction.
16479 (ix86_warn_func_return): New function.
16480 (ix86_attribute_table): Add "naked" attribute specification.
16481 (TARGET_WARN_FUNC_RETURN): Define.
16482 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16484 2017-07-31 Martin Liska <mliska@suse.cz>
16486 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16487 (dump_gimple_bb_header): Always dump BB info.
16488 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16490 2017-07-31 Martin Liska <mliska@suse.cz>
16493 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16494 also with current_function_decl non-null equality.
16496 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16499 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16500 change type to the element type, instead add eltype variable and
16501 use it where we are interested in the element type.
16503 PR tree-optimization/81603
16504 * ipa-polymorphic-call.c
16505 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16506 offset arithmetic in offset_int, bail out if the resulting bit offset
16507 doesn't fit into shwi.
16509 2017-07-31 Martin Liska <mliska@suse.cz>
16511 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16512 (gimplify_save_expr): Fix comment.
16514 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16517 * config/i386/i386.c (ix86_function_arg): Update arguments for
16519 (ix86_compute_frame_layout): Set the initial stack offset to
16520 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16521 INCOMING_FRAME_SP_OFFSET.
16522 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16523 stack before exception handler returns.
16524 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16525 the 'ERROR_CODE' for exception handler.
16527 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16529 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16530 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16531 (ASM_OUTPUT_REG_POP): Ditto.
16532 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16533 instead of asm_fprintf to output pure string.
16535 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16537 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16538 to imported_module_or_decl hook.
16539 (debug_nothing_tree_tree_tree_bool): Remove.
16540 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16541 * debug.c (do_nothing_debug_hooks): Use
16542 debug_nothing_tree_tree_tree_bool_bool instead of
16543 debug_nothing_tree_tree_tree_bool.
16544 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16545 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16546 * sdbout.c (sdb_debug_hooks): Likewise.
16547 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16548 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16550 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16551 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16552 attribute, don't add anything.
16554 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16556 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16557 (fold_build2_stat_loc): Likewise.
16558 (fold_build3_stat_loc): Likewise.
16559 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16560 (fold_build1_loc): Remove macro.
16561 (fold_build2_loc): Likewise.
16562 (fold_build3_loc): Likewise.
16564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16566 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16567 (gimple_build_debug_bind_source_stat): Likewise.
16568 * gimple.h (gimple_build_debug_bind): Remove macro.
16569 (gimple_build_debug_bind_source): Likewise.
16571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16573 * bitmap.c (bitmap_alloc): Adjust.
16574 (bitmap_gc_alloc): Likewise.
16575 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16579 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16580 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16581 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16582 (bitmap_gc_alloc_stat): Likewise.
16583 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16587 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16588 * rtl.h (shallow_copy_rtx): Remove macro.
16590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16592 * emit-rtl.c (gen_raw_REG): Adjust.
16593 * gengenrtl.c (gendef): Likewise.
16594 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16595 * rtl.h (rtx_alloc): Remove macro.
16597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16599 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16600 (build_tree_list_stat): Likewise.
16601 * tree.h (build_tree_list): Remove macro.
16602 (build_tree_list_vec): Likewise.
16604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16606 * tree.c (make_vector_stat): Remove _stat from name.
16607 (build_vector_stat): Likewise.
16608 * tree.h (make_vector_stat): Remove macro.
16609 (build_vector_stat): Likewise.
16611 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16613 * tree.h (build_var_debug_value): Remove prototype.
16615 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16617 * tree.c (tree_cons_stat): Remove _stat from name.
16618 * tree.h (tree_cons): Remove macro.
16620 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16622 * tree.c (build_vl_exp_stat): Remove _stat from name.
16623 * tree.h (build_vl_exp): Remove macro.
16625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16627 * tree.c (build_decl_stat): Remove _stat from name.
16628 * tree.h (build_decl): Remove macro.
16630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16632 * gimple.c (gimple_build_with_ops_stat): Adjust.
16633 (gimple_alloc_stat): Remove _stat from name.
16634 * gimple.h (gimple_alloc): Remove macro.
16636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16638 * tree.c (make_tree_vec_stat): Remove _stat from name.
16639 (grow_tree_vec_stat): Likewise.
16640 * tree.h (make_tree_vec_stat): Adjust prototype.
16641 (grow_tree_vec_stat): Likewise.
16642 (make_tree_vec): Remove macro.
16643 (grow_tree_vec): Likewise.
16645 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16647 * fold-const.c (fold_build1_stat_loc): Adjust.
16648 (fold_build2_stat_loc): Likewise.
16649 (fold_build3_stat_loc): Likewise.
16650 * tree.c (build0_stat): Remove _stat from name.
16651 (build1_stat): Likewise.
16652 (build2_stat): Likewise.
16653 (build3_stat): Likewise.
16654 (build4_stat): Likewise.
16655 (build5_stat): Likewise.
16656 * tree.h (build1_loc): Remove macro, and rename _stat function
16658 (build2_loc): Likewise.
16659 (build3_loc): Likewise.
16660 (build4_loc): Likewise.
16661 (build5_loc): Likewise.
16663 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16665 * tree.c (make_int_cst_stat): Remove _stat from name.
16666 * tree.h (make_int_cst_stat): Adjust prototype.
16667 (make_int_cst): Remove macro.
16669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16671 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16672 * tree.h (make_tree_binfo_stat): Adjust prototype.
16673 (make_tree_binfo): Remove.
16675 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16677 * tree.c (copy_node_stat): Rename to copy_node.
16678 (build_distinct_type_copy): Adjust.
16679 * tree.h (copy_node_stat): Adjust prototype.
16680 (copy_node): Remove macro.
16682 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16684 * tree.c (make_node_stat): rename to make_node.
16685 (build_tree_list_stat): Adjust.
16686 (build0_stat): Likewise.
16687 (build2_stat): Likewise.
16688 (build3_stat): Likewise.
16689 (build4_stat): Likewise.
16690 (build5_stat): Likewise.
16691 (build_decl_stat): Likewise.
16692 * tree.h (make_node_stat): Adjust prototype.
16693 (make_node): remove macro.
16695 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16697 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16698 (PPC_FEATURE2_SCV): Likewise.
16699 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16701 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16703 * config/aarch64/aarch64.c
16704 (aarch64_internal_mov_immediate): Add new special pattern.
16705 * config/aarch64/aarch64.md (*movdi_aarch64):
16706 Add reg/32bit const mov case.
16708 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16709 Richard Sandiford <richard.sandiford@linaro.org>
16711 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16712 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16713 Add integer and movi cases.
16714 (movi-split-hf-df-sf split, fp16): New.
16715 (enabled): Added TARGET_FP_F16INST.
16716 * config/aarch64/iterators.md (GPF_HF): New.
16717 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16719 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16721 * config/aarch64/aarch64.c
16722 (aarch64_simd_container_mode): Add prototype.
16723 (aarch64_expand_mov_immediate): Add HI support.
16724 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16725 (aarch64_can_const_movi_rtx_p): New.
16726 (aarch64_preferred_reload_class):
16727 Remove restrictions of using FP registers for certain SIMD operations.
16728 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16729 (aarch64_valid_floating_const): Add integer move validation.
16730 (aarch64_simd_imm_scalar_p): Remove.
16731 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16732 (aarch64_legitimate_constant_p): Expand list of supported cases.
16733 * config/aarch64/aarch64-protos.h
16734 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16735 (aarch64_reinterpret_float_as_int): New.
16736 (aarch64_simd_imm_scalar_p): Remove.
16737 * config/aarch64/constraints.md (Uvi): New.
16738 (Dd): Split into Ds and new Dd.
16739 * config/aarch64/aarch64.md (*movsi_aarch64):
16741 (*movdi_aarch64): Add SIMD mov case.
16743 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16745 * tree-predcom.c: (struct chain): Handle store-store chain in which
16746 stores for elimination only store loop invariant values.
16747 (execute_pred_commoning_chain): Ditto.
16748 (prepare_initializers_chain_store_elim): Ditto.
16749 (prepare_finalizers): Ditto.
16750 (is_inv_store_elimination_chain): New function.
16751 (initialize_root_vars_store_elim_1): New function.
16753 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16755 * tree-predcom.c: Revise general description of the pass.
16756 (enum chain_type): New enum type for store elimination.
16757 (struct chain): New field supporting store elimination.
16758 (struct component): Ditto.
16759 (dump_chain): Dump store-stores chain.
16760 (release_chain): Release resources.
16761 (split_data_refs_to_components): Compute and create component
16762 contains only stores for elimination.
16763 (get_chain_last_ref_at): New function.
16764 (make_invariant_chain): Initialization.
16765 (make_rooted_chain): Specify chain type in parameter and record it.
16766 (add_looparound_copies): Skip for store-stores chain.
16767 (determine_roots_comp): Compute type of chain and pass it to
16769 (initialize_root_vars_store_elim_2): New function.
16770 (finalize_eliminated_stores): New function.
16771 (remove_stmt): Handle store for elimination.
16772 (execute_pred_commoning_chain): Execute predictive commoning on
16773 store-store chains.
16774 (determine_unroll_factor): Skip unroll for store-stores chain.
16775 (prepare_initializers_chain_store_elim): New function.
16776 (prepare_initializers_chain): Hanlde store-store chain.
16777 (prepare_finalizers_chain, prepare_finalizers): New function.
16778 (tree_predictive_commoning_loop): Return integer value indicating
16779 if loop is unrolled or lcssa form is corrupted.
16780 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16782 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16784 * tree-predcom.c (initialize_root): Delete.
16785 (execute_pred_commoning_chain): Initialize root vars and replace
16786 reference of non-combined chain directly, rather than call above
16789 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16791 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16792 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16794 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16796 * tree-predcom.c (struct chain): New field init_seq.
16797 (release_chain): Release init_seq.
16798 (prepare_initializers_chain): Record intialization stmts in above
16800 (insert_init_seqs): New function.
16801 (tree_predictive_commoning_loop): Call insert_init_seqs.
16803 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16805 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16807 2017-07-28 Richard Biener <rguenther@suse.de>
16809 * match.pd: Remove superfluous :c.
16810 * genmatch.c (simplify::id): Add member.
16811 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16813 (current_id): New global.
16814 (dt_node::parent): Move from ...
16815 (dt_operand::parent): ... here. Add for_id member.
16816 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16817 (decision_tree::find_node): Relax order requirement when
16818 merging DT_TRUE nodes to ones inbetween the current simplify
16819 and the one we try to merge with. Add diagnostic whenever
16820 we need to enforce pattern order by not merging.
16821 (decision_tree::insert): Set current_id.
16822 (decision_tree::print_node): Dump parent node and for_id.
16823 (parser::last_id): Add member.
16824 (parser::push_simplify): Assign unique id.
16825 (parser::parser): Initialize last_id.
16827 2017-07-28 Martin Liska <mliska@suse.cz>
16830 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16831 gimple_build_debug_bind.
16833 2017-07-28 Richard Biener <rguenther@suse.de>
16835 PR tree-optimization/81502
16836 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16838 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16840 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16841 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16842 for BIT_FIELD_REF args.
16843 * fold-const.c (make_bit_field_ref): Likewise.
16844 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16846 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16849 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16850 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16851 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16852 Or it into SANITIZER_UNDEFINED.
16853 * ubsan.c: Include gimple-fold.h and varasm.h.
16854 (ubsan_expand_ptr_ifn): New function.
16855 (instrument_pointer_overflow): New function.
16856 (maybe_instrument_pointer_overflow): New function.
16857 (instrument_object_size): Formatting fix.
16858 (pass_ubsan::execute): Call instrument_pointer_overflow
16859 and maybe_instrument_pointer_overflow.
16860 * internal-fn.c (expand_UBSAN_PTR): New function.
16861 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16862 * sanitizer.def (__ubsan_handle_pointer_overflow,
16863 __ubsan_handle_pointer_overflow_abort): New builtins.
16864 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16865 * internal-fn.def (UBSAN_PTR): New internal function.
16866 * opts.c (sanitizer_opts): Add pointer-overflow.
16867 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16868 * fold-const.c (build_range_check): Compute pointer range check in
16869 integral type if pointer arithmetics would be needed. Formatting
16872 2017-07-28 Martin Liska <mliska@suse.cz>
16875 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16876 parameters that are of a variable-length.
16878 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16880 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16882 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16883 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16884 (CRT_CALL_STATIC_FUNCTION): Likewise.
16885 (ASM_DEFAULT_SPEC): New define.
16886 (ASM_SPEC32): Likewise.
16887 (ASM_SPEC64): Likewise.
16888 (ASM_SPEC_COMMON): Likewise.
16889 (ASM_SPEC): Likewise.
16890 (INVALID_64BIT): Likewise.
16891 (LINK_OS_DEFAULT_SPEC): Likewise.
16892 (LINK_OS_SPEC32): Likewise.
16893 (LINK_OS_SPEC64): Likewise.
16894 (POWERPC_LINUX): Likewise.
16895 (PTRDIFF_TYPE): Likewise.
16896 (RESTORE_FP_PREFIX): Likewise.
16897 (RESTORE_FP_SUFFIX): Likewise.
16898 (SAVE_FP_PREFIX): Likewise.
16899 (SAVE_FP_SUFFIX): Likewise.
16900 (SIZE_TYPE): Likewise.
16901 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16902 (TARGET_64BIT): Likewise.
16903 (TARGET_64BIT): Likewise.
16904 (TARGET_AIX): Likewise.
16905 (WCHAR_TYPE_SIZE): Likewise.
16906 (WCHAR_TYPE): Undefine.
16907 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16908 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16909 (CPP_OS_RTEMS_SPEC): Delete.
16910 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16911 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16913 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16915 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16917 PR tree-optimization/81578
16918 * tree-parloops.c (build_new_reduction): Bail out if
16919 reduction_code isn't one of the standard OpenMP reductions.
16920 Move the details printing after that decision.
16922 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16924 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16925 related to reload_in_progress.
16926 (splat_input_operand): Likewise.
16927 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16929 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16931 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16932 (TARGET_INSTANTIATE_DECLS): Likewise.
16933 (legitimate_indexed_address_p): Delete reload_in_progress code.
16934 (rs6000_debug_legitimate_address_p): Likewise.
16935 (rs6000_eliminate_indexed_memrefs): Likewise.
16936 (rs6000_emit_le_vsx_store): Likewise.
16937 (rs6000_emit_move_si_sf_subreg): Likewise.
16938 (rs6000_emit_move): Likewise.
16939 (register_to_reg_type): Likewise.
16940 (rs6000_pre_atomic_barrier): Likewise.
16941 (rs6000_machopic_legitimize_pic_address): Likewise.
16942 (rs6000_allocate_stack_temp): Likewise.
16943 (rs6000_address_for_fpconvert): Likewise.
16944 (rs6000_address_for_altivec): Likewise.
16945 (rs6000_secondary_memory_needed_rtx): Delete function.
16946 (rs6000_check_sdmode): Likewise.
16947 (rs6000_alloc_sdmode_stack_slot): Likewise.
16948 (rs6000_instantiate_decls): Likewise.
16949 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16950 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16951 Delete reload_in_progress.
16952 (*vec_reload_and_plus_<mptrsize>): Likewise.
16953 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16954 (vsx_div_v2di): Likewise.
16955 (vsx_udiv_v2di): Likewise.
16957 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16959 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16960 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16961 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16962 (rs6000_debug_reg_global): Delete print of LRA status.
16963 (rs6000_option_override_internal): Delete dead LRA related code.
16964 (rs6000_lra_p): Delete function.
16965 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16967 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16969 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16970 * config/riscv/rtems.h: New file.
16972 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16973 Sudakshina Das <sudi.das@arm.com>
16975 * config/aarch64/aarch64.md
16976 (define_split for and<mode>3nr_compare): Move
16977 non aarch64_logical_operand to a register.
16978 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16979 register immediate operand to a register.
16980 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16982 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16984 PR middle-end/81564
16985 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16987 2017-07-27 Richard Biener <rguenther@suse.de>
16989 PR tree-optimization/81573
16990 PR tree-optimization/81494
16991 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16992 multi defuse cycle case.
16994 2017-07-27 Richard Biener <rguenther@suse.de>
16996 PR tree-optimization/81571
16997 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17000 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17002 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17003 earlier and only if MASK_FPU is set. Adjust formatting.
17005 2017-07-27 Martin Liska <mliska@suse.cz>
17007 * opt-functions.awk: Add validation of value of Init.
17008 * optc-gen.awk: Pass new argument.
17010 2017-07-27 Martin Liska <mliska@suse.cz>
17012 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17013 Fix wrong condition.
17015 2017-07-27 Martin Liska <mliska@suse.cz>
17017 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17018 BBs and edges seen by autoFDO.
17020 2017-07-27 Richard Biener <rguenther@suse.de>
17022 PR tree-optimization/81502
17023 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17024 with incompatible but same sized type.
17025 (execute_update_addresses_taken): Likewise.
17027 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17029 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17030 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17032 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17035 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17036 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17037 Change s_operand to memory_operand.
17039 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17041 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17042 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17043 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17044 Emit instructions rather than returning an expression. Handle TFmode
17045 and KFmode by casting to TImode.
17046 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17047 (rs6000_emit_le_vsx_store): Likewise.
17048 * config/rs6000/vsx.md (VSX_TI): New iterator.
17049 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17050 (*vsx_le_undo_permute_<mode>): Likewise.
17051 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17052 emit the split sequence.
17053 (*vsx_le_perm_store_<mode>): Likewise.
17055 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17057 PR tree-optimization/81555
17058 PR tree-optimization/81556
17059 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17060 if true, force CHANGED for the recursive invocation.
17061 (reassociate_bb): Remember original length of ops array, pass
17062 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17064 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17065 attributes for noipa attribute. For naked attribute use
17066 lookup_attribute first before lookup_attribute_spec.
17067 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17069 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17070 for functions with noipa attribute.
17071 (cgraph_externally_visible_p): Return true for functions with noipa
17073 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17074 for functions with noipa attribute.
17075 * doc/extend.texi: Document noipa function attribute.
17076 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17077 also for functions with noipa attribute.
17078 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17080 2017-07-26 Andrew Pinski <apinski@cavium.com>
17082 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17083 vec_unalign_load_cost and vec_unalign_store_cost.
17085 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17087 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17088 -mvsx-small-integer option.
17089 (ISA_3_0_MASKS_IEEE): Likewise.
17090 (OTHER_VSX_VECTOR_MASKS): Likewise.
17091 (POWERPC_MASKS): Likewise.
17092 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17093 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17094 code, only testing for DImode being allowed in non-VSX floating
17096 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17097 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17099 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17100 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17101 TARGET_P8_VECTOR test.
17102 (rs6000_secondary_reload_simple_move): Likewise.
17103 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17104 since TARGET_P9_VECTOR was already tested.
17105 (rs6000_opt_masks): Remove -mvsx-small-integer.
17106 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17107 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17109 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17110 test for TARGET_VEXTRACTUB was used, and that uses
17112 (p9 extract splitter): Likewise.
17113 (vsx_extract_<mode>_di_p9): Likewise.
17114 (vsx_extract_<mode>_store_p9): Likewise.
17115 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17116 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17117 the elimination of TARGET_VSX_SMALL_INTEGER.
17118 (vsx_extract_<mode>_p8): Likewise.
17119 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17120 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17121 (vsx_set_<mode>_p9): Likewise.
17122 (vsx_set_v4sf_p9): Likewise.
17123 (vsx_set_v4sf_p9_zero): Likewise.
17124 (vsx_insert_extract_v4sf_p9): Likewise.
17125 (vsx_insert_extract_v4sf_p9_2): Likewise.
17126 * config/rs6000/rs6000.md (sign extend splitter): Change
17127 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17128 (floatsi<mode>2_lfiwax_mem): Likewise.
17129 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17130 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17131 since a test for TARGET_P9_VECTOR was used.
17132 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17133 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17134 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17135 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17136 TARGET_P8_VECTOR test.
17137 (fix_trunc<mode>si2_stfiwx): Likewise.
17138 (fix_trunc<mode>si2_internal): Likewise.
17139 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17140 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17142 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17143 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17144 TARGET_P8_VECTOR test.
17145 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17146 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17147 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17149 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17150 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17151 since a test for TARGET_P9_VECTOR was used.
17152 (splitter for loading small constants): Likewise.
17154 2017-07-26 Andrew Pinski <apinski@cavium.com>
17156 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17159 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17162 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17163 (fp_valid_at): Likewise.
17165 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17167 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17168 (qdf24xx_addrcost_table): Likewise.
17169 (cortexa57_tunings): Update to use generic_branch_cost.
17170 (cortexa72_tunings): Likewise.
17171 (cortexa73_tunings): Likewise.
17172 (qdf24xx_tunings): Likewise.
17174 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17176 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17177 (thunderx2t99_branch_cost): Likewise.
17178 (cortexa35_tunings): Update to use generic_branch_cost.
17179 (cortexa53_tunings): Likewise.
17180 (cortexa57_tunings): Likewise.
17181 (cortexa72_tunings): Likewise.
17182 (cortexa73_tunings): Likewise.
17183 (thunderx2t99_tunings): Likewise.
17185 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17187 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17188 (sparc_option_override): Honour MASK_FSMULD.
17189 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17190 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17191 * config/sparc/sparc.opt (mfsmuld): New option.
17192 * doc/invoke.texi (mfsmuld): Document option.
17194 2017-07-26 Marek Polacek <polacek@redhat.com>
17196 PR middle-end/70992
17197 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17199 2017-07-26 Richard Biener <rguenther@suse.de>
17201 * gimple-match-head.c (do_valueize): Return OP if valueize
17203 (get_def): New helper to get at the def stmt of a SSA name
17204 if valueize allows.
17205 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17206 do_valueize to get at the def stmt.
17207 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17209 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17211 PR middle-end/46932
17212 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17214 2017-07-26 Martin Liska <mliska@suse.cz>
17217 * function.c (expand_function_start): Make expansion of
17218 nonlocal_goto_save_area after parm_birth_insn.
17220 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17222 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17223 from all CPU target flags enable members.
17225 2017-07-26 Richard Biener <rguenther@suse.de>
17227 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17228 (decision_tree::gen): Make 'type' const.
17229 (write_predicate): Likewise.
17231 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17233 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17234 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17235 (rs6000_option_override_internal): Likewise.
17236 (rs6000_expand_vector_set): Likewise.
17237 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17238 (TARGET_UPPER_REGS_SF): Likewise.
17239 (TARGET_UPPER_REGS_DI): Likewise.
17240 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17241 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17242 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17243 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17244 (Splitters for DI constants in Altivec registers): Likewise.
17245 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17246 (vsx_set_v4sf_p9): Likewise.
17247 (vsx_set_v4sf_p9_zero): Likewise.
17248 (vsx_insert_extract_v4sf_p9): Likewise.
17249 (vsx_insert_extract_v4sf_p9_2): Likewise.
17251 2017-07-25 Carl Love <cel@us.ibm.com>
17253 * doc/extend.texi: Update the built-in documentation file for the
17254 existing built-in functions
17255 vector signed char vec_cnttz (vector signed char);
17256 vector unsigned char vec_cnttz (vector unsigned char);
17257 vector signed short vec_cnttz (vector signed short);
17258 vector unsigned short vec_cnttz (vector unsigned short);
17259 vector signed int vec_cnttz (vector signed int);
17260 vector unsigned int vec_cnttz (vector unsigned int);
17261 vector signed long long vec_cnttz (vector signed long long);
17262 vector unsigned long long vec_cnttz (vector unsigned long long);
17264 2017-07-25 Andrew Pinski <apinski@cavium.com>
17266 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17267 accesses where the use is for the first operand of a BIT_INSERT.
17269 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17272 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17273 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17275 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17277 * config/i386/gstabs.h: Delete.
17278 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17280 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
17282 * config/i386/i386.c (ix86_decompose_address): Do not check for
17283 register RTX when looking at index_reg or base_reg.
17284 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
17286 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
17288 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
17289 to update EH info here.
17291 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17293 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
17295 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17297 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
17299 2017-07-25 Torsten Duwe <duwe@suse.de>
17301 * common.opt: Introduce -fpatchable-function-entry
17302 command line option, and its variables function_entry_patch_area_size
17303 and function_entry_patch_area_start.
17304 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17305 including a two-value parser.
17306 * target.def (print_patchable_function_entry): New target hook.
17307 * targhooks.h (default_print_patchable_function_entry): New function.
17308 * targhooks.c (default_print_patchable_function_entry): Likewise.
17309 * toplev.c (process_options): Switch off IPA-RA if
17310 patchable function entries are being generated.
17311 * varasm.c (assemble_start_function): Look at the
17312 patchable-function-entry command line switch and current
17313 function attributes and maybe generate NOP instructions by
17314 calling the print_patchable_function_entry hook.
17315 * doc/extend.texi: Document patchable_function_entry attribute.
17316 * doc/invoke.texi: Document -fpatchable_function_entry
17317 command line option.
17318 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17320 * doc/tm.texi: Re-generate.
17322 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17325 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17326 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17328 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17330 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17332 2017-07-25 Richard Biener <rguenther@suse.de>
17334 PR tree-optimization/81455
17335 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17336 not walk in cycles when looking for guards.
17338 2017-07-25 Richard Biener <rguenther@suse.de>
17340 PR tree-optimization/81529
17341 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17342 when optimizing backedge uses.
17344 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17346 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17348 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17349 to dl_section_ref. On AIX, append an expression to subtract
17350 the size of the section length to dl_section_ref.
17352 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17354 * configure.ac: If any of the config.* scripts fail, exit 1.
17355 * configure: Regenerate.
17357 2017-07-25 Richard Biener <rguenther@suse.de>
17359 PR middle-end/81546
17360 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17361 of immediate uses, be more verbose on errors.
17363 2017-07-25 Richard Biener <rguenther@suse.de>
17365 PR tree-optimization/81510
17366 * tree-vect-loop.c (vect_is_simple_reduction): When the
17367 reduction stmt is not inside the loop bail out.
17369 2017-07-25 Richard Biener <rguenther@suse.de>
17371 PR tree-optimization/81303
17372 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17373 profitability check against LOOP_VINFO_NITERSM1.
17375 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17377 * domwalk.c (cmp_bb_postorder): Simplify.
17378 (sort_bbs_postorder): New function. Use it...
17379 (dom_walker::walk): ...here to optimize common cases.
17381 2017-07-25 Martin Liska <mliska@suse.cz>
17384 * ipa-visibility.c (function_and_variable_visibility): Make the
17385 redirection just on target that supports aliasing.
17386 Fix GNU coding style.
17388 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17391 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17393 * config/arm/rtems.h: Likewise.
17394 * config/bfin/rtems.h: Likewise.
17395 * config/i386/rtemself.h: Likewise.
17396 * config/lm32/rtems.h: Likewise.
17397 * config/m32c/rtems.h: Likewise.
17398 * config/m68k/rtemself.h: Likewise.
17399 * config/microblaze/rtems.h: Likewise.
17400 * config/mips/rtems.h: Likewise.
17401 * config/moxie/rtems.h: Likewise.
17402 * config/nios2/rtems.h: Likewise.
17403 * config/powerpcspe/rtems.h: Likewise.
17404 * config/rs6000/rtems.h: Likewise.
17405 * config/rtems.h: Likewise.
17406 * config/sh/rtems.h: Likewise.
17407 * config/sh/rtemself.h: Likewise.
17408 * config/sparc/rtemself.h: Likewise.
17410 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17413 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17414 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17415 * tree-ssa-structalias.c (alias_get_name): Same.
17417 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17420 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17421 instructions if no du chain is found.
17423 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17425 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17427 2017-07-25 Richard Biener <rguenther@suse.de>
17429 PR middle-end/81505
17430 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17433 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17435 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17436 upper-regs options.
17437 (ISA_2_7_MASKS_SERVER): Likewise.
17438 (ISA_3_0_MASKS_IEEE): Likewise.
17439 (OTHER_P8_VECTOR_MASKS): Likewise.
17440 (OTHER_VSX_VECTOR_MASKS): Likewise.
17441 (POWERPC_MASKS): Likewise.
17442 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17443 duplicate list of options.
17444 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17445 explicit -mupper-regs options.
17446 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17447 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17448 alias for -mupper-regs-df.
17449 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17450 (rs6000_init_hard_regno_mode_ok): Likewise.
17451 (rs6000_option_override_internal): Likewise.
17452 (rs6000_opt_masks): Likewise.
17453 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17454 options in terms of whether -mvsx or -mpower8-vector was used.
17455 (TARGET_UPPER_REGS_DI): Likewise.
17456 (TARGET_UPPER_REGS_SF): Likewise.
17457 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17458 -mupper-regs-* options.
17460 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17462 * passes.c (emergency_dump_function): Print some empty lines and a
17463 header before the RTL dump.
17465 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17467 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17469 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17472 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17473 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17475 2017-07-24 Carl Love <cel@us.ibm.com>
17477 * config/rs6000/rs6000-c.c: Add support for built-in functions
17478 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17479 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17480 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17481 vec_extract_fp_from_shortl): Add defines for the two builtins.
17482 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17483 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17485 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17486 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17487 * doc/extend.texi: Update the built-in documentation file for the
17488 new built-in function.
17490 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17493 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17495 * doc/generic.texi: Likewise.
17496 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17497 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17499 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17501 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17502 (aarch64_mls_elt_merge<mode>): Likewise.
17504 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17506 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17507 having __cxa_atexit.
17509 2017-07-23 Michael Collison <michael.collison@arm.com>
17511 * config/arm/arm.c (arm_option_override): Deprecate
17512 use of -mstructure-size-boundary.
17513 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17514 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17516 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17519 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17520 Reduce cost estimate for direct moves.
17522 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17525 * config/i386/i386.c (ix86_option_override_internal): Disable
17526 BMI, BMI2 and TBM instructions for -m16.
17528 2017-07-21 Carl Love <cel@us.ibm.com>
17530 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17531 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17532 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17534 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17535 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17536 VMULOSW): New enum "unspec" values.
17537 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17538 altivec_vmulosw): New patterns.
17539 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17540 VMULOSW): Add definitions.
17542 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17544 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17545 (qdf24xx): Likewise.
17546 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17547 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17548 (AARCH64_FL_V8_1): Renumber.
17549 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17550 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17551 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17552 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17553 rdma to feature modifiers list.
17555 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17557 PR middle-end/56727
17558 * ipa-visibility (function_and_variable_visibility): Convert
17559 recursive PLT call to direct call if appropriate.
17561 2017-07-21 Andrew Pinski <apinski@cavium.com>
17563 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17564 operand 1 to see if the types precision matches.
17565 * fold-const.c (operand_equal_p): Likewise.
17567 2017-07-21 Richard Biener <rguenther@suse.de>
17569 PR tree-optimization/81303
17570 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17571 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17572 (vect_peeling_hash_get_lowest_cost): Adjust.
17573 (vect_enhance_data_refs_alignment): Likewise. Use
17574 vect_get_peeling_costs_all_drs to compute the penalty for no
17575 peeling to match up costs.
17577 2017-07-21 Richard Biener <rguenther@suse.de>
17579 PR tree-optimization/81500
17580 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17581 we didn't identify a reduction path.
17583 2017-07-21 Tom de Vries <tom@codesourcery.com>
17584 Cesar Philippidis <cesar@codesourcery.com>
17586 PR gcov-profile/81442
17587 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17590 2017-07-21 Tom de Vries <tom@codesourcery.com>
17593 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17595 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17596 nvptx_override_options_after_change.
17598 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17600 * dwarf2out.c (output_file_names): Avoid double testing for
17601 dwarf_version >= 5.
17603 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17605 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17607 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17609 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17611 (try_crossjump_to_edge): Do not punt on partitioned functions.
17613 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17615 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17616 Put all BBs reachable only via paths crossing cold region to cold
17618 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17620 2016-07-21 Richard Biener <rguenther@suse.de>
17622 PR tree-optimization/81303
17623 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17624 into account prologue and epilogue iterations when raising
17625 min_profitable_iters to sth at least covering one vector iteration.
17627 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17629 * config/arm/arm.c (arm_test_cpu_arch_dat):
17632 2017-07-20 Nathan Sidwell <nathan@acm.org>
17634 Remove TYPE_METHODS.
17635 * tree.h (TYPE_METHODS): Delete.
17636 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17637 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17638 (dbxout_type_methods): Scan TYPE_FIELDS.
17639 (dbxout_type): Don't check TYPE_METHODS here.
17640 * function.c (use_register_for_decl): Always ignore register for
17641 class types when not optimizing.
17642 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17643 * tree.c (free_lang_data_in_type): Stitch out member functions and
17644 templates from TYPE_FIELDS.
17645 (build_distinct_type_copy, verify_type_variant,
17646 verify_type): Member fns are on TYPE_FIELDS.
17647 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17648 * tree-pretty-print.c (dump_generic_node): Likewise.
17650 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17653 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17654 V2TImode and V4TImode.
17655 (ix86_expand_vector_extract): Likewise.
17656 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17657 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17658 (ssescalarmode): Handle V4TImode and V2TImode.
17659 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17660 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17661 (VEXTRACTI128_MODE): New mode iterator.
17662 (splitter for *vec_extractv?ti first element): New.
17663 (VEC_INIT_MODE): New mode iterator.
17664 (vec_init<mode>): Consolidate 3 expanders into one using
17665 VEC_INIT_MODE mode iterator.
17667 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17669 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17670 non_spilled_static_chain_regno_p.
17672 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17674 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17676 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17678 * bb-reorder.c (connect_traces): Allow copying of blocks within
17681 2017-07-20 Richard Biener <rguenther@suse.de>
17683 * gimple.h (gimple_phi_result): Add gphi * overload.
17684 (gimple_phi_result_ptr): Likewise.
17685 (gimple_phi_arg): Likewise. Adjust index assert to only
17686 allow actual argument accesses rather than all slots available
17688 (gimple_phi_arg_def): Add gphi * overload.
17689 * tree-phinodes.c (make_phi_node): Initialize only actual
17691 (resize_phi_node): Clear memory not covered by old node,
17692 do not initialize excess argument slots.
17693 (reserve_phi_args_for_new_edge): Initialize new argument slot
17696 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17698 PR tree-optimization/81388
17700 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17702 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17703 by removing computation of may_be_zero.
17705 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17706 Tom de Vries <tom@codesourcery.com>
17708 PR middle-end/81030
17709 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17710 when gimple level profile disagrees with what RTL expander did.
17712 2017-07-20 Richard Biener <rguenther@suse.de>
17714 PR tree-optimization/61171
17715 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17716 (vect_analyze_stmt): Add slp instance parameter.
17717 (vectorizable_reduction): Likewise.
17718 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17719 (vect_is_simple_reduction): Deal with chains not detected
17720 as SLP reduction chain, specifically not properly associated
17721 chains containing a mix of plus/minus.
17722 (get_reduction_op): Remove.
17723 (get_initial_defs_for_reduction): Simplify, pass in whether
17724 this is a reduction chain, pass in the SLP node for the PHIs.
17725 (vect_create_epilog_for_reduction): Get the SLP instance as
17727 (vectorizable_reduction): Get the SLP instance as arg.
17728 During analysis remember the SLP node with the PHIs in the
17729 instance. Simplify getting at the vectorized reduction PHIs.
17730 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17731 through SLP instance.
17732 (vect_slp_analyze_operations): Likewise.
17733 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17734 (vect_transform_stmt): Likewise.
17736 2017-07-20 Tom de Vries <tom@codesourcery.com>
17738 PR tree-optimization/81489
17739 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17740 read of phi arg location to before loop that modifies phi.
17742 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17744 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17747 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17749 PR middle-end/81331
17750 * except.c (execute): Fix ordering issue.
17752 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17754 PR rtl-optimization/81423
17755 * combine.c (make_compound_operation_int): Don't try to optimize
17756 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17758 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17760 PR rtl-optimization/81423
17761 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17762 with a constant that is -1 in the truncated to mode.
17764 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17766 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17767 (determine_unlikely_bbs): ... here.
17768 * predict.h (propagate_unlikely_bbs_forward): Declare.
17769 * cfgexpand.c (pass_expand::execute): Use it.
17770 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17772 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17773 propagate_unlikely_bbs_forward.
17775 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17777 PR middle-end/81331
17778 * except.c (maybe_add_nop_after_section_switch): New function.
17781 2017-07-19 Tom de Vries <tom@codesourcery.com>
17783 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17785 2017-07-19 Tom de Vries <tom@codesourcery.com>
17787 * gimple.h (gimple_phi_arg): Make assert more strict.
17789 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17791 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17792 * config/rs6000/mmintrin.h: New file.
17793 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17795 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17797 PR tree-optimization/81346
17798 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17800 2017-07-19 Tom de Vries <tom@codesourcery.com>
17802 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17804 2017-07-19 Tom de Vries <tom@codesourcery.com>
17806 * config/nvptx/nvptx-modes.def: Add V2DImode.
17807 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17808 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17809 (nvptx_output_mov_insn): Handle lack of mov.b128.
17810 (nvptx_print_operand): Handle 'H' and 'L' codes.
17811 (nvptx_vector_mode_supported): Allow V2DImode.
17812 (nvptx_preferred_simd_mode): New function.
17813 (nvptx_data_alignment): New function.
17814 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17815 nvptx_preferred_simd_mode.
17816 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17818 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17820 2017-07-19 Tom de Vries <tom@codesourcery.com>
17822 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17823 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17824 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17825 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17826 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17827 (mov<VECIM>_insn): New define_insn.
17828 (define_expand "mov<VECIM>): New define_expand.
17830 2017-07-19 Tom de Vries <tom@codesourcery.com>
17832 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17834 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17836 PR tree-optimization/81346
17837 * fold-const.h (fold_div_compare, range_check_type): Declare.
17838 * fold-const.c (range_check_type): New function.
17839 (build_range_check): Use range_check_type.
17840 (fold_div_compare): No longer static, rewritten into
17841 a match.pd helper function.
17842 (fold_comparison): Don't call fold_div_compare here.
17843 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17844 as helper function.
17846 2017-07-19 Nathan Sidwell <nathan@acm.org>
17848 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17849 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17850 * tree.c (find_decls_types_r, verify_type): Use
17851 TYPE_{MIN,MAX}_VALUE_RAW.
17852 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17853 (hash_tree): Likewise.
17854 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17856 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17859 2017-07-18 Tom de Vries <tom@codesourcery.com>
17861 PR middle-end/81464
17862 * omp-expand.c (expand_omp_for_static_chunk): Handle
17863 equal-argument loop exit phi.
17865 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17868 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17869 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17870 operand 2 predicate.
17871 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17872 operand 2 predicate.
17873 (ror,rol -> rorx splitters): Use const_int_operand as
17874 operand 2 predicate.
17876 2017-06-18 Richard Biener <rguenther@suse.de>
17878 PR tree-optimization/81410
17879 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17880 the gap in the ! slp_perm SLP case after each group.
17882 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17884 PR middle-end/81463
17885 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17888 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17890 PR middle-end/81462
17891 * predict.c (set_even_probabilities): Cleanup; do not affect
17892 probabilities that are already known.
17893 (combine_predictions_for_bb): Call even when count is set.
17895 2017-07-18 Nathan Sidwell <nathan@acm.org>
17897 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17900 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17903 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17904 optimization for loop niter analysis.
17906 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17909 * config/avr/avr.c (avr_optimize_casesi): Don't use
17910 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17912 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17914 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17915 body_cost_vec from _vect_peel_extended_info.
17916 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17917 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17920 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17922 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17924 2017-07-18 Richard Biener <rguenther@suse.de>
17926 PR tree-optimization/80620
17927 PR tree-optimization/81403
17928 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17929 info when re-using a VN table entry.
17931 2017-07-18 Richard Biener <rguenther@suse.de>
17933 PR tree-optimization/81418
17934 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17935 vectype_in. Verify that with lane-reducing reduction operations
17936 we have a single def-use cycle.
17938 2017-07-17 Carl Love <cel@us.ibm.com>
17940 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17942 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17943 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17944 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17945 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17946 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17947 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17948 VMULOSW): New enum "unspec" values.
17949 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17950 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17951 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17952 altivec_vmulosw): New patterns.
17953 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17954 VMULOSW): Add definitions.
17955 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17957 * config/alpha/alpha.c: Include predict.h.
17959 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17961 * tree-vrp.c (compare_assert_loc): Fix comparison function
17962 to return predictable results.
17964 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17966 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17968 (subdi3): Likewise.
17969 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17970 * doc/invoke.texi (mexpand-adddi): Update text.
17972 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17974 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17975 that also clobbers the CC register. The old expand code is moved
17977 (*arc_clzsi2): ... here.
17978 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17979 the CC register. The old expand code is moved to ...
17980 (arc_ctzsi2): ... here.
17982 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17984 * config/arc/arc.opt (mindexed-loads): Use initial value
17985 TARGET_INDEXED_LOADS_DEFAULT.
17986 (mauto-modify-reg): Use initial value
17987 TARGET_AUTO_MODIFY_REG_DEFAULT.
17988 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17989 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17990 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17991 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17993 2017-07-17 Martin Liska <mliska@suse.cz>
17996 * opts.c (finish_options): Do not allow -fgnu-tm
17997 w/ -fsanitize={kernel-,}address. Say sorry.
17999 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18002 * tree-loop-distribution.c (classify_partition): Only assert on
18003 numer of iterations.
18004 (merge_dep_scc_partitions): Delete prameter. Update function call.
18005 (distribute_loop): Remove code handling loop with unknown niters.
18006 (pass_loop_distribution::execute): Skip loop with unknown niters.
18008 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18011 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18012 function sort_partitions_by_post_order.
18014 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18016 PR tree-optimization/81374
18017 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18018 the max index of basic blocks, rather than number of basic blocks.
18020 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18022 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18024 (arc_legitimate_pic_operand_p): Likewise.
18025 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18027 (arc_needs_pcl_p): Likewise.
18028 (arc_legitimate_pc_offset_p): Likewise.
18029 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18030 function is also used in constrains.md.
18031 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18032 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18033 PLUS. Only return true/false in known cases, otherwise assert.
18034 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18035 is already called in arc_legitimate_constant_p.
18036 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18038 (LEGITIMATE_PIC_OPERAND_P): Use
18039 arc_raw_symbolic_reference_mentioned_p function.
18040 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18045 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18046 Andrew Burgess <andrew.burgess@embecosm.com>
18048 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18049 (arc_return_address_register): New function.
18050 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18051 (arc_handle_fndecl_attribute): Add naked attribute.
18052 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18053 (TARGET_WARN_FUNC_RETURN): Likewise.
18054 (arc_allocate_stack_slots_for_args): New function.
18055 (arc_warn_func_return): Likewise.
18056 (machine_function): Change type fn_type.
18057 (arc_compute_function_type): Consider new naked function type,
18058 change function return type.
18059 (arc_must_save_register): Adapt to handle new
18060 arc_compute_function_type's return type.
18061 (arc_expand_prologue): Likewise.
18062 (arc_expand_epilogue): Likewise.
18063 (arc_return_address_regs): Delete.
18064 (arc_return_address_register): New function.
18065 (arc_epilogue_uses): Use above function.
18066 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18067 (arc_function_type): Change encoding, add naked type.
18068 (ARC_INTERRUPT_P): Change to handle the new encoding.
18069 (ARC_FAST_INTERRUPT_P): Likewise.
18070 (ARC_NORMAL_P): Define.
18071 (ARC_NAKED_P): Likewise.
18072 (arc_compute_function_type): Delete prototype.
18073 * config/arc/arc.md (in_ret_delay_slot): Use
18074 arc_return_address_register function.
18075 (simple_return): Likewise.
18076 (p_return_i): Likewise.
18078 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18080 PR tree-optimization/81428
18081 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18082 can't be built for those types.
18084 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18086 Remove stuff dead since r239246.
18088 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18089 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18090 (avr_inform_devices): Remove dead stuff.
18092 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18094 * config/arm/arm_neon.h: Fix softp typo.
18096 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18098 PR tree-optimization/81365
18099 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18100 aggregate moves onto bb predecessor edges, make sure there are no
18101 loads that could alias the lhs in between the start of bb and the
18104 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18107 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18108 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18109 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18111 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18113 PR tree-optimization/81396
18114 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18115 (init_symbolic_number): Initialize it to 1.
18116 (perform_symbolic_merge): Add n_ops from both operands into the new
18118 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18119 without base_addr as useless if they need more than one operation.
18120 (bswap_replace): Handle !bswap case for NULL base_addr.
18122 2017-07-17 Tom de Vries <tom@codesourcery.com>
18125 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18128 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18130 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18131 conditional builtin define __FIX_LEON3FT_B2BST.
18133 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18135 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18136 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18139 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18141 PR rtl-optimization/81424
18142 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18143 to remove potential trapping from operands if -fnon-call-exceptions.
18145 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18147 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18148 profile_proability for scalling.
18149 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18151 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18153 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18155 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18157 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18158 fixpoint arithmetics.
18160 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18162 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18163 fixpoint arithmetics.
18165 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18167 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18168 fixpoint arithmetics.
18170 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18172 * profile-count.h (profile_probability::from_reg_br_prob_note,
18173 profile_probability::to_reg_br_prob_note): New functions.
18174 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18175 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18176 * predict.c (probability_reliable_p): Update.
18177 (edge_probability_reliable_p): Update.
18178 (br_prob_note_reliable_p): Update.
18179 (invert_br_probabilities): Update.
18180 (add_reg_br_prob_note): New function.
18181 (combine_predictions_for_insn): Update.
18182 * asan.c (asan_clear_shadow): Update.
18183 * cfgbuild.c (compute_outgoing_frequencies): Update.
18184 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18185 (update_br_prob_note): Update.
18186 (rtl_verify_edges): Update.
18187 (purge_dead_edges): Update.
18188 (fixup_reorder_chain): Update.
18189 * emit-rtl.c (try_split): Update.
18190 * ifcvt.c (cond_exec_process_insns): Update.
18191 (cond_exec_process_if_block): Update.
18192 (dead_or_predicable): Update.
18193 * internal-fn.c (expand_addsub_overflow): Update.
18194 (expand_neg_overflow): Update.
18195 (expand_mul_overflow): Update.
18196 * loop-doloop.c (doloop_modify): Update.
18197 * loop-unroll.c (compare_and_jump_seq): Update.
18198 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18199 * predict.h: Update.
18200 * reorg.c (mostly_true_jump): Update.
18202 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18203 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18204 * config/arc/arc.c: (emit_unlikely_jump): Update.
18205 * config/arm/arm.c: (emit_unlikely_jump): Update.
18206 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18207 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18208 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18209 (ix86_print_operand): Update.
18210 (ix86_split_fp_branch): Update.
18211 (predict_jump): Update.
18212 * config/ia64/ia64.c (ia64_print_operand): Update.
18213 * config/mmix/mmix.c (mmix_print_operand): Update.
18214 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18215 (rs6000_expand_split_stack_prologue): Update.
18216 * config/rs6000/rs6000.c: Update.
18217 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18218 (s390_expand_vec_movstr): Update.
18219 (s390_expand_cs_tdsi): Update.
18220 (s390_expand_split_stack_prologue): Update.
18221 * config/sh/sh.c (sh_print_operand): Update.
18222 (expand_cbranchsi4): Update.
18223 (expand_cbranchdi4): Update.
18224 * config/sparc/sparc.c (output_v9branch): Update.
18225 * config/spu/spu.c (get_branch_target): Update.
18226 (ea_load_store_inline): Update.
18227 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18228 * config/tilepro/tilepro.c: Update.
18230 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18232 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18233 (gimplify_save_expr): Likewise.
18235 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18237 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18239 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18241 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18243 * timevar.def (TV_IPA_FNSUMMARY): Define.
18245 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18247 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18248 to back store errata sensitive sequence from being generated.
18249 (sqrtdf2_fix): Likewise.
18251 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18253 * tree-ssa-threadupdate.c (compute_path_counts,
18254 update_joiner_offpath_counts): Use profile_probability.
18256 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18259 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18261 * config/arm/arm-c.c (arm_cpu_builtins): Define
18262 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18264 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18266 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18267 array entries to represent __ieee128 versions of the
18268 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
18269 scalar_extract_sig, and scalar_insert_exp built-in functions.
18270 (altivec_resolve_overloaded_builtin): Add special case handling
18271 for the __builtin_scalar_insert_exp function, as represented by
18272 the P9V_BUILTIN_VEC_VSIEDP constant.
18273 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
18274 exponent support for __ieee128 argument.
18275 (VSESQP): Add scalar extract signature support for __ieee128
18277 (VSTDCNQP): Add scalar test negative support for __ieee128
18279 (VSIEQP): Add scalar insert exponent support for __int128 argument
18280 with __ieee128 result.
18281 (VSIEQPF): Add scalar insert exponent support for __ieee128
18282 argument with __ieee128 result.
18283 (VSTDCQP): Add scalar test data class support for __ieee128
18285 (VSTDCNQP): Add overload support for scalar test negative with
18286 __ieee128 argument.
18287 (VSTDCQP): Add overload support for scalar test data class
18288 __ieee128 argument.
18289 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
18290 UNSPEC_VSX_SXSIGDP.
18291 (UNSPEC_VSX_SIEXPQP): New constant.
18292 (xsxexpqp): New insn for VSX scalar extract exponent quad
18294 (xsxsigqp): New insn for VSX scalar extract significand quad
18296 (xsiexpqpf): New insn for VSX scalar insert exponent quad
18297 precision with floating point argument.
18298 (xststdcqp): New expand for VSX scalar test data class quad
18300 (xststdcnegqp): New expand for VSX scalar test negative quad
18302 (xststdcqp): New insn to match expansions for VSX scalar test data
18303 class quad precision and VSX scalar test negative quad precision.
18304 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18305 special case operand checking to enforce that second operand of
18306 VSX scalar test data class with quad precision argument is a 7-bit
18308 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18309 prototypes and descriptions of __ieee128 versions of
18310 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18311 scalar_test_data_class, and scalar_test_neg built-in functions.
18313 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18315 PR tree-optimization/81162
18316 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18317 replace a negate with an add.
18319 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18321 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18323 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18325 * config/arm/arm-c.c (arm_cpu_builtins): Define
18326 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18328 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18330 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18331 (armv8-r): Set ARM Cortex-R52 as default CPU.
18332 * config/arm/arm-tables.opt: Regenerate.
18333 * config/arm/arm-tune.md: Regenerate.
18334 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18336 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18337 extension for -mcpu=cortex-r52.
18339 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18341 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18342 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18343 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18344 (fp-armv8): Define it as FP_ARMv8 only.
18345 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18346 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18348 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18349 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18350 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18351 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18352 than TARGET_FPU_ARMV8.
18353 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18354 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18355 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18357 * config/arm/neon.md (neon_vrint): Likewise.
18358 (neon_vcvt): Likewise.
18359 (neon_<fmaxmin_op><mode>): Likewise.
18360 (<fmaxmin><mode>3): Likewise.
18361 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18362 * config/arm/predicates.md (arm_cond_move_operator): Check against
18363 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18365 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18367 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18368 to top of function.
18370 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18372 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18373 loop in comment with memset.
18375 2017-07-14 Martin Liska <mliska@suse.cz>
18377 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18378 * dwarf2out.c (is_java): Remove the function.
18379 (output_pubname): Remove usage of the function.
18380 (lower_bound_default): Remove usage of DW_LANG_Java.
18381 (gen_compile_unit_die): Likewise.
18382 * gcc.c: Remove compiler defaults for .java and .zip files.
18383 * gimple-expr.c (remove_suffix): Change as there's no longer
18384 extension than 4-letter one.
18385 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18386 (gimplify_save_expr): Likewise.
18387 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18388 as it's possible even for other languages than Java.
18389 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18390 * lto-opts.c (lto_write_options): Remove reference to Java.
18391 * opts.c (strip_off_ending): Update file extension handling.
18392 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18393 * tree-eh.c (lower_resx): Likewise.
18394 * tree.c (free_lang_data_in_type): Remove dead code.
18395 (find_decls_types_r): Likewise.
18396 (build_common_builtin_nodes): Remove Java from a comment.
18397 (verify_type): Remove dead code.
18398 * varasm.c (assemble_external): Remove Java from a comment.
18400 2017-07-14 Martin Liska <mliska@suse.cz>
18402 * opts.c (finish_options): Add quotes.
18403 (common_handle_option): Likewise.
18405 2017-07-14 Martin Liska <mliska@suse.cz>
18407 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18408 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18409 Remove N_SO_PASCAL.
18410 * dwarf2out.c (lower_bound_default): Do not handle
18412 (gen_compile_unit_die): Likewise.
18413 * gcc.c: Remove default extension binding for GNU Pascal.
18414 * stmt.c: Remove Pascal language from a comment.
18415 * xcoffout.c: Likewise.
18417 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18420 * diagnostic-show-locus.c (fixit_cmp): New function.
18421 (layout::layout): Sort m_fixit_hints.
18422 (column_range::column_range): Assert that the values are valid.
18423 (struct char_span): New struct.
18424 (correction::overwrite): New method.
18425 (struct source_line): New struct.
18426 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18427 calls in terms of classes source_line and char_span, and
18428 correction::overwrite.
18429 (selftest::test_overlapped_fixit_printing_2): New function.
18430 (selftest::diagnostic_show_locus_c_tests): Call it.
18432 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18434 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18435 early if there is no lhs.
18437 2017-07-13 Martin Liska <mliska@suse.cz>
18439 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18440 (gen_reference_type_die): Likewise.
18441 * stor-layout.c: Remove Pascal-related comment.
18443 2017-07-13 Martin Liska <mliska@suse.cz>
18445 * opts.c (finish_options): Add quotes to error messages.
18446 (parse_sanitizer_options): Likewise.
18448 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18450 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18452 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18454 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18456 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18458 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18460 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18462 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18465 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18466 provides the hardware capability bits, define the macro
18467 __BUILTIN_CPU_SUPPORTS__.
18468 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18469 if GLIBC does not provide the hardware capability bits. Add a
18470 gcc_unreachable call if the built-in cpu function is neither
18471 __builtin_cpu_is nor __builtin_cpu_supports.
18472 (rs6000_get_function_versions_dispatcher): Change the warning
18473 that an old GLIBC is used which does not export the capability
18474 bits to be an error.
18475 * doc/extend.texi (target_clones attribute): Document the
18476 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18477 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18478 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18479 the macros defined by GCC if the newer GLIBC is available.
18481 2017-07-12 Jeff Law <law@redhat.com>
18483 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18484 remaining includes slightly.
18485 * config/riscv/riscv-builtins.c: Include profile-count.h.
18487 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18490 * config/avr/avr.c (avr_set_current_function): In diagnostic
18491 messages: Quote keywords and (parts of) identifiers.
18492 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18495 2017-07-12 Carl Love <cel@us.ibm.com>
18497 * config/rs6000/rs6000-c.c: Add support for built-in functions
18498 vector bool char vec_revb (vector bool char);
18499 vector bool short vec_revb (vector short char);
18500 vector bool int vec_revb (vector bool int);
18501 vector bool long long vec_revb (vector bool long long);
18502 * doc/extend.texi: Update the built-in documentation file for the
18503 new built-in functions.
18505 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18507 * config/s390/s390.md: Remove movcc splitter.
18509 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18511 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18512 load/store on condition.
18514 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18517 * config/avr/avr.c (avr_encode_section_info)
18518 [progmem && !TREE_READONLY]: Error if progmem object needs
18521 2017-07-11 Michael Collison <michael.collison@arm.com>
18523 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18526 2017-07-11 Carl Love <cel@us.ibm.com>
18528 * config/rs6000/rs6000-c.c: Add support for builtins
18529 vector unsigned int vec_parity_lsbb (vector signed int);
18530 vector unsigned int vec_parity_lsbb (vector unsigned int);
18531 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18532 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18533 vector unsigned long long vec_parity_lsbb (vector signed long long);
18534 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18535 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18536 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18537 * doc/extend.texi: Update the built-in documentation file for the
18538 new built-in functions.
18540 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18542 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18543 (layout::m_primary_loc): New field.
18544 (layout::layout): Initialize new field. Move location filtering
18545 logic from here to...
18546 (layout::maybe_add_location_range): ...this new method. Add
18547 support for filtering to just the lines already specified by other
18549 (layout::will_show_line_p): New method.
18550 (gcc_rich_location::add_location_if_nearby): New method.
18551 (selftest::test_add_location_if_nearby): New test function.
18552 (selftest::diagnostic_show_locus_c_tests): Call it.
18553 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18556 2017-07-11 Tom de Vries <tom@codesourcery.com>
18558 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18559 (bb_first_real_insn): New function.
18560 (nvptx_single): Add extra initialization of broadcasted condition
18563 2017-07-11 Nathan Sidwell <nathan@acm.org>
18565 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18567 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18569 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18570 Binutils doc as TEXI will mess them up.
18571 * doc/invoke.texi (AVR Options): Same here.
18573 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18575 * config/sparc/sparc.opt (mfix-ut700): New option.
18576 (mfix-gr712rc): Likewise.
18577 (sparc_fix_b2bst): New variable.
18578 * doc/invoke.texi (SPARC options): Document them.
18579 (ARM options): Fix warnings.
18580 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18581 instructions to prevent sequences that can trigger the store-store
18582 errata for certain LEON3FT processors.
18583 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18584 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18585 * config/sparc/sparc.md (fix_b2bst): New attribute.
18586 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18588 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18591 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18593 (*rsqrtsf2_sse): Ditto.
18595 (div<mode>3): Macroize insn from divdf3 and divsf3
18596 using MODEF mode iterator.
18598 2017-07-10 Martin Sebor <msebor@redhat.com>
18600 PR tree-optimization/80397
18601 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18602 instead of testing for equality to INTEGER_TYPE.
18604 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18606 * config.gcc: Remove uclibc from arc target spec.
18608 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18610 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18612 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18615 * lto-wrapper.c (remove_option): New function.
18616 (merge_and_complain): Merge PIC/PIE options more realistically.
18618 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18620 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18624 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18625 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18626 * config.in: Regenerate.
18627 * configure: Regenerate.
18628 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18629 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18630 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18631 (TARGET_GASISR_PROLOGUES): ...target mask.
18632 * common/config/avr/avr-common.c
18633 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18634 Set -mgas-isr-prologues.
18635 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18636 INSERT_PASS_BEFORE for it.
18637 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18638 * config/avr/avr.c (avr_option_override)
18639 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18640 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18641 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18642 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18643 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18645 (make_avr_pass_pre_proep): New function.
18646 (emit_push_sfr) <treg>: Add argument to function and use it
18647 instead of TMP_REG.
18648 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18649 and set machine->gasisr.yes.
18650 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18651 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18652 __gcc_isr.n_pushed to .L__stack_usage.
18653 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18654 (avr_asm_final_postscan_insn): ...this new static function.
18655 * config/avr/avr.h (machine_function)
18656 <is_no_gccisr, use_L__stack_usage>: New fields.
18657 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18658 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18659 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18660 (gasisr, *gasisr): New expander and insn.
18661 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18662 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18663 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18665 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18667 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18670 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18672 Move jump-tables out of .text again.
18675 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18676 (ASM_OUTPUT_ADDR_VEC): New function.
18677 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18678 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18679 INSN_ADDRESSes as asm comment.
18680 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18681 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18682 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18683 * config/avr/avr.md (*tablejump): Adjust comment.
18684 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18685 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18687 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18688 (avr_output_addr_vec): New proto.
18689 (avr_log_t) <insn_addresses>: New field.
18691 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18694 * config/i386/i386.c (ix86_function_arg_advance): Set
18695 outgoing_args_on_stack to true if there are outgoing arguments
18697 (ix86_function_arg): Likewise.
18698 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18699 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18700 * config/i386/i386.h (machine_function): Add
18701 outgoing_args_on_stack.
18703 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18705 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18706 supporting pthreds.
18707 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18709 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18711 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18712 (REAL_H): Remove $(MACHMODE_H).
18713 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18715 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18716 $(MACHMODE_H) and double-int.h.
18717 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18719 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18722 2017-07-07 Andrew Pinski <apinski@cavium.com>
18724 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18725 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18727 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18729 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18730 Add warning if GCC was not configured to link against a GLIBC that
18731 exports the hardware capability bits.
18732 (make_resolver_func): Make resolver function private and not a
18733 COMDAT function. Create the name with clone_function_name instead
18734 of make_unique_name.
18737 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18738 correct operand in doing the split.
18740 2017-07-07 Carl Love <cel@us.ibm.com>
18742 * config/rs6000/rs6000-c: Add support for built-in function
18743 vector unsigned short vec_pack_to_short_fp32 (vector float,
18745 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18746 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18747 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18748 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18749 (convert_4f32_8i16): Add define_expand.
18750 * doc/extend.texi: Update the built-in documentation file for the
18751 new built-in function.
18753 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18755 * config/sparc/m8.md: New file.
18756 * config/sparc/sparc.md: Include m8.md.
18758 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18760 * config/sparc/sparc.opt: New option -mvis4b.
18761 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18762 (sparc_option_override): Handle VIS4B.
18763 (enum sparc_builtins): Define
18764 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18765 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18766 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18767 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18768 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18769 (check_constant_argument): New function.
18770 (sparc_vis_init_builtins): Define builtins
18771 __builtin_vis_dictunpack{8,16,32},
18772 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18773 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18774 __builtin_vis_fpcmpde{8,16,32}shl and
18775 __builtin_vis_fpcmpur{8,16,32}shl.
18776 (sparc_expand_builtin): Check that the constant operands to
18777 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18778 constant and in range.
18779 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18781 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18782 (SPARC_IMM5_P): Likewise.
18783 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18784 (enabled): Handle vis4b.
18785 (UNSPEC_DICTUNPACK): New unspec.
18786 (UNSPEC_FPCMPSHL): Likewise.
18787 (UNSPEC_FPUCMPSHL): Likewise.
18788 (UNSPEC_FPCMPDESHL): Likewise.
18789 (UNSPEC_FPCMPURSHL): Likewise.
18790 (cpu_feature): New CPU feature `vis4b'.
18791 (dictunpack{8,16,32}): New insns.
18792 (FPCSMODE): New mode iterator.
18793 (fpcscond): New code iterator.
18794 (fpcsucond): Likewise.
18795 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18796 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18797 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18798 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18799 * config/sparc/constraints.md: Define constraints `q' for unsigned
18800 2-bit integer constants and `t' for unsigned 5-bit integer
18802 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18804 (imm5_operand_dictunpack16): Likewise.
18805 (imm5_operand_dictunpack32): Likewise.
18806 (imm2_operand): Likewise.
18807 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18808 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18809 ditunpack* and fpcmp*shl builtins.
18811 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18813 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18814 * config.in: Add HAVE_AS_SPARC6 define.
18815 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18817 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18819 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18820 (CPP_CPU_SPEC): Handle m8.
18821 (ASM_CPU_SPEC): Likewise.
18822 * config/sparc/sparc-opts.h (enum processor_type): Add
18824 * config/sparc/sparc.c (m8_costs): New struct.
18825 (sparc_option_override): Handle TARGET_CPU_m8.
18826 (sparc32_initialize_trampoline): Likewise.
18827 (sparc64_initialize_trampoline): Likewise.
18828 (sparc_issue_rate): Likewise.
18829 (sparc_register_move_cost): Likewise.
18830 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18831 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18832 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18833 (CPP_CPU_SPEC): Handle M8.
18834 (ASM_CPU_SPEC): Likewise.
18835 (AS_M8_FLAG): Define.
18836 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18837 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18838 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18840 * configure: Regenerate.
18841 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18844 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18846 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18848 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18849 ("*movdi_insn_sp32"): Do not set v3pipe.
18850 ("*movsi_insn"): Likewise.
18851 ("*movdi_insn_sp64"): Likewise.
18852 ("*movsf_insn"): Likewise.
18853 ("*movdf_insn_sp32"): Likewise.
18854 ("*movdf_insn_sp64"): Likewise.
18855 ("*zero_extendsidi2_insn_sp64"): Likewise.
18856 ("*sign_extendsidi2_insn"): Likewise.
18857 ("*mov<VM32:mode>_insn"): Likewise.
18858 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18859 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18860 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18861 ("<vlop:code><VL:mode>3"): Likewise.
18862 ("*not_<vlop:code><VL:mode>3"): Likewise.
18863 ("*nand<VL:mode>_vis"): Likewise.
18864 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18865 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18866 ("one_cmpl<VL:mode>2"): Likewise.
18867 ("faligndata<VM64:mode>_vis"): Likewise.
18868 ("alignaddrsi_vis"): Likewise.
18869 ("alignaddrdi_vis"): Likweise.
18870 ("alignaddrlsi_vis"): Likewise.
18871 ("alignaddrldi_vis"): Likewise.
18872 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18873 ("bmaskdi_vis"): Likewise.
18874 ("bmasksi_vis"): Likewise.
18875 ("bshuffle<VM64:mode>_vis"): Likewise.
18876 ("cmask8<P:mode>_vis"): Likewise.
18877 ("cmask16<P:mode>_vis"): Likewise.
18878 ("cmask32<P:mode>_vis"): Likewise.
18879 ("pdistn<P:mode>_vis"): Likewise.
18880 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18882 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18884 * config/sparc/sparc.md ("subtype"): New insn attribute.
18885 ("*wrgsr_sp64"): Set insn subtype.
18886 ("*rdgsr_sp64"): Likewise.
18887 ("alignaddrsi_vis"): Likewise.
18888 ("alignaddrdi_vis"): Likewise.
18889 ("alignaddrlsi_vis"): Likewise.
18890 ("alignaddrldi_vis"): Likewise.
18891 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18892 ("fexpand_vis"): Likewise.
18893 ("fpmerge_vis"): Likewise.
18894 ("faligndata<VM64:mode>_vis"): Likewise.
18895 ("bshuffle<VM64:mode>_vis"): Likewise.
18896 ("cmask8<P:mode>_vis"): Likewise.
18897 ("cmask16<P:mode>_vis"): Likewise.
18898 ("cmask32<P:mode>_vis"): Likewise.
18899 ("fchksm16_vis"): Likewise.
18900 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18901 ("fmean16_vis"): Likewise.
18902 ("fp<plusminus_insn>64_vis"): Likewise.
18903 ("<plusminus_insn>v8qi3"): Likewise.
18904 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18905 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18906 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18907 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18908 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18909 ("*movqi_insn"): Likewise.
18910 ("*movhi_insn"): Likewise.
18911 ("*movsi_insn"): Likewise.
18912 ("movsi_pic_gotdata_op"): Likewise.
18913 ("*movdi_insn_sp32"): Likewise.
18914 ("*movdi_insn_sp64"): Likewise.
18915 ("movdi_pic_gotdata_op"): Likewise.
18916 ("*movsf_insn"): Likewise.
18917 ("*movdf_insn_sp32"): Likewise.
18918 ("*movdf_insn_sp64"): Likewise.
18919 ("*zero_extendhisi2_insn"): Likewise.
18920 ("*zero_extendqihi2_insn"): Likewise.
18921 ("*zero_extendqisi2_insn"): Likewise.
18922 ("*zero_extendqidi2_insn"): Likewise.
18923 ("*zero_extendhidi2_insn"): Likewise.
18924 ("*zero_extendsidi2_insn_sp64"): Likewise.
18925 ("ldfsr"): Likewise.
18926 ("prefetch_64"): Likewise.
18927 ("prefetch_32"): Likewise.
18928 ("tie_ld32"): Likewise.
18929 ("tie_ld64"): Likewise.
18930 ("*tldo_ldub_sp32"): Likewise.
18931 ("*tldo_ldub1_sp32"): Likewise.
18932 ("*tldo_ldub2_sp32"): Likewise.
18933 ("*tldo_ldub_sp64"): Likewise.
18934 ("*tldo_ldub1_sp64"): Likewise.
18935 ("*tldo_ldub2_sp64"): Likewise.
18936 ("*tldo_ldub3_sp64"): Likewise.
18937 ("*tldo_lduh_sp32"): Likewise.
18938 ("*tldo_lduh1_sp32"): Likewise.
18939 ("*tldo_lduh_sp64"): Likewise.
18940 ("*tldo_lduh1_sp64"): Likewise.
18941 ("*tldo_lduh2_sp64"): Likewise.
18942 ("*tldo_lduw_sp32"): Likewise.
18943 ("*tldo_lduw_sp64"): Likewise.
18944 ("*tldo_lduw1_sp64"): Likewise.
18945 ("*tldo_ldx_sp64"): Likewise.
18946 ("*mov<VM32:mode>_insn"): Likewise.
18947 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18948 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18950 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18952 * config/sparc/sparc.md ("type"): New insn type viscmp.
18953 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18955 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18956 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18957 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18958 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18960 ("n7_vis_logical_11cycle"): Likewise.
18961 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18962 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18963 * config/sparc/niagara.md ("niag_vis"): Likewise.
18964 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18965 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18967 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18969 * config/sparc/sparc.md: New instruction type `bmask'.
18970 (bmaskdi_vis): Use the `bmask' type.
18971 (bmasksi_vis): Likewise.
18972 * config/sparc/ultra3.md (us3_array): Likewise.
18973 * config/sparc/niagara7.md (n7_array): Likewise.
18974 * config/sparc/niagara4.md (n4_array): Likewise.
18975 * config/sparc/niagara2.md (niag2_vis): Likewise.
18976 (niag3_vis): Likewise.
18977 * config/sparc/niagara.md (niag_vis): Likewise.
18979 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18981 * ipa-comdats.c: Remove optimize check from gate.
18982 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18983 for functions not optimized.
18984 (ipa_fn_summary_read): Skip optimize check.
18985 (ipa_fn_summary_write): Likewise.
18986 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18988 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18990 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18991 (check_callers): Give up on caller which is not optimized.
18992 (inline_small_functions): Likewise.
18993 (ipa_inline): Do not give up when not optimizing.
18994 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18995 away unoptimizes cdtors.
18996 (whole_program_function_and_variable_visibility): Do
18997 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18998 * ipa.c (process_references): Do not check optimize.
18999 (symbol_table::remove_unreachable_nodes): Update optimize check.
19000 (set_writeonly_bit): Update optimize check.
19001 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19002 (pass_ipa_single_use::gate): Remove.
19004 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19006 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19007 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19008 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19009 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19010 permute_load, permute_store, adjust_extract, adjust_splat,
19011 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19012 replace_swap_with_copy, dump_swap_insn_table,
19013 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19014 recombine_lvx_pattern, recombine_stvx_pattern,
19015 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19016 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19017 to file rs6000-p8swap.c.
19018 * config/rs6000/rs6000-p8swap.c: New file.
19019 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19020 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19021 and rs6000*-*-* targets.
19023 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19025 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19027 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19029 * lto-wrapper.c (merge_and_complain): Do not merge
19030 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19031 fsigned_zeros, ftrapping_math, fwrapv.
19032 (append_compiler_options): Do not track these options.
19033 (append_linker_options): Likewie
19035 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19037 * cgraphunit.c (cgraph_node::finalize_function): When
19038 !flag_toplevel_reorde set no_reorder flag.
19039 (varpool_node::finalize_decl): Likewise.
19040 (symbol_table::compile): Drop no toplevel reorder path.
19042 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19044 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19045 edges; zero probability is not better than uninitialized.
19047 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19049 * asan.h (asan_sanitize_allocas_p): Declare.
19050 * asan.c (asan_sanitize_allocas_p): New function.
19051 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19052 (handle_builtin_alloca): Likewise.
19053 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19054 if !asan_sanitize_allocas_p.
19055 * params.def (asan-instrument-allocas): Add new option.
19056 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19057 * opts.c (common_handle_option): Disable allocas sanitization for
19060 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19062 * asan.c: Include gimple-fold.h.
19063 (get_last_alloca_addr): New function.
19064 (handle_builtin_stackrestore): Likewise.
19065 (handle_builtin_alloca): Likewise.
19066 (asan_emit_allocas_unpoison): Likewise.
19067 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19068 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19069 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19070 (instrument_builtin_call): Pass gimple iterator to
19071 get_mem_refs_of_builtin_call.
19072 (last_alloca_addr): New global.
19073 * asan.h (asan_emit_allocas_unpoison): Declare.
19074 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19075 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19076 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19077 if function calls alloca.
19078 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19079 * gimple-fold.h (replace_call_with_value): Declare.
19080 * internal-fn.c: Include asan.h.
19081 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19082 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19084 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19086 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19087 (C_SELFTEST_FLAGS): New.
19088 (CPP_SELFTEST_FLAGS): New.
19089 (SELFTEST_DEPS): New, from deps of s-selftest.
19090 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19091 (CPP_SELFTEST_DEPS): New.
19092 (selftest): Add dependency on s-selftest-c++.
19093 (s-selftest): Rename to...
19094 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19095 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19096 than SELFTEST_FLAGS.
19097 (selftest-gdb): Rename to...
19098 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19100 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19101 (selftest-valgrind): Rename to...
19102 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19104 (selftest-valgrind): Reintroduce as an alias for
19105 selftest-c-valgrind.
19106 (s-selftest-c++): New.
19107 (selftest-c++-gdb): New.
19108 (selftest-c++-valgrind): New.
19110 2017-07-06 Olivier Hainque <hainque@adacore.com>
19112 * gcc.c (process_command): When deciding if undefined variables
19113 should be ignored when processing specs, accept "gcc -v" as well.
19115 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19117 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19118 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19120 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19122 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19123 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19124 * config/arm/arm-tables.opt: Regenerate.
19125 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19127 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19129 2017-07-06 Carl Love <cel@us.ibm.com>
19131 * ChangeLog: Clean up from mid air collision
19133 2017-07-06 Carl Love <cel@us.ibm.com>
19135 * config/rs6000/rs6000-c.c: Add support for built-in functions
19136 vector signed int vec_subc (vector signed int, vector signed int);
19137 vector signed __int128 vec_subc (vector signed __int128,
19138 vector signed __int128);
19139 vector unsigned __int128 vec_subc (vector unsigned __int128,
19140 vector unsigned __int128);
19141 vector signed int vec_sube (vector signed int, vector signed int,
19142 vector signed int);
19143 vector unsigned int vec_sube (vector unsigned int,
19144 vector unsigned int,
19145 vector unsigned int);
19146 vector signed __int128 vec_sube (vector signed __int128,
19147 vector signed __int128,
19148 vector signed__int128);
19149 vector unsigned __int128 vec_sube (vector unsigned __int128,
19150 vector unsigned __int128,
19151 vector unsigned __int128);
19152 vector signed int vec_subec (vector signed int, vector signed int,
19153 vector signed int);
19154 vector unsigned int vec_subec (vector unsigned int,
19155 vector unsigned int,
19156 vector unsigned int);
19157 vector signed __int128 vec_subec (vector signed __int128,
19158 vector signed __int128,
19159 vector signed__int128);
19160 vector unsigned __int128 vec_subec (vector unsigned __int128,
19161 vector unsigned __int128,
19162 vector unsigned __int128);
19163 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19164 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19165 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19166 BU_ALTIVEC_OVERLOAD_X definitions.
19167 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19168 * doc/extend.texi: Update the built-in documentation file for the new
19169 built-in functions.
19171 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19174 * diagnostic-show-locus.c (layout::layout): Use start and finish
19175 spelling location for the start and finish of each range.
19176 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19177 Add unused aspect param.
19178 * input.c (expand_location_1): Add "aspect" param, and use it
19179 to access the correct part of the location.
19180 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19182 (expand_location_to_spelling_point): Likewise.
19183 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19184 param, and pass it to expand_location_1.
19186 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19188 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19189 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19190 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19191 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19192 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19193 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19194 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19195 _mm_maskz_getmant_ss): New intrinsics.
19196 (__builtin_ia32_getexpss128_mask): Changed to ...
19197 __builtin_ia32_getexpss128_round ... this.
19198 (__builtin_ia32_getexpsd128_mask): Changed to ...
19199 __builtin_ia32_getexpsd128_round ... this.
19200 * config/i386/i386-builtin-types.def
19201 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19202 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19203 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19204 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19205 __builtin_ia32_getmantss_mask_round): New builtins.
19206 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19207 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19208 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19209 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19210 * config/i386/sse.md
19211 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19212 avx512f_sgetexp<mode><mask_scalar_name>
19213 <round_saeonly_scalar_name> ... this.
19214 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19215 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19216 vgetexp<ssescalarmodesuffix>
19217 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19218 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19219 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19220 avx512f_vgetmant<mode><mask_scalar_name>
19221 <round_saeonly_scalar_name> ... this.
19222 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19223 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19224 vgetmant<ssescalarmodesuffix>
19225 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19226 %0<mask_scalar_operand4>, %1, %2
19227 <round_saeonly_scalar_mask_op4>, %3} ... this.
19228 * config/i386/subst.md (mask_scalar_operand4,
19229 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19230 round_saeonly_scalar_nimm_predicate): New subst attributes.
19232 2017-07-06 Julia Koval <julia.koval@intel.com>
19234 * config/i386/i386.c (ix86_erase_embedded_rounding):
19235 Remove code for old rounding pattern.
19237 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19239 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19241 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19243 * doc/sourcebuild.texi (Test Directives, Variants of
19244 dg-require-support): Add documentation for dg-require-stack-check.
19246 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19248 * config/i386/subst.md (mask_scalar, round_scalar,
19249 round_saeonly_scalar): New meta-templates.
19250 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19251 round_scalar_mask_operand3, round_scalar_mask_op3,
19252 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19253 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19254 round_saeonly_scalar_constraint,
19255 round_saeonly_scalar_prefix): New subst attribute.
19256 * config/i386/sse.md
19257 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19258 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19259 <round_scalar_name> ... this.
19260 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19261 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19262 <round_scalar_name> ... this.
19263 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19264 <sse>_vm<code><mode>3<mask_scalar_name>
19265 <round_saeonly_scalar_name> ... this.
19266 (v<plusminus_mnemonic><ssescalarmodesuffix>
19267 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19268 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19269 v<plusminus_mnemonic><ssescalarmodesuffix>
19270 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19271 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19272 (v<multdiv_mnemonic><ssescalarmodesuffix>
19273 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19274 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19275 v<multdiv_mnemonic><ssescalarmodesuffix>
19276 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19277 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19278 (v<maxmin_float><ssescalarmodesuffix>
19279 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
19280 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
19281 v<maxmin_float><ssescalarmodesuffix>
19282 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19283 %0<mask_scalar_operand3>, %1, %<iptr>2
19284 <round_saeonly_scalar_mask_op3>} ... this.
19286 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
19288 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
19289 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
19291 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
19292 Alan Hayward <alan.hayward@arm.com>
19293 David Sherwood <david.sherwood@arm.com>
19295 * combine.c (simplify_if_then_else): Remove "enum" before
19297 * compare-elim.c (can_eliminate_compare): Likewise.
19298 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19300 (aarch64_lookup_simd_builtin_type): Likewise.
19301 (aarch64_simd_builtin_type): Likewise.
19302 (aarch64_init_simd_builtin_types): Likewise.
19303 (aarch64_simd_expand_args): Likewise.
19304 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19306 (aarch64_reverse_mask): Likewise.
19307 (aarch64_simd_emit_reg_reg_move): Likewise.
19308 (aarch64_gen_adjusted_ldpstp): Likewise.
19309 (aarch64_ccmp_mode_to_code): Likewise.
19310 (aarch64_operands_ok_for_ldpstp): Likewise.
19311 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19312 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19314 (aarch64_min_divisions_for_recip_mul): Likewise.
19315 (aarch64_reassociation_width): Likewise.
19316 (aarch64_get_condition_code_1): Likewise.
19317 (aarch64_simd_emit_reg_reg_move): Likewise.
19318 (aarch64_simd_attr_length_rglist): Likewise.
19319 (aarch64_reverse_mask): Likewise.
19320 (aarch64_operands_ok_for_ldpstp): Likewise.
19321 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19322 (aarch64_gen_adjusted_ldpstp): Likewise.
19323 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19325 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19326 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19327 (arm_lookup_simd_builtin_type): Likewise.
19328 (arm_simd_builtin_type): Likewise.
19329 (arm_init_simd_builtin_types): Likewise.
19330 (arm_expand_builtin_args): Likewise.
19331 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19332 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19333 (ft32_setup_incoming_varargs): Likewise.
19334 (ft32_function_arg): Likewise.
19335 (ft32_function_arg_advance): Likewise.
19336 (ft32_pass_by_reference): Likewise.
19337 (ft32_arg_partial_bytes): Likewise.
19338 (ft32_valid_pointer_mode): Likewise.
19339 (ft32_addr_space_pointer_mode): Likewise.
19340 (ft32_addr_space_legitimate_address_p): Likewise.
19341 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19343 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19344 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19345 (iamcu_alignment): Likewise.
19346 (canonicalize_vector_int_perm): Likewise.
19347 (ix86_noce_conversion_profitable_p): Likewise.
19348 (ix86_mpx_bound_mode): Likewise.
19349 (ix86_operands_ok_for_move_multiple): Likewise.
19350 * config/microblaze/microblaze-protos.h
19351 (microblaze_expand_conditional_branch_reg): Likewise.
19352 * config/microblaze/microblaze.c
19353 (microblaze_expand_conditional_branch_reg): Likewise.
19354 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19356 (rs6000_reassociation_width): Likewise.
19357 (rs6000_invalid_binary_op): Likewise.
19358 (fusion_p9_p): Likewise.
19359 (emit_fusion_p9_load): Likewise.
19360 (emit_fusion_p9_store): Likewise.
19361 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19363 (riscv_hard_regno_mode_ok_p): Likewise.
19364 (riscv_address_insns): Likewise.
19365 (riscv_split_symbol): Likewise.
19366 (riscv_legitimize_move): Likewise.
19367 (riscv_function_value): Likewise.
19368 (riscv_hard_regno_nregs): Likewise.
19369 (riscv_expand_builtin): Likewise.
19370 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19371 (riscv_build_integer): Likewise.
19372 (riscv_split_integer): Likewise.
19373 (riscv_legitimate_constant_p): Likewise.
19374 (riscv_cannot_force_const_mem): Likewise.
19375 (riscv_regno_mode_ok_for_base_p): Likewise.
19376 (riscv_valid_base_register_p): Likewise.
19377 (riscv_valid_offset_p): Likewise.
19378 (riscv_valid_lo_sum_p): Likewise.
19379 (riscv_classify_address): Likewise.
19380 (riscv_legitimate_address_p): Likewise.
19381 (riscv_address_insns): Likewise.
19382 (riscv_load_store_insns): Likewise.
19383 (riscv_force_binary): Likewise.
19384 (riscv_split_symbol): Likewise.
19385 (riscv_force_address): Likewise.
19386 (riscv_legitimize_address): Likewise.
19387 (riscv_move_integer): Likewise.
19388 (riscv_legitimize_const_move): Likewise.
19389 (riscv_legitimize_move): Likewise.
19390 (riscv_address_cost): Likewise.
19391 (riscv_subword): Likewise.
19392 (riscv_output_move): Likewise.
19393 (riscv_canonicalize_int_order_test): Likewise.
19394 (riscv_emit_int_order_test): Likewise.
19395 (riscv_function_arg_boundary): Likewise.
19396 (riscv_pass_mode_in_fpr_p): Likewise.
19397 (riscv_pass_fpr_single): Likewise.
19398 (riscv_pass_fpr_pair): Likewise.
19399 (riscv_get_arg_info): Likewise.
19400 (riscv_function_arg): Likewise.
19401 (riscv_function_arg_advance): Likewise.
19402 (riscv_arg_partial_bytes): Likewise.
19403 (riscv_function_value): Likewise.
19404 (riscv_pass_by_reference): Likewise.
19405 (riscv_setup_incoming_varargs): Likewise.
19406 (riscv_print_operand): Likewise.
19407 (riscv_elf_select_rtx_section): Likewise.
19408 (riscv_save_restore_reg): Likewise.
19409 (riscv_for_each_saved_reg): Likewise.
19410 (riscv_register_move_cost): Likewise.
19411 (riscv_hard_regno_mode_ok_p): Likewise.
19412 (riscv_hard_regno_nregs): Likewise.
19413 (riscv_class_max_nregs): Likewise.
19414 (riscv_memory_move_cost): Likewise.
19415 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19416 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19417 (rl78_addr_space_address_mode): Likewise.
19418 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19420 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19421 (rs6000_reassociation_width): Likewise.
19422 (rs6000_invalid_binary_op): Likewise.
19423 (fusion_p9_p): Likewise.
19424 (emit_fusion_p9_load): Likewise.
19425 (emit_fusion_p9_store): Likewise.
19426 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19427 (ok_for_simple_move_operands): Likewise.
19428 (ok_for_simple_move_strict_operands): Likewise.
19429 (ok_for_simple_arith_logic_operands): Likewise.
19430 (visium_legitimize_reload_address): Likewise.
19431 (visium_select_cc_mode): Likewise.
19432 (output_cbranch): Likewise.
19433 (visium_split_double_move): Likewise.
19434 (visium_expand_copysign): Likewise.
19435 (visium_expand_int_cstore): Likewise.
19436 (visium_expand_fp_cstore): Likewise.
19437 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19438 (visium_function_arg): Likewise.
19439 (visium_function_arg_advance): Likewise.
19440 (visium_libcall_value): Likewise.
19441 (visium_setup_incoming_varargs): Likewise.
19442 (visium_legitimate_constant_p): Likewise.
19443 (visium_legitimate_address_p): Likewise.
19444 (visium_legitimize_address): Likewise.
19445 (visium_secondary_reload): Likewise.
19446 (visium_register_move_cost): Likewise.
19447 (visium_memory_move_cost): Likewise.
19448 (prepare_move_operands): Likewise.
19449 (ok_for_simple_move_operands): Likewise.
19450 (ok_for_simple_move_strict_operands): Likewise.
19451 (ok_for_simple_arith_logic_operands): Likewise.
19452 (visium_function_value_1): Likewise.
19453 (rtx_ok_for_offset_p): Likewise.
19454 (visium_legitimize_reload_address): Likewise.
19455 (visium_split_double_move): Likewise.
19456 (visium_expand_copysign): Likewise.
19457 (visium_expand_int_cstore): Likewise.
19458 (visium_expand_fp_cstore): Likewise.
19459 (visium_split_cstore): Likewise.
19460 (visium_select_cc_mode): Likewise.
19461 (visium_split_cbranch): Likewise.
19462 (output_cbranch): Likewise.
19463 (visium_print_operand_address): Likewise.
19464 * expmed.c (flip_storage_order): Likewise.
19465 * expmed.h (emit_cstore): Likewise.
19466 (flip_storage_order): Likewise.
19467 * genrecog.c (validate_pattern): Likewise.
19468 * hsa-gen.c (gen_hsa_addr): Likewise.
19469 * internal-fn.c (expand_arith_overflow): Likewise.
19470 * ira-color.c (allocno_copy_cost_saving): Likewise.
19471 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19472 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19473 (process_invariant_for_inheritance): Likewise.
19474 * lra-eliminations.c (move_plus_up): Likewise.
19475 * omp-low.c (lower_oacc_reductions): Likewise.
19476 * simplify-rtx.c (simplify_subreg): Likewise.
19477 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19478 (TARGET_CHKP_BOUND_MODE): Likewise..
19479 * targhooks.c (default_chkp_bound_mode): Likewise.
19480 (default_setup_incoming_vararg_bounds): Likewise.
19481 * targhooks.h (default_chkp_bound_mode): Likewise.
19482 (default_setup_incoming_vararg_bounds): Likewise.
19483 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19484 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19485 (have_whole_vector_shift): Likewise.
19486 * tree-vect-stmts.c (vectorizable_load): Likewise.
19487 * doc/tm.texi: Regenerate.
19489 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19491 Graceful degrade if Binutils PR21472 is not available.
19494 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19495 .rodata in flash test fails.
19496 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19497 * confgure: Regenerate.
19498 * config.in: Regenerate.
19499 * config/avr/avr.c (avr_asm_named_section)
19500 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19501 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19502 (avr_asm_init_sections): Same.
19504 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19506 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19507 (fma<VH:mode>4_intrinsic): Likewise.
19508 (*fmsub<VCVTF:mode>4): Likewise.
19509 (*fmsub<VH:mode>4_intrinsic): Likewise.
19511 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19514 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19515 Don't depend on "optimize > 0".
19516 (out_movhi_r_mr, out_movqi_mr_r): Same.
19517 (out_movhi_mr_r, out_movqi_r_mr): Same.
19518 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19519 io_address_operand on "optimize > 0".
19521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19523 * tree-loop-distribution.c: Add general explanantion on the pass.
19524 (generate_loops_for_partition): Mark distributed loop.
19525 (pg_add_dependence_edges): New parameter. Handle alias data
19526 dependence specially and record it in the parameter if asked.
19527 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19528 (init_partition_graph_vertices, add_partition_graph_edge): New.
19529 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19530 (free_partition_graph_vdata, build_partition_graph): New.
19531 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19532 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19533 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19534 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19535 (version_for_distribution_p, finalize_partitions): New.
19536 (distribute_loop): Handle alias data dependence specially. Factor
19537 out loop fusion code as functions and call these functions.
19539 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19541 * tree-loop-distribution.c (classify_partition): New parameter and
19542 better handle reduction statement.
19543 (rdg_build_partitions): Revise comment.
19544 (distribute_loop): Compute statements in all partitions and pass it
19545 to classify_partition.
19547 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19549 * tree-loop-distribution.c (enum partition_type): New.
19550 (struct partition): New field type.
19551 (partition_merge_into): Add parameter. Update partition type.
19552 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19553 (build_rdg_partition_for_vertex): Compute partition type.
19554 (rdg_build_partitions): Dump partition type.
19555 (distribute_loop): Update calls to partition_merge_into.
19557 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19559 * tree-loop-distribution.c (struct ddr_hasher): New.
19560 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19562 (classify_partition): Call get_data_dependence.
19563 (pg_add_dependence_edges): Ditto.
19564 (distribute_loop): Release data dependence hash table.
19566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19568 * tree-loop-distribution.c (ref_base_address): Delete.
19569 (similar_memory_accesses): Rename ...
19570 (share_memory_accesses): ... to this. Check if partitions access
19571 the same memory reference.
19572 (distribute_loop): Call share_memory_accesses.
19574 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19576 * tree-loop-distribution.c (struct partition): New field recording
19577 its data reference.
19578 (partition_alloc, partition_free): Init and release data refs.
19579 (partition_merge_into): Merge data refs.
19580 (build_rdg_partition_for_vertex): Collect data refs for partition.
19581 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19583 (distribute_loop): Remve data refs from vertice data of partition
19586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19588 * tree-loop-distribution.c (params.h): Include header file.
19589 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19590 (datarefs_vec): New global var.
19591 (create_rdg_vertices): Use datarefs_vec directly.
19592 (free_rdg): Don't free data references.
19593 (build_rdg): Update use. Don't free data references.
19594 (distribute_loop): Compute global variable for data references.
19595 Bail out if there are too many data references.
19597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19599 * tree-loop-distribution.c (loop_nest): New global var.
19600 (build_rdg): Use loop directly, rather than loop nest.
19601 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19603 (distribute_loop): Compute global variable loop nest. Update use.
19605 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19607 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19608 (partition_merge_into): New parameter. Dump reason for fusion.
19609 (distribute_loop): Update use of partition_merge_into.
19611 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19613 * tree-loop-distribution.c (bb_top_order_index): New.
19614 (bb_top_order_index_size, bb_top_order_cmp): New.
19615 (stmts_from_loop): Use topological order.
19616 (pass_loop_distribution::execute): Compute and release topological
19617 order for basic blocks.
19619 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19621 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19624 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19626 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19627 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19628 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19629 * internal-fn.def (LOOP_DIST_ALIAS): New.
19630 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19631 (fold_loop_internal_call): ... this.
19632 (vect_loop_dist_alias_call): New function.
19633 (set_uid_loop_bbs): Call fold_loop_internal_call.
19634 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19637 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19640 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19641 Require dead FLAGS_REG at the beginning of a peephole.
19643 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19646 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19647 arguments in the call to __builtin_ia32_sbb_u32.
19648 (_subborrow_u64): Swap _X and _Y arguments in the call to
19649 __builtin_ia32_sbb_u64.
19651 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19654 * tree-vrp.c (compare_assert_loc): Turn into a function template
19655 with stable template parameter. Only test if a->e is NULL,
19656 !a->e == !b->e has been verified already. Use e == NULL or
19657 e != NULL instead of e or ! e tests. If stable is true, don't use
19658 iterative_hash_expr, on the other side allow a or b or both NULL
19659 and sort the NULLs last.
19660 (process_assert_insertions): Sort using compare_assert_loc<false>
19661 instead of compare_assert_loc, later sort using
19662 compare_assert_loc<true> before calling process_assert_insertions_for
19663 in a loop. Use break instead of continue once seen NULL pointer.
19665 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19667 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19668 Cortex-R7 and Cortex-R8 processors.
19670 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19672 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19673 uninitialized while src is not.
19675 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19677 * common/config/arm/arm-common.c: Adjust include path for
19679 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19680 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19681 (arm-cpu-data.h): Likewise.
19682 (arm-cpu-cdata.h): Likewise.
19683 * config/arm/arm-cpu.h: Delete.
19684 * config/arm/arm-cpu-cdata.h: Delete.
19685 * config/arm/arm-cpu-data.h: Delete.
19687 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19689 * config/arm/arm-cpus.in (cortex-a55): New.
19690 (cortex-a75): Likewise.
19691 (cortex-a75.cortex-a55): Likewise.
19692 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19694 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19695 * config/arm/arm-cpu-cdata.h: Regenerate.
19696 * config/arm/arm-cpu-data.h: Regenerate.
19697 * config/arm/arm-cpu.h: Regenerate.
19698 * config/arm/arm-tables.opt: Regenerate.
19699 * config/arm/arm-tune.md: Regenerate.
19701 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19703 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19705 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19707 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19710 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19712 PR tree-optimization/81292
19713 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19714 full_string_p, also call adjust_related_strinfos if the adjustment
19715 is simple, otherwise invalidate related strinfos.
19717 2017-07-04 Martin Liska <mliska@suse.cz>
19720 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19721 newly created variable as DECL_IGNORED_P.
19723 2017-07-04 Martin Liska <mliska@suse.cz>
19726 * ipa-inline.c (inline_small_functions):
19727 Use xstrdup_for_dump.
19729 2017-07-04 Tom de Vries <tom@codesourcery.com>
19731 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19733 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19736 * config/darwin.c (darwin_function_switched_text_sections):
19739 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19741 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19743 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19745 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19747 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19749 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19750 min_profitable_iters, and th as inclusive lower bounds.
19751 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19752 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19753 for min_profitable_iters and min_profitable_estimate.
19754 (vect_transform_loop): Treat th as an inclusive lower bound.
19755 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19757 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19760 * config/darwin.c (darwin_function_switched_text_sections):
19761 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19762 in two pieces, and suppress the use of buf.
19764 2017-07-03 Nathan Sidwell <nathan@acm.org>
19766 * hash-table.h (hash_table_mod1): Fix indentation.
19768 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19770 PR middle-end/81290
19771 * predict.c (force_edge_cold): Be more careful about propagation
19773 * profile-count.h (profile_probability::guessed,
19774 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19776 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19778 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19780 * doc/invoke.texi (rcpc architecture extension): Document it.
19782 2017-07-03 Richard Biener <rguenther@suse.de>
19784 PR tree-optimization/60510
19785 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19786 the scalar reduction PHI and use it.
19787 (vectorizable_reduction): Properly guard the single_defuse_cycle
19788 path for non-SLP reduction chains where we cannot use it.
19789 Rework reduc_def/index and vector type deduction. Rework
19790 vector operand gathering during reduction op code-gen.
19791 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19792 chains dissolve the chain and leave it to non-SLP reduction
19795 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19797 * tree-data-ref.h (dr_alignment): Declare.
19798 * tree-data-ref.c (dr_alignment): New function.
19799 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19800 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19802 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19804 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19806 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19807 and base_misalignment fields.
19808 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19809 * tree-data-ref.c: Include builtins.h.
19810 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19811 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19812 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19813 * tree-vect-data-refs.c: Include tree-cfg.h.
19814 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19815 fields instead of calculating an alignment here.
19816 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19817 innermost_loop_behavior fields.
19819 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19821 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19823 (DR_STEP_ALIGNMENT): New macro.
19824 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19825 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19826 (create_data_ref): Print it.
19827 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19828 to tell whether the step preserves vector (mis)alignment.
19829 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19830 Move the check for an integer step and generalise to all INTEGER_CST.
19831 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19832 Print the outer step alignment.
19834 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19836 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19837 with offset_alignment.
19838 (DR_ALIGNED_TO): Delete.
19839 (DR_OFFSET_ALIGNMENT): New macro.
19840 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19841 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19842 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19843 (create_data_ref): Likewise.
19844 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19845 (vect_analyze_data_refs): Likewise.
19846 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19847 creating dummy innermost behavior.
19849 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19851 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19852 with a "innermost_loop_behavior *" and refeence tree.
19853 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19854 (create_data_ref): Update call accordingly.
19855 * tree-predcom.c (find_looparound_phi): Likewise.
19857 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19859 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19860 fields with dr_wrt_vec_loop.
19861 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19862 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19863 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19864 (vect_dr_behavior): New function.
19865 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19866 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19867 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19868 track whether the step preserves the misalignment.
19869 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19870 Use vect_dr_behavior.
19871 (vect_setup_realignment): Update call accordingly.
19872 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19873 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19874 call to vect_create_addr_base_for_vector_ref.
19875 (vect_create_cond_for_align_checks): Likewise.
19876 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19877 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19878 (vect_recog_mask_conversion_pattern): Likewise.
19879 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19880 (new_stmt_vec_info): Remove redundant zeroing.
19882 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19884 * common/config/arm/arm-common.c (arm_be8_option): New function.
19885 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19886 (ISA_ARMv6): Add isa_bit_be8.
19887 * config/arm/arm.h (arm_be8_option): Add prototype.
19888 (BE8_SPEC_FUNCTION): New define.
19889 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19890 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19891 (mlittle-endian): Similarly.
19892 (mbe8, mbe32): New options.
19893 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19894 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19896 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19898 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19900 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19902 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19903 (cleanup_tree_cfg_bb): Use it.
19904 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19906 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19908 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19911 * loop-doloop.c (add_test): Update profile.
19913 2017-07-03 Martin Liska <mliska@suse.cz>
19916 * sanopt.c (rewrite_usage_of_param): New function.
19917 (sanitize_rewrite_addressable_params): Likewise.
19918 (pass_sanopt::execute): Call rewrite_usage_of_param.
19920 2017-07-03 Richard Biener <rguenther@suse.de>
19922 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19923 back to using VIEW_CONVERT_EXPR.
19925 2017-07-03 Martin Liska <mliska@suse.cz>
19928 * doc/extend.texi: Document when a resolver function is
19929 generated for target_clones.
19931 2017-07-03 Martin Liska <mliska@suse.cz>
19933 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19934 and shadow memory of auto variables which are subject of
19935 use-after-scope sanitization.
19936 (asan_expand_mark_ifn): Add do set only when is_poison.
19938 2016-07-03 Richard Biener <rguenther@suse.de>
19940 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19942 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19943 (vectorizable_reduction): Perform reduction PHI creation when
19944 visiting a reduction PHI and adjust and simplify code generation
19945 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19946 (vect_transform_loop): Visit reduction PHIs.
19947 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19948 defs into the SLP tree.
19949 (vect_build_slp_tree): Reduction defs terminate the recursion.
19950 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19952 (vect_get_vec_defs_for_stmt_copy): Export.
19953 (vect_get_vec_defs): Likewise.
19954 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19956 (vect_get_vec_defs_for_stmt_copy): Declare.
19957 (vect_get_vec_defs): Likewise.
19959 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19961 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19962 parameter with a "loop" parameter and use it instead of the
19963 loop containing DR_STMT. Don't check simple_iv when doing
19964 BB analysis. Describe the two analysis modes in the comment.
19966 2017-07-03 Tom de Vries <tom@codesourcery.com>
19968 PR tree-optimization/69468
19969 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19970 (find_same_succ_bb): Handle ignore_edge_flags.
19972 2017-07-03 Tom de Vries <tom@codesourcery.com>
19974 PR tree-optimization/81192
19975 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19977 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19979 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19981 2017-07-03 Tom de Vries <tom@codesourcery.com>
19983 PR tree-optimization/81192
19984 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19985 BB_SAME_SUCC (bb) == NULL.
19987 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19989 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19992 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19994 * dumpfile.c: Include profile-count.h
19995 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19997 (insert_cond_bb): Update profile.
19998 * tree-cfg.h (insert_cond_bb): Update prototype.
19999 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20000 * tree-dump.c: Do not include tree-cfg.
20002 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20004 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20006 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20008 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20011 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20013 * tree-complex.c (expand_complex_div_wide): update profile.
20015 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20016 Alan Hayward <alan.hayward@arm.com>
20017 David Sherwood <david.sherwood@arm.com>
20019 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20020 (CORETYPES_H): New define.
20021 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20022 (insn-modes-inline.h, s-modes-inline-h): New rules.
20023 (generated_files): Add insn-modes-inline.h.
20024 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20025 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20026 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20027 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20028 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20029 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20030 (build/gencodes.o, build/genconditions.o): Likewise.
20031 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20032 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20033 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20034 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20035 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20036 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20037 * coretypes.h: Include everything up to real.h for generators.
20038 Include insn-modes.h first. Include wide-int-print.h after
20039 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20040 * machmode.h: Don't include insn-modes.h here.
20041 * function-tests.c: Remove includes of signop.h, machmode.h,
20042 double-int.h and wide-int.h.
20044 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20046 * optc-save-gen.awk: Likewise.
20047 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20048 * godump.c: Remove include of wide-int-print.h.
20049 * pretty-print.h: Likewise.
20050 * wide-int-print.cc: Likewise.
20051 * wide-int.cc: Likewise.
20052 * hash-map-tests.c: Remove include of signop.h.
20053 * hash-set-tests.c: Likewise.
20054 * rtl-tests.c: Likewise.
20055 * mkconfig.sh: Remove include of machmode.h.
20056 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20058 (emit_insn_modes_inline_h): ...this new function. Emit the code
20059 into an insn-modes-inline.h header file, adding appropriate
20060 include guards and end comments.
20061 (emit_insn_modes_c_header): Remove include of machmode.h.
20062 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20064 (main): Handle -i flag and call emit_insn_modes_inline_h when
20067 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20069 * tree-ssa-strlen.c (strinfo): Rename the length field to
20070 nonzero_chars. Add a full_string_p field.
20071 (compare_nonzero_chars, zero_length_string_p): New functions.
20072 (get_addr_stridx): Add an offset_out parameter.
20073 Use compare_nonzero_chars.
20074 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20075 (new_strinfo): Update after above changes to strinfo.
20076 (set_endptr_and_length): Set full_string_p.
20077 (get_string_length): Update after above changes to strinfo.
20078 (unshare_strinfo): Update call to new_strinfo.
20079 (maybe_invalidate): Likewise.
20080 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20081 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20082 as a uhwi instead of an shwi. Update after above changes to
20083 strinfo and new_strinfo.
20084 (zero_length_string): Assert that chainsi contains full strings.
20085 Use zero_length_string_p. Update call to new_strinfo.
20086 (adjust_related_strinfos): Update after above changes to strinfo.
20087 Copy full_string_p from origsi.
20088 (adjust_last_stmt): Use zero_length_string_p.
20089 (handle_builtin_strlen): Update after above changes to strinfo and
20090 new_strinfo. Install the lhs as the string length if the previous
20091 entry didn't describe a full string.
20092 (handle_builtin_strchr): Update after above changes to strinfo
20094 (handle_builtin_strcpy): Likewise.
20095 (handle_builtin_strcat): Likewise.
20096 (handle_builtin_malloc): Likewise.
20097 (handle_pointer_plus): Likewise.
20098 (handle_builtin_memcpy): Likewise. Track nonzero characters
20099 that aren't necessarily followed by a nul terminator.
20100 (handle_char_store): Likewise.
20102 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20104 PR tree-optimization/80769
20105 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20106 for malloc and calloc. Document the new invariant that all related
20107 strinfos have delayed lengths or none do.
20108 (verify_related_strinfos): Move earlier in file.
20109 (set_endptr_and_length): New function, split out from...
20110 (get_string_length): ...here. Also set the lengths of related
20112 (zero_length_string): Assert that chainsi has known (rather than
20114 (adjust_related_strinfos): Likewise.
20116 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20118 PR tree-optimization/81136
20119 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20120 assert that two references with the same misalignment have the same
20121 compile-time misalignment if those compile-time misalignments
20124 2017-07-01 Andi Kleen <ak@linux.intel.com>
20126 * print-tree.c (print_node): Print all attributes.
20128 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20130 * cfg.c (scale_bbs_frequencies): New function.
20131 * cfg.h (scale_bbs_frequencies): Declare it.
20132 * cfgloopanal.c (single_likely_exit): Cleanup.
20133 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20135 (scale_loop_profile): Likewise.
20136 (loop_version): Likewise.
20137 (create_empty_loop_on_edge): Update.
20138 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20139 scale_loop_frequencies, scale_loop_profile, loopify,
20140 loop_version): Update prototypes.
20141 * modulo-sched.c (sms_schedule): Update.
20142 * predict.c (unlikely_executed_edge_p): Also check probability.
20143 (probably_never_executed_edge_p): Fix typo.
20144 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20145 * tree-parloops.c (gen_parallel_loop): Update.
20146 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20147 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20148 * tree-ssa-loop-split.c (split_loop): Update.
20149 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20150 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20151 (vect_loop_versioning): Update.
20152 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20154 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20156 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20158 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20160 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20161 to keep profile consistent.
20163 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20165 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20166 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20167 * profile-count.h (max_safe_multiplier): Make unsigned.
20168 (profile_count::guessed_zero): New.
20170 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20172 * bb-reorder.c (fix_up_crossing_landing_pad,
20173 fix_crossing_conditional_branches): Use make_single_succ_edge
20174 to keep profile consistent.
20176 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20178 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20181 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20184 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20185 the right scopes, make sure cond_jump isn't preserved between multiple
20186 iterations. Search for fallthru edge whenever there are 3+ edges and
20187 use find_fallthru_edge for it.
20189 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20191 Patch by Alexander Monakov <amonakov@ispras.ru>
20192 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20193 probabilities consistently.
20195 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20197 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20198 emit_cmp_and_jump_insns.
20200 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20203 * tree-inline.c (expand_call_inline): Combine profile statuses.
20205 2017-06-30 Andrew Pinski <apinski@cavium.com>
20207 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20208 fold_stmt returned true.
20210 2017-06-30 Nathan Sidwell <nathan@acm.org>
20212 * ggc.h (empty_string): Delete.
20213 * cfgexpand.c (expand_asm_stmt): Use plain "".
20214 * optabs.c (expand_asm_memory_barrier): Likewise.
20215 * stringpool.c (empty_string): Delete.
20216 (digit_vector, digit_string): Delete.
20217 (ggc_alloc_string): Use plain "", don't optimize single digit
20218 strings. Use ggc_alloc_atomic.
20220 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20222 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20223 comparison set and one other set, use the cost of the non-comparison
20226 2017-06-30 Nathan Sidwell <nathan@acm.org>
20228 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20231 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20233 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20234 loops. Remove now unneeded calls to gimple_switch_set_label() that
20235 just set removed labels to NULL_TREE.
20237 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20239 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20240 (set_range_info): ...here. Only call set_range_info_raw if domain
20242 (set_nonzero_bits): Call set_range_info_raw.
20243 * tree-ssanames.h (set_range_info_raw): New.
20245 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20248 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20249 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20250 of nonimmediate_operand and <store_mask_constraint> instead of m
20251 for the input operand. For V8FI iterator, always split if input
20252 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20253 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20254 <store_mask_predicate> instead of register_operand and
20255 <store_mask_constraint> instead of v for the input operand. Make
20256 sure both operands aren't MEMs for if not <mask_applied>.
20258 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20260 * lto-wrapper.c (copy_file) Close both file descriptors before
20263 2017-06-30 Martin Liska <mliska@suse.cz>
20266 * multiple_target.c (create_dispatcher_calls): Make ifunc
20267 also for function that don't have calls or are not referenced.
20269 2017-06-30 Richard Biener <rguenther@suse.de>
20271 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
20272 analyze the first scalar stmt. Move vector type computation
20273 for the BB case here from ...
20274 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
20275 live operation processing in the SLP case properly.
20277 2017-06-30 Richard Biener <rguenther@suse.de>
20279 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
20281 2017-06-30 Martin Liska <mliska@suse.cz>
20284 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
20285 before BUILT_IN_UNWIND_RESUME when ASAN is used.
20287 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
20289 * doc/invoke.texi (AArch64): Add missing options and remove redundant
20292 2017-06-30 Richard Biener <rguenther@suse.de>
20294 PR tree-optimization/81249
20295 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
20296 condition reduction result to original scalar type.
20298 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20300 * profile-count.h (enum profile_quality): Fix typos and whitespace
20303 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20305 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20306 type for branch probabilities.
20308 2017-06-29 Julian Brown <julian@codesourcery.com>
20309 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20311 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20312 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20313 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20314 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20316 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20318 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20319 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20320 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20321 CC usage from generic code to here.
20322 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20323 CC usage into the target macros.
20325 2017-06-29 Maya Rashish <coypu@sdf.org>
20327 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20330 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20332 * arm/arm-builtins.c: Include profile-count.h
20333 * except.c (sjlj_emit_function_enter): Use
20334 profile_probability::unlikely.
20336 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20338 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20339 and tocrel_offset be pointer args rather than implicitly using
20341 (legitimate_constant_pool_address_p, rs6000_emit_move,
20342 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20343 tocrel_offset and use in toc_relative_expr_p call.
20344 (print_operand, print_operand_address): Use static tocrel_base_oac
20345 and tocrel_offset_oac.
20346 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20349 2017-06-29 Maya Rashish <coypu@sdf.org>
20351 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20353 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20355 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20356 objects, take into account only the alignment of 'op0' and 'mode1' if
20359 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20361 * ccmp.c (ccmp_tree_comparison_p): New function.
20362 (ccmp_candidate_p): Update to use above function.
20363 (get_compare_parts): New function.
20364 (expand_ccmp_next): Update to use new functions.
20365 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20367 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20368 take mode as argument.
20369 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20370 * expr.c (expand_expr_real_1): Pass mode as argument.
20372 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20374 * combine.c (combine_instructions): Print insns to dump_file, together
20377 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20379 * asan.c (asan_emit_stack_protection): Update.
20380 (create_cond_insert_point): Update.
20381 * auto-profile.c (afdo_propagate_circuit): Update.
20382 * basic-block.h (struct edge_def): Turn probability to
20383 profile_probability.
20384 (EDGE_FREQUENCY): Update.
20385 * bb-reorder.c (find_traces_1_round): Update.
20386 (better_edge_p): Update.
20387 (sanitize_hot_paths): Update.
20388 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20389 (make_single_succ_edge): Update.
20390 (check_bb_profile): Update.
20391 (dump_edge_info): Update.
20392 (update_bb_profile_for_threading): Update.
20393 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20394 probabilitycount to 0.
20395 * cfgbuild.c (compute_outgoing_frequencies): Update.
20396 * cfgcleanup.c (try_forward_edges): Update.
20397 (outgoing_edges_match): Update.
20398 (try_crossjump_to_edge): Update.
20399 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20400 (expand_gimple_tailcall): Update.
20401 (construct_init_block): Use make_single_succ_edge.
20402 (construct_exit_block): Use make_single_succ_edge.
20403 * cfghooks.c (verify_flow_info): Update.
20404 (redirect_edge_succ_nodup): Update.
20405 (split_edge): Update.
20406 (account_profile_record): Update.
20407 * cfgloopanal.c (single_likely_exit): Update.
20408 * cfgloopmanip.c (scale_loop_profile): Update.
20409 (set_zero_probability): Remove.
20410 (duplicate_loop_to_header_edge): Update.
20411 * cfgloopmanip.h (loop_version): Update prototype.
20412 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20413 (force_nonfallthru_and_redirect): Update.
20414 (update_br_prob_note): Update.
20415 (rtl_verify_edges): Update.
20416 (purge_dead_edges): Update.
20417 (rtl_lv_add_condition_to_bb): Update.
20418 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20419 * cgraphunit.c (init_lowered_empty_function): Update.
20420 (cgraph_node::expand_thunk): Update.
20421 * cilk-common.c: Include profile-count.h
20422 * dojump.c (inv): Remove.
20423 (jumpifnot): Update.
20424 (jumpifnot_1): Update.
20425 (do_jump_1): Update.
20427 (do_jump_by_parts_greater_rtx): Update.
20428 (do_compare_rtx_and_jump): Update.
20429 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20430 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20431 * dwarf2cfi.c: Include profile-count.h
20432 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20433 (sjlj_emit_dispatch_table): Likewise.
20434 * explow.c: Include profile-count.h
20435 * expmed.c (emit_store_flag_force): Update.
20436 (do_cmp_and_jump): Update.
20437 * expr.c (compare_by_pieces_d::generate): Update.
20438 (compare_by_pieces_d::finish_mode): Update.
20439 (emit_block_move_via_loop): Update.
20440 (store_expr_with_bounds): Update.
20441 (store_constructor): Update.
20442 (expand_expr_real_2): Update.
20443 (expand_expr_real_1): Update.
20444 * expr.h (try_casesi, try_tablejump): Update prototypes.
20445 * gimple-pretty-print.c (dump_probability): Update.
20446 (dump_profile): New.
20447 (dump_gimple_label): Update.
20448 (dump_gimple_bb_header): Update.
20449 * graph.c (draw_cfg_node_succ_edges): Update.
20450 * hsa-gen.c (convert_switch_statements): Update.
20451 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20452 (find_if_case_1): Update.
20453 (find_if_case_2): Update.
20454 * internal-fn.c (expand_arith_overflow_result_store): Update.
20455 (expand_addsub_overflow): Update.
20456 (expand_neg_overflow): Update.
20457 (expand_mul_overflow): Update.
20458 (expand_vector_ubsan_overflow): Update.
20459 * ipa-cp.c (good_cloning_opportunity_p): Update.
20460 * ipa-split.c (split_function): Use make_single_succ_edge.
20461 * ipa-utils.c (ipa_merge_profiles): Update.
20462 * loop-doloop.c (add_test): Update.
20463 (doloop_modify): Update.
20464 * loop-unroll.c (compare_and_jump_seq): Update.
20465 (unroll_loop_runtime_iterations): Update.
20466 * lra-constraints.c (lra_inheritance): Update.
20467 * lto-streamer-in.c (input_cfg): Update.
20468 * lto-streamer-out.c (output_cfg): Update.
20469 * mcf.c (adjust_cfg_counts): Update.
20470 * modulo-sched.c (sms_schedule): Update.
20471 * omp-expand.c (expand_omp_for_init_counts): Update.
20472 (extract_omp_for_update_vars): Update.
20473 (expand_omp_ordered_sink): Update.
20474 (expand_omp_for_ordered_loops): Update.
20475 (expand_omp_for_generic): Update.
20476 (expand_omp_for_static_nochunk): Update.
20477 (expand_omp_for_static_chunk): Update.
20478 (expand_cilk_for): Update.
20479 (expand_omp_simd): Update.
20480 (expand_omp_taskloop_for_outer): Update.
20481 (expand_omp_taskloop_for_inner): Update.
20482 * omp-simd-clone.c (simd_clone_adjust): Update.
20483 * optabs.c (expand_doubleword_shift): Update.
20484 (expand_abs): Update.
20485 (emit_cmp_and_jump_insn_1): Update.
20486 (expand_compare_and_swap_loop): Update.
20487 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20488 * predict.c (predictable_edge_p): Update.
20489 (edge_probability_reliable_p): Update.
20490 (set_even_probabilities): Update.
20491 (combine_predictions_for_insn): Update.
20492 (combine_predictions_for_bb): Update.
20493 (propagate_freq): Update.
20494 (estimate_bb_frequencies): Update.
20495 (force_edge_cold): Update.
20496 * profile-count.c (profile_count::dump): Add missing space into dump.
20497 (profile_count::debug): Add newline.
20498 (profile_count::differs_from_p): Explicitly convert to unsigned.
20499 (profile_count::stream_in): Update.
20500 (profile_probability::dump): New member function.
20501 (profile_probability::debug): New member function.
20502 (profile_probability::differs_from_p): New member function.
20503 (profile_probability::differs_lot_from_p): New member function.
20504 (profile_probability::stream_in): New member function.
20505 (profile_probability::stream_out): New member function.
20506 * profile-count.h (profile_count_quality): Rename to ...
20507 (profile_quality): ... this one.
20508 (profile_probability): New.
20509 (profile_count): Update.
20510 * profile.c (compute_branch_probabilities): Update.
20511 * recog.c (peep2_attempt): Update.
20512 * sched-ebb.c (schedule_ebbs): Update.
20513 * sched-rgn.c (find_single_block_region): Update.
20514 (compute_dom_prob_ps): Update.
20515 (schedule_region): Update.
20516 * sel-sched-ir.c (compute_succs_info): Update.
20517 * stmt.c (struct case_node): Update.
20518 (do_jump_if_equal): Update.
20519 (get_outgoing_edge_probs): Update.
20520 (conditional_probability): Update.
20521 (emit_case_dispatch_table): Update.
20522 (expand_case): Update.
20523 (expand_sjlj_dispatch_table): Update.
20524 (emit_case_nodes): Update.
20525 * targhooks.c: Update.
20526 * tracer.c (better_p): Update.
20527 (find_best_successor): Update.
20528 * trans-mem.c (expand_transaction): Update.
20529 * tree-call-cdce.c: Update.
20530 * tree-cfg.c (gimple_split_edge): Upate.
20531 (move_sese_region_to_fn): Upate.
20532 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20533 * tree-eh.c (lower_resx): Upate.
20534 (cleanup_empty_eh_move_lp): Upate.
20535 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20536 * tree-inline.c (copy_edges_for_bb): Update.
20537 (copy_cfg_body): Update.
20538 * tree-parloops.c (gen_parallel_loop): Update.
20539 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20540 (gimple_gen_time_profiler): Update.
20541 * tree-ssa-dce.c (remove_dead_stmt): Update.
20542 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20543 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20544 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20545 (unloop_loops): Update.
20546 (try_peel_loop): Update.
20547 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20548 * tree-ssa-loop-split.c (connect_loops): Update.
20549 (split_loop): Update.
20550 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20551 (hoist_guard): Update.
20552 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20553 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20554 (value_replacement): Update.
20555 * tree-ssa-reassoc.c (branch_fixup): Update.
20556 * tree-ssa-tail-merge.c (replace_block_by): Update.
20557 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20558 (create_edge_and_update_destination_phis): Update.
20559 (compute_path_counts): Update.
20560 (recompute_probabilities): Update.
20561 (update_joiner_offpath_counts): Update.
20562 (freqs_to_counts_path): Update.
20563 (duplicate_thread_path): Update.
20564 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20565 (struct switch_conv_info): Update.
20566 (gen_inbound_check): Update.
20567 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20568 (vect_do_peeling): Update.
20569 (vect_loop_versioning): Update.
20570 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20571 (optimize_mask_stores): Update.
20572 * ubsan.c (ubsan_expand_null_ifn): Update.
20573 * value-prof.c (gimple_divmod_fixed_value): Update.
20574 (gimple_divmod_fixed_value_transform): Update.
20575 (gimple_mod_pow2): Update.
20576 (gimple_mod_pow2_value_transform): Update.
20577 (gimple_mod_subtract): Update.
20578 (gimple_mod_subtract_transform): Update.
20579 (gimple_ic): Update.
20580 (gimple_stringop_fixed_value): Update.
20581 (gimple_stringops_transform): Update.
20582 * value-prof.h: Update.
20584 2017-06-29 Carl Love <cel@us.ibm.com>
20586 * config/rs6000/rs6000-c.c: Add support for built-in functions
20587 vector signed int vec_signed (vector float);
20588 vector signed long long vec_signed (vector double);
20589 vector signed int vec_signed2 (vector double, vector double);
20590 vector signed int vec_signede (vector double);
20591 vector signed int vec_signedo (vector double);
20592 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20593 instruction generator.
20594 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20595 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20596 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20598 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20599 vunsignede_v2df): Add define_expands.
20600 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20601 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20602 VEC_UNSIGNEDO): Add definitions.
20603 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20604 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20605 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20606 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20607 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20608 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20609 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20610 vec_unsignede, vec_unsignedo): Add builtin defines.
20611 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20613 * doc/extend.texi: Update the built-in documentation file for the
20614 new built-in functions.
20616 2017-06-29 Richard Biener <rguenther@suse.de>
20618 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20619 reduction chains to LOOP_VINFO_REDUCTIONS.
20620 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20621 SLP reductions after processing reduction chains.
20623 2017-06-29 Nathan Sidwell <nathan@acm.org>
20625 * builtins.c (fold_builtin_FUNCTION): Use
20626 lang_hooks.decl_printable_name.
20628 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20630 PR middle-end/81194
20631 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20632 with only one label.
20633 * stmt.c (expand_case): Assert NCASES is greater than one.
20635 2017-06-29 Richard Biener <rguenther@suse.de>
20637 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20639 (group_case_labels): Likewise.
20640 (find_taken_edge): Push sanity checking on val to workers...
20641 (find_taken_edge_cond_expr): ... here
20642 (find_taken_edge_switch_expr): ... and here, handle cases
20643 with just a default label.
20644 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20645 (group_case_labels): Likewise.
20646 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20647 group_case_labels does anything cleanup the CFG again.
20649 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20651 PR tree-optimization/81196
20652 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20653 exit condition comparing two IVs.
20655 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20657 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20658 profile to the dummy entry at the end of the list of architectures.
20659 * config/arm/arm-cpu-cdata.h: Regenerated.
20661 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20662 Michael Collison <michael.collison@arm.com>
20665 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20667 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20668 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20669 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20670 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20671 of shift when the shift amount is masked with constant equal to
20672 the size of the mode.
20673 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20676 2017-06-29 Martin Liska <mliska@suse.cz>
20678 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20680 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20683 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20686 (edge_growth_cache_entry::edge_growth_cache_entry): New
20688 (reset_edge_growth_cache): Update to use constructor.
20690 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20692 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20693 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20694 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20696 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20698 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20699 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20701 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20703 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20704 (*-linux-uclibc*): Add t-uclibc tmake_file.
20705 * config/t-musl: New.
20706 * config/t-uclibc: New.
20708 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20710 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20712 (gen_comm_data): Emit architectural setting of arch_prof.
20713 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20715 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20716 (armv8-m.base, armv8-m.main): Likewise.
20717 * arm-protos.h (arm_build_target): Add profile field.
20718 (arch_option): Likewise.
20719 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20721 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20722 arm_active_target.profile.
20724 2017-06-28 Richard Biener <rguenther@suse.de>
20726 PR middle-end/81227
20727 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20728 TYPE_OVERFLOW_WRAPS.
20729 * match.pd (negate_expr_p): Likewise.
20730 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20731 fold_build2, not fold_binary.
20733 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20735 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20736 Convert memory address to Pmode.
20737 (aarch64_print_operand): Assert MEM operands are always Pmode.
20739 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20742 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20743 Remove redundant if.
20744 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20745 * config/arm/aarch-common-protos.h
20746 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20747 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20749 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20752 * multiple_target.c (create_dispatcher_calls): Set the default
20753 clone to be static, not public.
20755 2017-06-28 Richard Biener <rguenther@suse.de>
20757 * tree-vect-loop.c (vectorizable_reduction): Move special
20758 cond reduction IV var creation ...
20759 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20760 parameter. Use STMT_VINFO_VECTYPE.
20761 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20764 2017-06-28 Martin Liska <mliska@suse.cz>
20767 * ipa-visibility.c (non_local_p): Handle visibility.
20769 2017-06-28 Martin Liska <mliska@suse.cz>
20772 * common.opt: Add IntegerRange to various options.
20773 * opt-functions.awk (integer_range_info): New function.
20774 * optc-gen.awk: Add integer_range_info to cl_options struct.
20775 * opts-common.c (decode_cmdline_option): Handle
20776 CL_ERR_INT_RANGE_ARG.
20777 (cmdline_handle_error): Likewise.
20778 * opts.c (print_filtered_help): Show valid interval in
20779 when --help is provided.
20780 * opts.h (struct cl_option): Add range_min and range_max fields.
20781 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20783 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20785 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20786 (x * C EQ/NE y * C): New transformation.
20788 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20790 * genmultilib (combination_space): Accept '+' in option names.
20792 2017-06-28 Martin Liska <mliska@suse.cz>
20795 * asan.c (instrument_derefs): Bail out inner references
20796 that are hard register variables.
20798 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20801 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20802 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20804 2017-06-28 Richard Biener <rguenther@suse.de>
20806 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20807 (vect_get_slp_defs): Adjust.
20808 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20810 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20812 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20813 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20814 (vectorizable_reduction): Adjust.
20815 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20817 (vect_get_slp_defs): Likewise.
20818 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20819 (vectorizable_bswap): Adjust.
20820 (vectorizable_call): Likewise.
20821 (vectorizable_conversion): Likewise.
20822 (vectorizable_assignment): Likewise.
20823 (vectorizable_shift): Likewise.
20824 (vectorizable_operation): Likewise.
20825 (vectorizable_store): Likewise.
20826 (vectorizable_condition): Likewise.
20827 (vectorizable_comparison): Likewise.
20829 2017-06-28 Michael Collison <michael.collison@arm.com>
20832 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20834 (arm_gen_movmemqi): Removed unused variable 'i'.
20835 Convert 'for' loop into 'while' loop.
20836 (arm_expand_prologue): Remove last unnecessary set of insn.
20837 (thumb_pop): Remove unused variable 'pushed_words'.
20838 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20840 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20842 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20843 * config/s390/s390-protos.h: Add prototype of
20844 s390_rel_address_ok_p.
20845 * config/s390/s390.c (s390_got_symbol): New function.
20846 (s390_rel_address_ok_p): New function.
20847 (legitimize_pic_address): Use s390_rel_address_ok_p.
20848 (s390_load_got): Use s390_got_symbol.
20849 (s390_option_override): Issue error if
20850 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20851 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20853 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20855 2017-06-27 Andrew Pinski <apinski@cavium.com>
20857 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20858 (X * copysign (1.0, X)): New pattern.
20859 (X * copysign (1.0, -X)): New pattern.
20860 (copysign (-1.0, CST)): New pattern.
20862 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20864 * genmultilib (combination_space): Remove variable.
20865 Validate reuse rules against regular expression for any sequence
20866 of multilib options in any order.
20868 2017-06-27 Michael Collison <michael.collison@arm.com>
20870 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20871 call aarch64_split_simd_combine.
20872 * (aarch64_combine_internal<mode>): Delete pattern.
20873 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20874 Allow register and subreg operands.
20876 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20878 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20879 specific need, just fallback on defaults.
20880 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20882 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20883 Olivier Hainque <hainque@adacore.com>
20885 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20887 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20888 targets. Pick a default if no particular attempt applied.
20889 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20892 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20894 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20895 (x86_64-wrs-vxworks7): Likewise.
20897 2017-06-27 Marek Polacek <polacek@redhat.com>
20900 * ubsan.c (instrument_null): Check get_base_address's result for null.
20902 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20904 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20906 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20908 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20909 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20910 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20911 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20912 New function types.
20913 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20914 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20915 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20916 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20917 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20918 BUILT_IN_FEUPDATEENV): New builtins.
20919 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20920 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20921 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20922 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20924 (builtin_structptr_types): Adjust size.
20925 * tree.c (builtin_structptr_types): Add four entries.
20927 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20928 Olivier Hainque <hainque@adacore.com>
20930 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20931 (TLS_SYM): New local macro, forcing reference to __tls__ on
20932 link command lines for VxWorks 7 RTPs, triggering initialization
20934 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20935 OS features TLS support, true for RTPs on VxWorks 7.
20936 * config/vxworks.c (vxworks_override_options): Setup emutls
20939 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20941 * predict.c (test_prediction_value_range): Use -1U instead of -1
20942 to avoid narrowing conversion warning.
20943 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20944 to avoid narrowing conversion warning.
20945 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20947 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20949 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20951 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20952 64bit configurations.
20953 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20954 (SIZE_TYPE): Likewise.
20955 * config/vxworks.c (vxworks_emutls_var_fields): Use
20956 long_unsigned_type_node instead of unsigned_type_node as the offset
20957 field type, which is "pointer" mode in emutls.c.
20959 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20962 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20964 PR middle-end/81207
20965 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20966 gimple_vuse copying separately from gimple_vdef copying.
20968 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20970 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20973 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20974 Olivier Hainque <hainque@adacore.com>
20976 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20977 for all vxworks7 targets.
20978 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20979 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20980 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20981 variations for VX6/VX7 and 32/64bits later on in ...
20982 (VXWORKS_LIB_SPEC): Leverage new macros.
20983 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20984 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20986 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20988 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20989 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20991 2017-06-26 Carl Love <cel@us.ibm.com>
20993 * config/rs6000/rs6000-c.c: Add support for built-in functions
20994 vector bool char vec_reve (vector bool char);
20995 vector signed char vec_reve (vector signed char);
20996 vector unsigned char vec_reve (vector unsigned char);
20997 vector bool int vec_reve (vector bool int);
20998 vector signed int vec_reve (vector signed int);
20999 vector unsigned int vec_reve (vector unsigned int);
21000 vector bool long long vec_reve (vector bool long long);
21001 vector signed long long vec_reve (vector signed long long);
21002 vector unsigned long long vec_reve (vector unsigned long long);
21003 vector bool short vec_reve (vector bool short);
21004 vector signed short vec_reve (vector signed short);
21005 vector double vec_reve (vector double);
21006 vector float vec_reve (vector float);
21007 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21008 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21009 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21010 (altivec_vreve): New pattern.
21011 * config/rs6000/altivec.h (vec_reve): New define.
21012 * doc/extend.texi (vec_rev): Update the built-in documentation file
21013 for the new built-in functions.
21015 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21017 PR tree-optimization/71815
21018 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21020 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21022 (slsr_process_phi): Likewise.
21023 (replace_uncond_cands_and_profitable_phis): Don't replace a
21024 multiply candidate with a stride of 1 (copy or cast).
21025 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21027 (lowest_cost_path): Likewise.
21028 (total_savings): Likewise.
21030 2017-06-26 Richard Biener <rguenther@suse.de>
21033 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21034 Use def_builtin_pure for all gather builtins.
21036 2017-06-26 Richard Biener <rguenther@suse.de>
21038 PR tree-optimization/81203
21039 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21040 non-dominating BBs.
21042 2017-06-26 Marek Polacek <polacek@redhat.com>
21045 * doc/invoke.texi: Document -Wmultistatement-macros.
21047 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21049 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21050 arm_neon_ok_no_float_abi effective target.
21052 2017-06-26 Richard Biener <rguenther@suse.de>
21054 PR tree-optimization/80928
21055 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21056 (copy_bbs): Set BB_DUPLICATED flag early.
21057 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21059 (execute_on_shrinking_pred): Likewise.
21060 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21061 BB_DUPLICATED blocks.
21062 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21063 iterate over all PHIs considering removal of *gsi.
21065 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21067 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21070 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21072 * config/rs6000/rs6000-string.c: (expand_block_clear,
21073 do_load_for_compare, select_block_compare_mode,
21074 compute_current_alignment, expand_block_compare,
21075 expand_strncmp_align_check, expand_strn_compare,
21076 expand_block_move, rs6000_output_load_multiple)
21077 Move functions related to string/block move/compare
21078 to a separate file.
21079 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21080 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21081 for this function which is now used in two files.
21082 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21083 * config.gcc: Add rs6000-string.o to extra_objs for
21084 targets powerpc*-*-* and rs6000*-*-*.
21086 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21089 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21090 32-bit, since indexed is not valid for DImode.
21091 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21092 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21093 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21094 (define_peephole2 for Altivec d-form store): Likewise.
21097 * multiple_target.c (create_dispatcher_calls): Only create the
21098 dispatcher call if the function is the default clone of a
21099 versioned function.
21101 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21103 PR middle-end/80902
21104 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21105 a call, force the call to not be a tail call.
21107 2017-06-23 Jeff Law <law@redhat.com>
21109 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21112 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21114 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21115 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21116 handling for early expansion of vector shifts (sl,sr,sra,rl).
21117 (builtin_function_type): Add vector shift right instructions
21118 to the unsigned argument list.
21120 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21122 rtl-optimizatoin/79286
21123 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21124 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21125 trap. PIC register plus a const unspec without offset can never trap.
21127 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21129 * tree.h (builtin_structptr_type): New type.
21130 (builtin_structptr_types): Declare new array.
21131 * tree.c (builtin_structptr_types): New array.
21132 (free_lang_data, build_common_tree_nodes): Use it.
21134 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21137 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21140 2017-06-22 Matt Turner <mattst88@gmail.com>
21142 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21143 Lake models to skylake case. Assume skylake for unknown
21144 models with clflushopt.
21146 2017-06-22 Jeff Law <law@redhat.com>
21148 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21149 frame sizes that do not satisfy aarch64_uimm12_shift.
21151 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21153 * profile-count.h (apply_probability,
21154 apply_scale, probability_in): Fix checks for zero.
21156 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21158 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21159 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21161 2016-06-22 Richard Biener <rguenther@suse.de>
21163 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21164 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21165 REDUC_MAX_EXPR support.
21166 (vectorizable_reduction): Likewise.
21167 (vect_create_epilog_for_reduction): Likewise.
21169 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21171 * match.pd (A / (1 << B) -> A >> B): New.
21172 * generic-match-head.c: Include optabs-tree.h.
21173 * gimple-match-head.c: Likewise.
21174 * optabs-tree.h (target_supports_op_p): New.
21175 * optabs-tree.c (target_supports_op_p): New.
21177 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21179 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21180 $gcc_cv_ld --help output.
21181 (gcc_cv_ld_demangle): Likewise.
21182 (gcc_cv_ld_eh_frame_hdr): Likewise.
21183 (gcc_cv_ld_pie): Likewise.
21184 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21185 (gcc_cv_ld_buildid): Likewise.
21186 (gcc_cv_ld_sysroot): Likewise.
21187 (ld_bndplt_support): Likewise.
21188 (ld_pushpopstate_support): Likewise.
21189 * configure: Regenerate.
21190 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21192 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21195 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21196 operands indexes to avoid gap between operands and match_dups.
21198 2017-06-21 Andrew Pinski <apinski@cavium.com>
21200 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21201 Increment Arith_shift and Arith_shift_reg by 1.
21202 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21204 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21205 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21206 (aarch64_strip_extend): Add new argument and test for it.
21207 (aarch64_cheap_mult_shift_p): New function.
21208 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21209 add a cost if it is true.
21210 Update calls to aarch64_strip_extend.
21211 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21213 2017-06-21 Andrew Pinski <apinski@cavium.com>
21215 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21217 (thunderxt88): Likewise.
21218 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21219 (thunderx_prefetch_tune): New variable.
21220 (thunderx2t99_prefetch_tune): Update for the correct values.
21221 (thunderxt88_tunings): New variable.
21222 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21223 generic_prefetch_tune.
21224 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21226 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21228 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21229 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21230 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21231 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21232 (aarch64_atomic_cas<mode>, GPI): Likewise.
21234 2017-06-21 Martin Liska <mliska@suse.cz>
21236 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21237 statements on cold and hot labels.
21238 * predict.c (tree_estimate_probability_bb): Remove the
21239 prediction from this place.
21241 2017-06-21 Martin Liska <mliska@suse.cz>
21243 PR tree-optimization/79489
21244 * gimplify.c (maybe_add_early_return_predict_stmt): New
21246 (gimplify_return_expr): Call the function.
21247 * predict.c (tree_estimate_probability_bb): Remove handling
21249 * predict.def: Update comment about early return predictor.
21250 * gimple-predict.h (is_gimple_predict): New function.
21251 * predict.def: Change default value of early return to 66.
21252 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21254 * passes.def: Put pass_strip_predict_hints to the beginning of
21257 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21259 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21260 FUNCTION_DECL declarations.
21261 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21263 (dwaf2out_decl): Likewise.
21264 * godump.c (go_early_global_decl): Skip call to the real debug hook
21265 for FUNCTION_DECL declarations.
21266 * passes.c (rest_of_decl_compilation): Skip call to the
21267 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21268 -fdump-go-spec is passed.
21270 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21272 * config/i386/i386.c (struct builtin_isa): New field pure_p.
21273 Reorder for compactness.
21274 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
21275 (def_builtin_pure, def_builtin_pure2): New functions.
21276 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
21278 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21280 * match.pd (nop_convert): New predicate.
21281 ((A +- CST1) +- CST2): Allow some NOP conversions.
21283 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21286 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
21287 with ctors/dtors if GOVD_SHARED is set.
21289 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21291 * config/aarch64/aarch64.md (movti_aarch64):
21292 Emit mov rather than orr.
21293 (movtf_aarch64): Likewise.
21294 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
21295 Emit mov rather than orr.
21297 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21299 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
21300 Swap alternatives, make integer dup more expensive.
21302 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21304 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21305 Return true for non-tls symbols.
21307 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21309 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21310 (cortex-a75): Likewise.
21311 (cortex-a75.cortex-a55): Likewise.
21312 * config/aarch64/aarch64-tune.md: Regenerate.
21313 * doc/invoke.texi (-mtune): Document new values for -mtune.
21315 2017-06-21 Tom de Vries <tom@codesourcery.com>
21317 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21318 stack_size feature.
21319 (Effective-Target Keywords, Other attributes): Suggest using
21320 dg-add-options stack_size feature to get stack limit in stack_size
21321 effective target documentation.
21323 2017-06-21 Julian Brown <julian@codesourcery.com>
21324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21326 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21327 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21328 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21330 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21331 attribute type list for neon_multiply.
21332 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21333 attribute type list for neon_multiply.
21334 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21335 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21336 attribute type list for neon_multiply.
21337 * config/arm/types.md (crypto_pmull): Add.
21338 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21339 attribute type list.
21341 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21343 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21346 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21348 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21349 to make sure not to dereference a NULL cost_classes_ptr pointer.
21351 2017-06-20 Carl Love <cel@us.ibm.com>
21353 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21354 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21355 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21356 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21357 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21358 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21359 VMULOSW): New enum "unspec" values.
21360 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21361 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21362 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21363 altivec_vmulosw): New patterns.
21364 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21365 VMULOSW): Add definitions.
21367 2017-06-20 Julia Koval <julia.koval@intel.com>
21369 * config/i386/i386.c: Fix rounding expand for new pattern.
21370 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21372 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21374 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21375 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21377 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21379 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21382 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21384 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21385 architecture, then by implementer ID.
21386 * config/aarch64/aarch64-tune.md: Regenerate.
21388 2017-06-20 Richard Biener <rguenther@suse.de>
21390 PR middle-end/81097
21391 * fold-const.c (split_tree): Fold to type before negating.
21393 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21395 * diagnostic-show-locus.c
21396 (selftest::test_fixit_deletion_affecting_newline): New function.
21397 (selftest::diagnostic_show_locus_c_tests): Call it.
21399 2017-06-20 Andreas Schwab <schwab@suse.de>
21402 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21405 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21407 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21408 __ARM_FEATURE_COPROC according to support.
21410 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21412 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21413 Rewritten to avoid overflow for > 32-bit pointers.
21416 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21417 by removing enum keyword.
21418 (ubsan_type_descriptor): Likewise. Formatting fix.
21421 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21422 splitter): Require TARGET_SSE2 in the condition.
21424 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21427 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21428 for doing vector set of SFmode on ISA 3.0.
21429 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21430 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21432 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21433 SFmode value into a V4SF variable that was extracted from another
21434 V4SF variable without converting the element to double precision
21435 and back to single precision vector format.
21436 (vsx_insert_extract_v4sf_p9_2): Likewise.
21438 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21440 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21441 in UWHI to avoid undefined overflow.
21444 * ubsan.h (enum ubsan_encode_value_phase): New.
21445 (ubsan_encode_value): Change second argument to
21446 enum ubsan_encode_value_phase with default value of
21447 UBSAN_ENCODE_VALUE_GENERIC.
21448 * ubsan.c (ubsan_encode_value): Change second argument to
21449 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21450 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21451 create_tmp_var_raw instead of create_tmp_var and use a
21453 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21454 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21455 ubsan_encode_value callers.
21458 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21459 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21460 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21462 2017-06-19 Richard Biener <rguenther@suse.de>
21464 PR middle-end/81118
21465 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21466 estimates if we changed anything.
21468 2017-06-19 Richard Biener <rguenther@suse.de>
21470 PR tree-optimization/80887
21471 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21472 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21473 simplified lookups, then reset mprts_hook.
21474 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21476 (try_to_simplify): Likewise.
21478 2017-06-19 Martin Liska <mliska@suse.cz>
21481 * gimplify.c (gimplify_switch_expr):
21482 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21484 2017-06-19 Martin Liska <mliska@suse.cz>
21486 * doc/install.texi: Document that PGO runs in 4 stages.
21488 2017-06-19 Martin Liska <mliska@suse.cz>
21491 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21492 to dispatcher function name.
21493 * multiple_target.c (replace_function_decl): New function.
21494 (create_dispatcher_calls): Redirect both edges and references.
21496 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21498 * profile-count.c (profile_count::dump): Dump quality.
21499 (profile_count::differs_from_p): Update for unsigned val.
21500 * profile-count.h (profile_count_quality): New enum.
21501 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21503 2017-06-19 Richard Biener <rguenther@suse.de>
21505 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21506 struct function as arg.
21507 (estimate_numbers_of_iterations): Export overload with loop arg.
21508 (free_numbers_of_iterations_estimates_loop): Use an overload of
21509 free_numbers_of_iterations_estimates instead.
21510 * tree-cfg.c (remove_bb): Adjust.
21511 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21512 * tree-parloops.c (gen_parallel_loop): Likewise.
21513 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21515 (tree_unroll_loops_completely): Likewise.
21516 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21517 Use an overload instead and export.
21518 (estimated_loop_iterations): Adjust.
21519 (max_loop_iterations): Likewise.
21520 (likely_max_loop_iterations): Likewise.
21521 (estimate_numbers_of_iterations): Take struct function as arg
21523 (loop_exits_before_overflow): Adjust.
21524 (free_numbers_of_iterations_estimates_loop): Use an overload.
21525 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21526 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21528 2017-06-19 Richard Biener <rguenther@suse.de>
21531 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21532 RANGE_EXPR conservatively.
21534 2017-06-16 Carl Love <cel@us.ibm.com>
21536 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21537 definitions for vec_float, vec_float2, vec_floato,
21538 vec_floate built-ins.
21539 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21540 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21542 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21543 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21544 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21545 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21546 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21547 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21548 vec_floato): Add builtin defines.
21549 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21550 Update the built-in documentation file for the new built-in
21553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21555 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21556 (mthumb): Mark as the negative of -marm.
21558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21560 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21562 (ARM Options, -mtune): Document that this accepts the same
21563 extension options as -mcpu.
21564 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21568 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21569 permitted extensions.
21571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21573 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21574 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21575 (armv8-m.main): Add option +nodsp.
21576 * config/arm/arm-cpu-cdata.h: Regenerated.
21578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21580 * config/arm/t-fuchsia: New file.
21581 * config.gcc (arm*-*-fuchsia*): Use it.
21583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21585 * config/arm/t-symbian: Rewrite for new option infrastructure.
21587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21589 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21590 (MULTILIB_REQUIRED): Likewise.
21592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21594 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21595 (MULTILIB_RESUE): Likewise.
21596 (MULTILIB_MATCHES): Likewise.
21597 (MULTLIB_REQUIRED): Likewise.
21599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21601 * config/arm/t-rtems: Rewrite for new option framework.
21603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21605 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21606 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21607 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21608 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21609 * config/arm/t-multilib: ... here.
21610 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21611 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21612 armv7-a and armv8*-a when A-profile libraries have not been built.
21613 * config/arm/t-rmprofile: Rewrite.
21615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21617 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21618 with a backslash. Remove the backslash after substituting unescaped
21620 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21624 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21625 the driver to pass through the default setting of -mfloat-abi.
21626 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21628 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21629 (all_feat_combs): New rule.
21630 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21632 * config/arm/t-aprofile: Rewrite.
21634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21636 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21637 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21638 fpu is not specified by the user/command-line.
21639 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21640 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21641 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21642 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21643 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21644 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21648 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21649 * config/arm/t-arm-elf: Rewritten.
21651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21653 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21654 have some floating-point instructions.
21655 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21656 (TARGET_MAYBE_HARD_FLOAT): New macro.
21657 * config/arm/arm-builtins.c (arm_init_builtins): Use
21658 TARGET_MAYBE_HARD_FLOAT.
21659 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21663 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21664 (configargs.h): Include it.
21665 (arm_print_hint_for_fpu_option): New function.
21666 (arm_parse_fpu_option): New function.
21667 (candidate_extension): New class.
21668 (arm_canon_for_multilib): New function.
21669 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21670 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21671 (ARCH_CANONICAL_SPECS): New macro.
21672 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21676 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21677 are set after handling multilib fragments. Set target_cpu_default2
21680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21682 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21684 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21685 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21687 (check_arch): Likewise.
21688 * config/arm/arm.c (arm_configure_build_target): Handle
21689 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21690 options in the default.
21692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21694 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21695 when an option is an alias of another.
21696 * config/arm/parsecpu.awk (optalias): New parser token.
21697 (gen_comm_data): Mark non-alias options as such. Emit entries
21698 for extension aliases.
21699 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21700 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21701 (armv6kz, armv6zk, armv6t2): Likewise.
21702 (armv7): Make vfpv3-d16 an alias.
21703 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21705 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21706 Sort in canonical order.
21707 (armv8-a): Sort in canonical order.
21708 (armv8.1-a, armv8.2-a): Likewise.
21709 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21711 (cortex-a9): Sort in canonical order.
21712 * config/arm/arm.c (selftests.h): Include it.
21713 (arm_test_cpu_arch_data): New function.
21714 (arm_run_self_tests): New function.
21715 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21716 (targetm): Move declaration to the end of the file.
21717 * arm-cpu-cdata.h: Regenerated.
21719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21721 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21722 call to target_mode_check describing the type of option passed.
21723 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21724 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21725 arm_parse_cpu_option_name to match parameters against list of
21727 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21728 arm_arch_core_flags data structure.
21729 * config/arm/arm-cpu_cdata.h: Regenerated.
21731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21733 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21735 (arm_print_hint_for_cpu_option): Likewise.
21736 (arm_print_hint_for_arch_option): Likewise.
21737 (arm_parse_cpu_option_name): Likewise.
21738 (arm_parse_arch_option_name): Likewise.
21739 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21740 of entries in the all_fpus list.
21741 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21742 (arm_parse_cpu_option_name): Declare.
21743 (arm_parse_arch_option_name): Declare.
21744 (arm_parse_option_features): Declare.
21745 (arm_intialize_isa): Declare.
21746 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21748 (gen_comm_data): ... here. Make definitions non-static.
21749 * config/arm/arm-cpu-data.h: Regenerated.
21750 * config/arm/arm-cpu-cdata.h: Regenerated.
21752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21754 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21755 (cpu_arch_extension): New structure.
21756 (cpu_arch_option, arch_option, cpu_option): New structures.
21757 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21758 architecture types.
21759 (gen_data): Generate new format data tables.
21760 * config/arm/arm.c (cpu_tune): New structure.
21761 (cpu_option, processors): Delete.
21762 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21763 (arm_print_hint_for_cpu_option): ... this and ...
21764 (arm_print_hint_for_arch_option): ... this.
21765 (arm_parse_arch_cpu_name): Delete. Replace with ...
21766 (arm_parse_cpu_option_name): ... this and ...
21767 (arm_parse_arch_option_name): ... this.
21768 (arm_unrecognized_feature): Change type of target parameter to
21770 (arm_parse_arch_cpu_features): Delete. Replace with ...
21771 (arm_parse_option_features): ... this.
21772 (arm_configure_build_target): Rework to use new configuration data
21774 (arm_print_tune_info): Rework for new configuration data tables.
21775 * config/arm/arm-cpu-data.h: Regenerated.
21776 * config/arm/arm-cpu.h: Regenerated.
21778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21780 * Makefile.in (OBJS): Move sbitmap.o from here ...
21781 (OBJS-libcommon): ... to here.
21783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21785 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21786 (ISA_ALL_CRYPTO): New macro.
21787 (ISA_ALL_SIMD): New macro
21788 (ISA_ALL_FP): New macro.
21789 * config/arm/arm.c (fpu_bitlist): Update initializer.
21790 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21792 (arm9e): Add fpu. Add option for nofp
21793 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21794 (arm926ej-s, arm1026ej-s): Likewise.
21795 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21796 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21797 neon-fp16, neon-vfpv4, nofp and nosimd.
21798 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21799 (cortex-a8): Add fpu. Add option for nofp.
21800 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21801 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21802 (cortex-r4f): Add fpu.
21803 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21804 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21806 (cortex-r8): Likewise.
21807 (cortex-m4): Add fpu. Add option for nofp.
21808 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21809 (cortex-a17.cortex-a7): Likewise.
21810 (cortex-a32): Add fpu. Add options for crypto and nofp.
21811 (cortex-a35, cortex-a53): Likewise.
21812 (cortex-a57): Add fpu. Add option for crypto.
21813 (cortex-a72, cortex-a73): Likewise.
21814 (exynos-m1): Likewise.
21815 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21816 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21817 (cortex-m33): Add fpu. Add option for nofp.
21818 * config/arm/arm-cpu-cdata.h: Regenerated
21819 * config/arm/arm-cpu-data.h: Regenerated.
21821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21823 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21824 (armv5te, armv5tej): Likewise.
21825 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21826 (armv7): Add options fp and vfpv3-d16.
21827 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21828 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21830 (armv7ve): Likewise.
21831 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21832 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21833 (armv8-a): Add nocrypto option.
21834 (armv8.1-a, armv8.2-a): Likewise.
21835 (armv8-m.main): add options fp, fp.dp and nofp.
21837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21839 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21841 (armv8-a+crc): Delete.
21842 (armv8.1-a): Add options simd, crypto and nofp.
21843 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21844 (armv8.2-a+fp16): Delete.
21845 (armv8-m.main): Add option dsp.
21846 (armv8-m.main+dsp): Delete.
21847 (cortex-a8): Add fpu. Add nofp option.
21848 (cortex-a9): Add fpu. Add nofp and nosimd options.
21849 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21850 link to main cpu and architecture data structures.
21851 (gen_comm_data): Only put isa attributes from the main architecture
21853 (option): New statement for architecture and CPU entries.
21854 * arm.c (struct cpu_option): New structure.
21855 (struct processors): Add entry for options.
21856 (arm_unrecognized_feature): New function.
21857 (arm_parse_arch_cpu_name): Ignore any characters after the first
21859 (arm_parse_arch_cpu_feature): New function.
21860 (arm_configure_build_target): Separate out any CPU and architecture
21861 features and parse separately. Don't error out if -mfpu=auto is
21862 used with only an architecture string.
21863 (arm_print_asm_arch_directives): New function.
21864 (arm_file_start): Call it.
21865 * config/arm/arm-cpu-cdata.h: Regenerated.
21866 * config/arm/arm-cpu-data.h: Likewise.
21867 * config/arm/arm-tables.opt: Likewise.
21869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21871 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21872 assembler when it is not -mfpu=auto.
21874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21876 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21877 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21878 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21879 (ASM_CPU_SPEC): Rewrite.
21880 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21881 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21882 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21883 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21884 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21885 copied string is NUL-terminated. Also strip any characters prefixed
21887 (arm_rewrite_selected_arch): New function.
21888 (arm_rewrite_march): New function.
21890 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21892 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21893 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21894 (march, mcpu, mtune): Convert to string-based options.
21895 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21896 (arm_parse_arch_cpu_name): New function.
21897 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21898 identify selected architecture or CPU.
21899 (arm_option_save): New function.
21900 (TARGET_OPTION_SAVE): Redefine.
21901 (arm_option_restore): Restore string options.
21902 (arm_option_print): Print string options.
21904 2017-06-16 Martin Sebor <msebor@redhat.com>
21906 PR tree-optimization/80933
21907 PR tree-optimization/80934
21908 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21909 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21910 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21911 (gimple_fold_builtin): Call them.
21913 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21915 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21916 as unlikely; update profile.
21918 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21920 * predict.c (force_edge_cold): Handle declaring edges impossible
21923 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21925 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21927 (try_unroll_loop_completely): Fix reporting.
21929 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21931 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21933 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21936 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21937 if given a non-constant argument for an intrinsic which requires a
21940 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21942 * profile.c (compare_freqs): New function.
21943 (branch_prob): Sort edge list.
21944 (find_spanning_tree): Assume that the list is priority sorted.
21946 2017-06-16 Richard Biener <rguenther@suse.de>
21948 PR tree-optimization/81090
21949 * passes.def (pass_record_bounds): Remove.
21950 * tree-pass.h (make_pass_record_bounds): Likewise.
21951 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21952 make_pass_record_bounds): Likewise.
21953 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21954 not free niter estimates at the beginning but at the end.
21955 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21957 2017-06-16 Richard Biener <rguenther@suse.de>
21959 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21960 initializer to workaround ICE in host GCC 4.8.
21962 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21964 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21966 (clone_inlined_nodes): Update.
21968 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21970 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21971 prefetch settings, and enable prefetching by default at -O3.
21973 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21975 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21976 Set flag_prefetch_loop_arrays according to tuning data.
21978 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21980 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21981 New tune structure.
21982 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21983 [Unrelated to main purpose of the patch] Place the pointer field last
21984 to enable type checking errors when tune structure are wrongly merged.
21985 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21986 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21987 (thunderx2t99_prefetch_tune): New tune constants.
21988 (tune_params *_tunings): Update all tunings (no functional change).
21989 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21990 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21991 from tunings structures.
21993 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21996 * ubsan.c (instrument_null): Add T argument, use it instead
21997 of computing it based on IS_LHS.
21998 (instrument_object_size): Likewise.
21999 (pass_ubsan::execute): Adjust instrument_null and
22000 instrument_object_size callers to pass gimple_get_lhs or
22001 gimple_assign_rhs1 result to it. Use instrument_null instead of
22002 calling get_base_address and instrument_mem_ref. Handle
22003 aggregate call arguments for object-size sanitization.
22005 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22007 PR tree-optimization/81089
22008 * tree-vrp.c (is_masked_range_test): Validate operands of
22011 2017-06-15 Martin Sebor <msebor@redhat.com>
22014 * dumpfile.c (dump_register): Avoid calling memset to initialize
22015 a class with a default ctor.
22016 * gcc.c (struct compiler): Remove const qualification.
22017 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22018 * hash-table.h: Ditto.
22019 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22021 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22022 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22024 * params.h (struct param_info): Make struct members non-const.
22025 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22026 with default initialization.
22027 * vec.h (vec_copy_construct, vec_default_construct): New helper
22029 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22030 with vec_copy_construct.
22031 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22032 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22033 * doc/invoke.texi (-Wclass-memaccess): Document.
22035 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22037 * emit-rtl.h (is_leaf): Update comment about local
22038 register allocator.
22040 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22043 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22044 for a variable to have a section before checking if the section has a
22046 Set section to.persistent if persistent attribute is set.
22047 Warn if .persistent attribute is used on an automatic variable.
22049 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22051 PR rtl-optimization/80474
22052 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22054 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22056 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22059 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22061 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22062 (casesi): Emit bounds checking as RTL.
22063 (casesi_internal_mips16_<mode>): Remove bounds checking.
22065 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22067 * config/xtensa/xtensa.c (xtensa_option_override): Append
22068 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22069 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22070 xtensa_doloop_hooks): Define unconditionally.
22071 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22073 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22074 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22075 for it in xtensa_option_override.
22076 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22077 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22079 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22081 * doc/cppopts.texi: Document '-' special value to -MF.
22083 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22085 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22086 (cortex_a53_fconst): Likewise.
22087 (cortex_a53_fpmul): Likewise.
22088 (cortex_a53_f_load_64): Likewise.
22089 (cortex_a53_f_load_many): Likewise.
22090 (cortex_a53_advsimd_alu): Likewise.
22091 (cortex_a53_advsimd_alu_q): Likewise.
22092 (cortex_a53_advsimd_mul): Likewise.
22093 (cortex_a53_advsimd_mul_q): Likewise.
22094 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22095 Add missing fmul, r2f_cvt and fconst cases.
22097 2017-06-14 Richard Biener <rguenther@suse.de>
22099 PR middle-end/81088
22100 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22102 (fold_binary_loc): When associating do not treat pre-existing
22103 TREE_OVERFLOW on literal constants as a reason to allow
22104 TREE_OVERFLOW on associated literal constants.
22106 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22108 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22109 (MASK_FEATURES): New macro.
22110 * config/sparc/sparc.c (sparc_option_override): Remove the special
22111 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22113 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22115 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22116 a division of 0 if non-call exceptions are enabled.
22118 2017-06-14 Andrew Pinski <apinski@cavium.com>
22119 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22122 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22123 Improve vector initialization code gen for only variable case.
22125 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22127 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22129 2017-06-14 Richard Biener <rguenther@suse.de>
22131 PR tree-optimization/81083
22132 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22137 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22138 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22139 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22140 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22141 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22142 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22144 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22146 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22147 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22151 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22153 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22155 * config/rs6000/t-rtems: Don't handle SPE.
22157 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22159 * config/rs6000/t-linux: Don't handle SPE.
22161 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22163 * config/rs6000/eabispe.h: Delete file.
22165 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22167 * config/rs6000/t-spe: Delete file.
22169 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22171 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22172 (rs6000_legitimate_offset_address_p): Return false for anything in
22173 V2SImode or V2SFmode.
22175 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22177 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22178 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22179 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22181 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22182 (rs6000_legitimate_offset_address_p): Ditto.
22183 (rs6000_emit_move): Ditto.
22184 (rs6000_init_builtins): Remove V4HI_type_node.
22186 2017-06-13 Martin Liska <mliska@suse.cz>
22189 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22190 (gate_asan): Likewise.
22191 * asan.h (asan_no_sanitize_address_p): Remove the function.
22192 (sanitize_flags_p): New function.
22193 * builtins.def: Fix coding style.
22194 * common.opt: Use renamed enum value.
22195 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22196 * doc/extend.texi: Document no_sanitize attribute.
22197 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22198 to SANITIZE_UNDEFINED_NONDEFAULT.
22199 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22200 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22201 Use sanitize_flags_p.
22202 * gimplify.c (gimplify_function_tree): Likewise.
22203 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22204 * opts.c (parse_no_sanitize_attribute): New function.
22205 (common_handle_option): Use renamed enum value.
22206 * opts.h (parse_no_sanitize_attribute): Declare.
22207 * tree.c (sanitize_flags_p): New function.
22208 * tree.h: Declared here.
22209 * tsan.c: Use sanitize_flags_p.
22210 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22211 (instrument_mem_ref): Likewise.
22212 (instrument_bool_enum_load): Likewise.
22213 (do_ubsan_in_current_function): Remove the function.
22214 (pass_ubsan::execute): Use sanitize_flags_p.
22215 * ubsan.h: Remove do_ubsan_in_current_function
22216 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22217 (dump_function_to_file): Use it here.
22219 2017-06-13 Martin Jambor <mjambor@suse.cz>
22221 PR tree-optimization/80803
22222 PR tree-optimization/81063
22223 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22224 (propagate_subaccesses_across_link): Enqueue subtree whenever
22225 necessary instead of relying on the caller.
22227 2017-06-13 Martin Jambor <mjambor@suse.cz>
22229 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22230 that have a first_link.
22231 (sort_and_splice_var_accesses): Do not check first_link before
22233 (subtree_mark_written_and_enqueue): Likewise.
22234 (propagate_all_subaccesses): Likewise and do not stop at first
22235 parent with a first_link.
22237 2017-06-13 Martin Jambor <mjambor@suse.cz>
22239 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22242 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22244 * match.pd: New pattern.
22246 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22248 * tree-vrp.c (is_masked_range_test): New function.
22249 (register_edge_assert_for): Determine ranges for
22252 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22254 PR tree-optimization/67328
22255 * fold-const.c (maskable_range_p): New function.
22256 (build_range_check): Generate bittests if possible.
22258 2017-06-13 Martin Liska <mliska@suse.cz>
22260 * gimple-pretty-print.c (dump_probability): Add new argument.
22261 (dump_edge_probability): Dump both probability and count.
22262 (dump_gimple_label): Likewise.
22263 (dump_gimple_bb_header): Likewise.
22265 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22268 * config/avr/avr-devices.c: Fix indentation.
22269 * config/avr/gen-avr-mmcu-specs.c: Dito.
22271 2017-06-13 Richard Biener <rguenther@suse.de>
22273 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
22274 instead get vector type from stmt_info.
22275 (vectorizable_reduction): Adjust. Remove dead code.
22277 2017-06-13 Richard Biener <rguenther@suse.de>
22279 PR middle-end/81065
22280 * fold-const.c (extract_muldiv_1): Remove bogus distribution
22281 case of C * (x * C2 + C3).
22282 (fold_addr_of_array_ref_difference): Properly fold index difference.
22284 2017-06-12 David S. Miller <davem@davemloft.net>
22287 * config/sparc/sparc.md (return expander): Emit frame blockage if
22288 function uses alloca.
22290 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
22292 * combine.c (make_field_assignment): Check len rather than the mode
22293 precision when calling force_to_mode.
22295 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
22297 Support multilibs and devices that see flash in RAM address range.
22300 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
22301 (avr_mcu_t) <flash_pm_offset>: New field.
22302 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22303 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22304 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22305 (AVR_TINY_PM_OFFSET): Remove macro.
22306 * config/avr/avr.opt (-mshort-calls): New option.
22307 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22308 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22309 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22310 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22311 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22312 instead of avr_arch->have_jmp_call.
22313 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22314 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22315 avr_arch->flash_pm_offset to define.
22316 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22317 new field flash_pm_offset. Add entry for avrxmega3.
22318 (avr_texinfo): Add entry for avrxmega3.
22319 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22320 attiny212, attiny214,
22321 attiny412, attiny414, attiny416, attiny417,
22322 attiny814, attiny816, attiny817,
22323 attiny1614, attiny1616, attiny1617,
22324 attiny3214, attiny3216, attiny3217.
22325 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22326 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22327 (avr_print_operand_address) [AVR_TINY]: Same.
22328 (avr_asm_init_sections) <readonly_data_section>: Only patch
22329 callback if avr_arch->flash_pm_offset = 0.
22330 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22331 for rodata if avr_arch->flash_pm_offset != 0.
22332 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22333 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22334 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22335 (m_options): Append opt_rcall.
22336 (m_dirnames): Append dir_rcall.
22337 * config/avr/t-multilib: Regenerate.
22339 * configure.ac [target=avr]: Check whether avrxmega3 default
22340 linker description file works as needed.
22341 * configure: Regenerate.
22342 * doc/avr-mmcu.texi: Regenerate.
22343 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22344 <__AVR_ARCH__>: Document avrxmega3 and 103.
22345 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22346 <__AVR_SHORT_CALLS__>: Document it.
22347 <__AVR_PM_BASE_ADDRESS__>: Document it.
22348 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22349 (AVR Variable Attributes) <progmem>: Document this is
22350 not needed for avrxmega3.
22351 (AVR Named Address Spaces) <__flash>: Dito.
22353 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22355 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22357 2017-06-12 Doug Rupp <rupp@adacore.com>
22359 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22360 Append vxworks-stdint.h to the tm_file list.
22361 * config/vxworks-stdint.h: New file.
22363 2017-06-12 Martin Liska <mliska@suse.cz>
22365 PR tree-optimization/81041
22366 * tree-profile.c (gimple_gen_ic_func_profiler):
22367 Create an extra BB in profile-generate
22368 (gimple_gen_time_profiler): Likewise.
22370 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22372 PR tree-optimization/81003
22373 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22374 (update_range_test): Use it instead of force_gimple_operand_gsi.
22376 2017-06-12 Richard Biener <rguenther@suse.de>
22378 PR tree-optimization/81053
22379 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22380 with backedge value not defined in loop. Simplify def stmt
22383 2017-06-11 Tom de Vries <tom@codesourcery.com>
22386 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22388 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22389 nvptx_cannot_force_const_mem.
22391 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22393 * opts.c (finish_options): Move test for flag_split_stack after
22394 it has been initialized.
22396 2017-06-11 Jason Merrill <jason@redhat.com>
22398 * tree.h (id_equal): New.
22399 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22400 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22401 instead of strcmp of IDENTIFIER_POINTER.
22403 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22405 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22406 (mark_all_inlined_calls_cdtor): Fix formating.
22407 (inline_transform): Rescale profile before inlining.
22409 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22411 * cgraph.h (cgraph_edge::clone): Update prototype.
22412 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22413 (cgraph_node::create_clone): Update.
22414 (cgraph_node::create_version_clone): Update.
22415 * tree-inline.c (copy_bb): Update.
22416 (expand_call_inline): Update.
22418 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22420 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22421 factored out from ...
22422 (rs6000_emit_prologue): ... here.
22424 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22426 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22427 factored out from ...
22428 (rs6000_emit_prologue): ... here.
22430 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22432 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22434 (handle_missing_profiles): Fix computation of tp_first_run.
22435 (counts_to_freqs): Do not touch freqs when count is 0.
22437 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22439 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22442 2017-06-10 Tom de Vries <tom@codesourcery.com>
22444 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22445 attributes): Document signal effective target.
22447 2017-06-10 Tom de Vries <tom@codesourcery.com>
22449 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22450 Document effective target stack_size.
22452 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22454 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22455 to the edit_context if they can be auto-applied.
22457 2017-06-9 Ian Lance Taylor <iant@golang.org>
22459 * opts.c (finish_options): If -fsplit-stack, disable implicit
22460 -forder-blocks-and-partition.
22461 * doc/invoke.texi (Optimize Options): Document that when using
22462 -fsplit-stack -forder-blocks-and-partition is not implicitly
22465 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22467 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22468 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22469 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22470 * builtins.def (abort, trap, unreachable): Declare cold.
22471 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22472 * tree-core.h (ECF_COLD): New.
22473 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22474 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22476 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22478 * predict.c (unlikely_executed_stmt_p): Cleanup.
22480 2017-06-09 Richard Biener <rguenther@suse.de>
22482 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22483 model if the ref is always written to.
22485 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22487 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22489 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22491 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22494 2017-06-09 Tom de Vries <tom@codesourcery.com>
22497 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22498 "target cannot support label values" when encountering LABEL_REF.
22500 2017-06-09 Martin Liska <mliska@suse.cz>
22502 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22503 (gimple_gen_ic_func_profiler): Emit direct comparison
22504 of __gcov_indirect_call_callee with NULL.
22505 (gimple_gen_time_profiler): Change probability from
22506 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22508 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22510 * profile.c (edge_gcov_counts): Turn to pointer.
22511 (compute_branch_probabilities, compute_branch_probabilities): Update.
22512 (branch_prob): Do not clear edge_gcov_count.
22513 * profile.h (edge_gcov_counts): Turn to pointer.
22514 (edge_gcov_count): Update.
22516 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22518 * gimple.h (gimple_check_failed): Mark cold.
22520 2017-06-09 Richard Biener <rguenther@suse.de>
22522 PR tree-optimization/66623
22523 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22524 refactor check_reduction into two parts, properly computing
22525 whether we have to check reduction validity for outer loop
22528 2017-06-09 Richard Biener <rguenther@suse.de>
22530 PR tree-optimization/79483
22531 * graphite-scop-detection.c (order): New global.
22532 (get_order): Compute bb to order mapping that satisfies code
22533 generation constraints.
22534 (cmp_pbbs): New helper.
22535 (build_scops): Start domwalk at entry block, sort generated
22538 2017-06-09 Richard Biener <rguenther@suse.de>
22540 PR middle-end/81007
22541 * ipa-polymorphic-call.c
22542 (ipa_polymorphic_call_context::restrict_to_inner_class):
22543 Skip FIELD_DECLs with error_mark_node type.
22544 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22547 2017-06-09 Martin Liska <mliska@suse.cz>
22549 * predict.c (struct branch_predictor): New struct.
22550 (test_prediction_value_range): New test.
22551 (predict_c_tests): New function.
22552 * selftest-run-tests.c (selftest::run_tests): Run the function.
22553 * selftest.h: Declare new tests.
22555 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22558 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22559 gen_add3_insn did not fail.
22560 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22561 r0, construct that number in a temporary reg and add that reg to r0.
22562 If asked to put the result in r0 as well, fail.
22564 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22566 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22567 for early expansion of vec_eqv.
22569 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22571 PR middle-end/81005
22572 * ubsan.c (instrument_null): Avoid pointless code temporary.
22573 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22575 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22579 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22582 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22585 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22587 * predict.c (unlikely_executed_edge_p): Move ahead.
22588 (probably_never_executed_edge_p): Use it.
22590 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22592 PR middle-end/79988
22593 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22594 gimple_call_builtin_p call.
22596 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22598 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22599 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22600 rtl_check_failed_type2, rtl_check_failed_code1,
22601 rtl_check_failed_code2, rtl_check_failed_code_mode,
22602 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22603 rtvec_check_failed_bounds, rtl_check_failed_flag,
22604 _fatal_insn_not_found, _fatal_insn): Likewise.
22605 * tree.h (tree_contains_struct_check_failed,
22606 tree_check_failed, tree_not_check_failed,
22607 tree_class_check_failed, tree_range_check_failed,
22608 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22609 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22610 tree_operand_check_failed, omp_clause_check_failed,
22611 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22614 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22616 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22617 flag_branch_probabilities.
22618 * ipa-inline.c (edge_badness): Likewise.
22619 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22620 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22621 * predict.c (maybe_hot_frequency_p): Likewise.
22622 (probably_never_executed): Likewise.
22623 * sched-ebb.c (schedule_ebbs): Likewise.
22624 * sched-rgn.c (find_single_block_region): Likewise.
22625 * tracer.c (tail_duplicate): Likewise.
22627 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22629 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22630 longer requires x_flag_profile_use.
22632 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22634 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22635 instead of flag_reorder_blocks_and_partition.
22636 * dbxout.c (dbxout_function_end): Likewise.
22637 * dwarf2out.c (gen_subprogram_die): Likewise.
22638 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22639 * hw-doloop.c (reorg_loops): Likewise.
22640 * varasm.c (assemble_start_function,
22641 assemble_end_function): Likewise.
22642 (decide_function_section): Do not check for
22643 flag_reorder_blocks_and_partition.
22645 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22647 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22649 (chkp_get_hard_register_fake_addr_expr): Ditto.
22650 (chkp_build_addr_expr): Add check for hard reg case.
22651 (chkp_parse_array_and_component_ref): Ditto.
22652 (chkp_find_bounds_1): Ditto.
22653 (chkp_process_stmt): Don't generate bounds store for
22656 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22658 * predict.c (maybe_hot_bb_p): Do not check profile status.
22659 (maybe_hot_edge_p): Likewise.
22660 (probably_never_executed): Check for zero counts even if profile
22662 (unlikely_executed_edge_p): New function.
22663 (unlikely_executed_stmt_p): New function.
22664 (unlikely_executed_bb_p): New function.
22665 (set_even_probabilities): Use unlikely predicates.
22666 (combine_predictions_for_bb): Likewise.
22667 (predict_paths_for_bb): Likewise.
22668 (predict_paths_leading_to_edge): Likewise.
22669 (determine_unlikely_bbs): New function.
22670 (estimate_bb_frequencies): Use it.
22671 (compute_function_frequency): Use zero counts even if profile is
22673 * profile-count.h: Fix typo.
22675 2017-08-08 Julia Koval <julia.koval@intel.com>
22677 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22678 _mm512_mask_cvtsepi16_storeu_epi8,
22679 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22680 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22681 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22682 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22683 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22684 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22685 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22686 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22687 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22688 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22689 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22690 __builtin_ia32_pmovuswb256mem_mask,
22691 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22692 __builtin_ia32_pmovwb512mem_mask): New builtins.
22694 2017-08-08 Julia Koval <julia.koval@intel.com>
22696 PR target/73350,80862
22697 * config/i386/subst.md (round): Fix round pattern.
22698 * config/i386/i386.c (ix86_erase_embedded_rounding):
22699 Fix erasing rounding for the fixed pattern.
22701 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22703 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22705 2017-06-08 Martin Liska <mliska@suse.cz>
22707 PR gcov-profile/80911
22708 * gcov.c (block_info::block_info): New constructor.
22710 2017-06-07 Carl Love <cel@us.ibm.com>
22712 * config/rs6000/rs6000-c: The return type of the following
22713 built-in functions was implemented as int not long long. Fix sign
22714 of return value for the unsigned version of vec_mulo and vec_mule.
22715 vector unsigned long long vec_bperm (vector unsigned long long,
22716 vector unsigned char)
22717 vector signed long long vec_mule (vector signed int,
22719 vector unsigned long long vec_mule (vector unsigned int,
22720 vector unsigned int)
22721 vector signed long long vec_mulo (vector signed int,
22723 vector unsigned long long vec_mulo (vector unsigned int,
22724 vector unsigned int)
22725 * doc/extend.texi: Fix the documentation for the built-in
22728 2017-06-07 Carl Love <cel@us.ibm.com>
22731 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22734 2017-06-07 Carl Love <cel@us.ibm.com>
22736 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22737 support, Generate doublehv for signed int/float for BE case only.
22739 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22741 * doc/invoke.texi (mcx16): Rewrite.
22743 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22745 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22746 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22747 *mov<mode>_softfloat, and an anonymous splitter): Use
22748 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22752 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22754 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22755 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22756 (rs6000_debug_reg_global): Adjust.
22757 (rs6000_init_hard_regno_mode_ok): Adjust.
22758 (rs6000_dbx_register_number): Adjust.
22759 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22760 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22761 Remove SPE_ACC and SPEFSCR.
22762 (REG_ALLOC_ORDER): Ditto.
22763 (FRAME_POINTER_REGNUM): Change to 111.
22764 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22765 (REG_CLASS_NAMES): Ditto.
22766 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22767 (REGISTER_NAMES): Ditto.
22768 (ADDITIONAL_REG_NAMES): Ditto.
22769 (rs6000_reg_names): Ditto.
22770 * config/rs6000/rs6000.md: Renumber some register number
22773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22775 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22777 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22778 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22780 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22781 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22782 Delete the SPE high registers.
22783 (REG_ALLOC_ORDER): Ditto.
22784 (enum reg_class): Remove SPE_HIGH_REGS.
22785 (REG_CLASS_NAMES): Ditto.
22786 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22787 (REGISTER_NAMES): Ditto.
22788 (rs6000_reg_names): Ditto.
22789 * doc/tm.texi.in: Remove SPE as example.
22790 * doc/tm.texi: Regenerate.
22792 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22794 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22795 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22796 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22797 * config/rs6000/rs6000.md (type): Remove "brinc".
22799 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22801 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22802 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22803 * config/rs6000/linuxspe.h: Delete file.
22804 * config/rs6000/rs6000.md: Don't include spe.md.
22805 * config/rs6000/spe.h: Delete file.
22806 * config/rs6000/spe.md: Delete file.
22807 * config/rs6000/t-rs6000: Remove spe.md.
22809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22811 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22812 (reg_or_none500mem_operand): Delete.
22813 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22814 instead of reg_or_none500mem_operand.
22816 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22818 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22819 handling of SPE flags.
22820 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22824 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22826 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22827 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22828 paired_divv2sf3): Similar.
22829 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22830 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22831 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22833 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22834 Rename the paired_* instruction patterns.
22835 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22837 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22838 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22839 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22840 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22841 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22842 PAIRED_VECTOR_MODE.
22843 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22844 (spe_func_has_64bit_regs_p): Delete.
22845 (spe_expand_predicate_builtin): Delete.
22846 (spe_expand_evsel_builtin): Delete.
22847 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22848 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22849 (invalid_e500_subreg): Delete.
22850 (rs6000_legitimize_address): Always force_reg op2 as well, for
22851 paired single memory accesses.
22852 (rs6000_member_type_forces_blk): Delete.
22853 (rs6000_spe_function_arg): Delete.
22854 (rs6000_expand_unop_builtin): Delete SPE handling.
22855 (rs6000_expand_binop_builtin): Ditto.
22856 (spe_expand_stv_builtin): Delete.
22857 (bdesc_2arg_spe): Delete.
22858 (spe_expand_builtin): Delete.
22859 (spe_expand_predicate_builtin): Delete.
22860 (spe_expand_evsel_builtin): Delete.
22861 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22862 (spe_init_builtins): Delete.
22863 (spe_func_has_64bit_regs_p): Delete.
22864 (savres_routine_name): Delete "info" parameter. Adjust callers.
22865 (rs6000_emit_stack_reset): Ditto.
22866 (rs6000_dwarf_register_span): Delete.
22867 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22868 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22869 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22871 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22873 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22874 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22875 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22876 mulv2sf3, divv2sf3): Delete expanders.
22878 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22880 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22884 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22885 * config/rs6000/rs6000.c: Ditto.
22887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22889 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22890 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22891 comparison_operator.
22893 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22895 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22896 * config/rs6000/rs6000.opt: Ditto.
22897 * config/rs6000/t-rtems: Ditto.
22899 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22901 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22902 TARGET_E500_SINGLE by 0, simplify.
22903 * config/rs6000/rs6000.c: Ditto.
22904 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22905 (spe_build_register_parallel): Delete.
22906 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22907 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22908 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22909 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22910 (E500_CONVERT): Delete.
22911 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22913 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22915 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22916 * config/rs6000/dfp.md: Ditto.
22917 (negdd2, *negdd2_fpr): Merge.
22918 (absdd2, *absdd2_fpr): Merge.
22919 (negtd2, *negtd2_fpr): Merge.
22920 (abstd2, *abstd2_fpr): Merge.
22921 * config/rs6000/e500.h: Delete file.
22922 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22923 TARGET_FPRS by 1 and simplify.
22924 * config/rs6000/rs6000-c.c: Ditto.
22925 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22926 TARGET_DF_SPE by 0.
22927 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22929 * config/rs6000/rs6000.md: Ditto.
22930 (floatdidf2, *floatdidf2_fpr): Merge.
22931 (move_from_CR_gt_bit): Delete.
22932 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22933 (E500_CR_IOR_COMPARE): Delete.
22934 (All patterns that require !TARGET_FPRS): Delete.
22935 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22937 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22939 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22941 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22943 * graphds.c (add_edge): Intitialize edge's attached data.
22944 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22945 pointer parameter. Call pointed function on each edge during
22946 graph traversing. Skip traversing the edge when the function
22948 (graphds_dfs, graphds_scc): Ditto.
22949 (for_each_edge): New parameter. Pass the new parameter to callback
22951 * graphds.h (skip_edge_callback): New function pointer type.
22952 (graphds_dfs, graphds_scc): New function pointer parameter.
22953 (graphds_edge_callback, for_each_edge): New parameter.
22955 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22957 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22958 out code checking if runtime alias check is possible to below ...
22959 Call the new function.
22960 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22961 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22963 2017-06-07 Marek Polacek <polacek@redhat.com>
22966 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22967 TYPE_OVERFLOW_WRAPS check.
22969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22971 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22972 if versioning is required.
22973 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22974 peeling with the check for versioning.
22976 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22978 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22979 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22980 Set true to new parameter if new ssa variable is defined.
22981 (vect_gen_vector_loop_niters): Refactor. Set range information
22982 for the new vector loop bound variable.
22983 (vect_do_peeling): Ditto.
22985 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22987 * tree-affine.c (ssa.h): Include header file.
22988 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22989 has wrapping overflow behavior.
22991 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22993 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22995 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22997 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22998 (tree_to_aff_combination): ... here.
23000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23002 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23003 reg_pressure model function.
23004 (ivopts_global_cost_for_size): Delete.
23005 (determine_set_costs, iv_ca_recount_cost): Call new model function
23006 ivopts_estimate_reg_pressure.
23008 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23010 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23011 expensive than udiv. Remove floating point cases from mod.
23013 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23015 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23016 Increase idiv cost.
23018 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23020 * config/aarch64/aarch64.md
23021 (copysignsf3): Fix mask generation.
23023 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23025 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23027 (class dump_manager): Add register_dumps method.
23028 * dumpfile.c: Include langhooks.h.
23029 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23030 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23031 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23032 (dump_manager::dump_register): Start with 512 entries instead of 32.
23033 (dump_manager::register_dumps): New method.
23034 * toplev.c (general_init): Instead of invoking register_dumps
23035 langhook, invoke register_dumps method on the dump manager.
23036 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23039 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23041 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23042 Say that # requires an associated define_split to exist, and that
23043 the define_split must be suitable for use after register allocation.
23045 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23047 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23048 (compute_outgoing_frequencies): Also initialize zero counts.
23049 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23050 around loops; preserve more of profile when nothing changes.
23052 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23054 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23056 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23057 * config/arm/arm-cpu-cdata.h: Regenerate.
23058 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23059 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23060 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23061 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23062 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23064 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23066 * config/arm/t-rmprofile: Likewise.
23067 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23069 2017-06-06 David S. Miller <davem@davemloft.net>
23072 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23073 blockage if function uses alloca.
23075 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23077 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23078 New "uid" fields to hold pretty-print IDs of group and ref.
23079 Memory references are now identified as <group_id>:<ref_id>
23080 instead of using [random] addresses.
23081 (dump_mem_details): Simplify, no functional change.
23082 (dump_mem_ref): Simplify and make output more concise.
23083 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23084 (find_or_create_group): Initialize group uid.
23085 (record_ref): Initialize ref uid. Improve debug output.
23086 (prune_group_by_reuse, should_issue_prefetch_p,)
23087 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23088 (mark_nontemporal_store, determine_loop_nest_reuse):
23089 Improve debug output.
23091 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23093 * dbgcnt.def (prefetch): New debug counter.
23094 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23095 (schedule_prefetches): Stop issueing prefetches if debug counter
23098 2017-06-06 Tom de Vries <tom@codesourcery.com>
23100 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23101 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23103 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23105 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23106 Use aarch64_reg_or_zero predicate for operand 4.
23107 (aarch64_compare_and_swap<mode> define_insn_and_split):
23108 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23109 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23111 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23113 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23114 (arm_compute_save_core_reg_mask): This.
23115 (thumb1_compute_save_reg_mask): Rename into ...
23116 (thumb1_compute_save_core_reg_mask): This.
23117 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23118 (arm_compute_frame_layout): Likewise.
23120 2017-06-06 Richard Biener <rguenther@suse.de>
23122 PR tree-optimization/80974
23123 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23124 keep or clear leaders SSA info.
23126 2017-06-06 Tom de Vries <tom@codesourcery.com>
23128 * config/nvptx/nvptx.c (split_mode_p): New function.
23129 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23131 2017-06-06 Tom de Vries <tom@codesourcery.com>
23133 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23135 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23138 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23141 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23143 * shrink-wrap.c (handle_simple_exit): Update profile.
23144 (try_shrink_wrapping): Upate profile.
23146 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23148 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23149 (tree_guess_outgoing_edge_probabilities): New.
23150 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23151 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23153 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23155 * ipa-split.c (split_function): Initialize return bb profile.
23157 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23159 * profile.c (compute_branch_probabilities): Also initialize
23160 EXIT_BLOCK profile.
23162 2017-06-06 Richard Biener <rguenther@suse.de>
23164 PR tree-optimization/80928
23165 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23166 (vect_analyze_loop_operations): Properly guard analysis for
23168 (vect_transform_loop): Likewise.
23169 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23170 (vect_model_induction_cost): Do not cost for pure SLP.
23171 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23172 of induction in inner loop vectorization.
23173 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23174 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23175 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23177 (vect_analyze_slp_cost_1): Cost induction.
23178 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23179 (vect_get_slp_vect_defs): Likewise.
23180 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23181 (vect_transform_stmt): Handle SLP reductions.
23182 * tree-vectorizer.h (vectorizable_induction): Adjust.
23184 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23186 * config/rs6000/rs6000.c (make_resolver_func): Update
23187 init_lowered_empty_function call.
23189 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23191 * doc/invoke.texi: Document the -fprofile-abs-path option.
23192 * common.opt (fprofile-abs-path): New option.
23193 * gcov-io.h (gcov_write_filename): Declare.
23194 * gcov-io.c (gcov_write_filename): New function.
23195 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23196 * profile.c (output_location): Likewise.
23198 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23200 * shring-wrap.c: Revert accidental commit.
23202 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23204 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23206 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23208 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23210 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23211 profile in callgraph edge.
23212 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23213 (apply_scale): Likewise.
23214 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23215 Also scale profile when inlining function with zero profile.
23216 (initialize_cfun): Update exit block profile even when it is zero.
23217 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23218 when profile is read.
23220 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23222 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23223 (CLONE_*): New constants to define the processors we can generate
23224 code for with the target_clone attribute.
23225 (rs6000_clone_map): New array to identify which clone processors
23226 the current program is running on.
23227 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23228 target_clone attribute.
23229 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23230 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23231 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23232 (cpu_expand_builtin): Add support for target_clone attribute.
23233 (rs6000_valid_attribute_p): Allow "default" attribute.
23234 (get_decl_name): New debug function to simplify printing the
23235 current function name in debugging statements.
23236 (rs6000_clone_priority): New functions to support the target_clone
23237 attribute, and be able to generate code to switch between ISA 2.05
23238 through ISA 3.0 (power6 through power9).
23239 (rs6000_compare_version_priority): Likewise.
23240 (rs6000_get_function_versions_dispatcher): Likewise.
23241 (make_resolver_func): Likewise.
23242 (add_condition_to_bb): Likewise.
23243 (dispatch_function_versions): Likewise.
23244 (rs6000_generate_version_dispatcher_body): Likewise.
23245 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23246 (fusion_gpr_load_p): Fix a spacing issue.
23247 * doc/extend.texi (Common Function Attributes): Document that the
23248 PowerPC supports the target_clone attribute.
23250 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23252 * config/arm/arm.h: explain F symbol found in description of ARM
23253 register allocation in its legend.
23255 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23257 * config/mips/frame-header-opt.c: Include profile-count.h.
23258 * config/riscv/riscv.c: Include profile-count.h
23260 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23262 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23264 (sm_set_flag_if_changed): Add bbs field.
23265 (execute_sm_if_changed_flag_set): Pass BBS.
23266 (execute_sm): Update.
23268 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23270 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
23273 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23275 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
23276 (peephole2): New peephole2 to emit the above.
23277 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
23279 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23281 * config/aarch64/aarch64.c (define_peephole2 above
23282 *sub_<shift>_<mode>): New peephole.
23284 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23286 * config/i386/i386.c (make_resolver_func): Update.
23287 * Makefile.in: Add profile-count.h and profile-count.o
23288 * auto-profile.c (afdo_indirect_call): Update to new API.
23289 (afdo_set_bb_count): Update.
23290 (afdo_propagate_edge): Update.
23291 (afdo_propagate_circuit): Update.
23292 (afdo_calculate_branch_prob): Update.
23293 (afdo_annotate_cfg): Update.
23294 * basic-block.h: Include profile-count.h
23295 (struct edge_def): Turn count to profile_count.
23296 (struct basic_block_def): Likewie.
23297 (REG_BR_PROB_BASE): Move to profile-count.h
23298 (RDIV): Move to profile-count.h
23299 * bb-reorder.c (max_entry_count): Turn to profile_count.
23300 (find_traces): Update.
23301 (rotate_loop):Update.
23302 (connect_traces):Update.
23303 (sanitize_hot_paths):Update.
23304 * bt-load.c (migrate_btr_defs): Update.
23305 * cfg.c (RDIV): Remove.
23306 (init_flow): Use alloc_block.
23307 (alloc_block): Uninitialize count.
23308 (unchecked_make_edge): Uninitialize count.
23309 (check_bb_profile): Update.
23310 (dump_edge_info): Update.
23311 (dump_bb_info): Update.
23312 (update_bb_profile_for_threading): Update.
23313 (scale_bbs_frequencies_int): Update.
23314 (scale_bbs_frequencies_gcov_type): Update.
23315 (scale_bbs_frequencies_profile_count): New.
23316 * cfg.h (update_bb_profile_for_threading): Update.
23317 (scale_bbs_frequencies_profile_count): Declare.
23318 * cfgbuild.c (compute_outgoing_frequencies): Update.
23319 (find_many_sub_basic_blocks): Update.
23320 * cfgcleanup.c (try_forward_edges): Update.
23321 (try_crossjump_to_edge): Update.
23322 * cfgexpand.c (expand_gimple_tailcall): Update.
23323 (construct_exit_block): Update.
23324 * cfghooks.c (verify_flow_info): Update.
23325 (dump_bb_for_graph): Update.
23326 (split_edge): Update.
23327 (make_forwarder_block): Update.
23328 (duplicate_block): Update.
23329 (account_profile_record): Update.
23330 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23331 (get_estimated_loop_iterations): Update.
23332 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23333 (single_likely_exit): Update.
23334 * cfgloopmanip.c (scale_loop_profile): Update.
23336 (set_zero_probability): Update.
23337 (lv_adjust_loop_entry_edge): Update.
23338 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23339 (purge_dead_edges): Update.
23340 (rtl_account_profile_record): Update.
23341 * cgraph.c (cgraph_node::create): Uninitialize count.
23342 (symbol_table::create_edge): Uninitialize count.
23343 (cgraph_update_edges_for_call_stmt_node): Update.
23344 (cgraph_edge::dump_edge_flags): Update.
23345 (cgraph_node::dump): Update.
23346 (cgraph_edge::maybe_hot_p): Update.
23347 * cgraph.h: Include profile-count.h
23348 (create_clone), create_edge, create_indirect_edge): Update.
23349 (cgraph_node): Turn count to profile_count.
23350 (cgraph_edge0: Likewise.
23351 (make_speculative, clone): Update.
23352 (create_edge): Update.
23353 (init_lowered_empty_function): Update.
23354 * cgraphclones.c (cgraph_edge::clone): Update.
23355 (duplicate_thunk_for_node): Update.
23356 (cgraph_node::create_clone): Update.
23357 * cgraphunit.c (cgraph_node::analyze): Update.
23358 (cgraph_node::expand_thunk): Update.
23359 * final.c (dump_basic_block_info): Update.
23360 * gimple-streamer-in.c (input_bb): Update.
23361 * gimple-streamer-out.c (output_bb): Update.
23362 * graphite.c (print_global_statistics): Update.
23363 (print_graphite_scop_statistics): Update.
23364 * hsa-brig.c: Include basic-block.h.
23365 * hsa-dump.c: Include basic-block.h.
23366 * hsa-gen.c (T sum_slice): Update.
23367 (convert_switch_statements):Update.
23368 * hsa-regalloc.c: Include basic-block.h.
23369 * ipa-chkp.c (chkp_produce_thunks): Update.
23370 * ipa-cp.c (struct caller_statistics): Update.
23371 (init_caller_stats): Update.
23372 (gather_caller_stats): Update.
23373 (ipcp_cloning_candidate_p): Update.
23374 (good_cloning_opportunity_p): Update.
23375 (get_info_about_necessary_edges): Update.
23376 (dump_profile_updates): Update.
23377 (update_profiling_info): Update.
23378 (update_specialized_profile): Update.
23379 (perhaps_add_new_callers): Update.
23380 (decide_about_value): Update.
23381 (ipa_cp_c_finalize): Update.
23382 * ipa-devirt.c (struct odr_type_warn_count): Update.
23383 (struct decl_warn_count): Update.
23384 (struct final_warning_record): Update.
23385 (possible_polymorphic_call_targets): Update.
23386 (ipa_devirt): Update.
23387 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23388 * ipa-icf.c (sem_function::merge): Update.
23389 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23390 * ipa-inline.c (compute_uninlined_call_time): Update.
23391 (compute_inlined_call_time): Update.
23392 (want_inline_small_function_p): Update.
23393 (want_inline_self_recursive_call_p): Update.
23394 (edge_badness): Update.
23395 (lookup_recursive_calls): Update.
23396 (recursive_inlining): Update.
23397 (inline_small_functions): Update.
23398 (dump_overall_stats): Update.
23399 (dump_inline_stats): Update.
23400 * ipa-profile.c (ipa_profile_generate_summary): Update.
23401 (ipa_propagate_frequency): Update.
23402 (ipa_profile): Update.
23403 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23404 * ipa-utils.c (ipa_merge_profiles): Update.
23405 * loop-doloop.c (doloop_modify): Update.
23406 * loop-unroll.c (report_unroll): Update.
23407 (unroll_loop_runtime_iterations): Update.
23408 * lto-cgraph.c (lto_output_edge): Update.
23409 (lto_output_node): Update.
23410 (input_node): Update.
23411 (input_edge): Update.
23412 (merge_profile_summaries): Update.
23413 * lto-streamer-in.c (input_cfg): Update.
23414 * lto-streamer-out.c (output_cfg): Update.
23415 * mcf.c (create_fixup_graph): Update.
23416 (adjust_cfg_counts): Update.
23417 (sum_edge_counts): Update.
23418 * modulo-sched.c (sms_schedule): Update.
23419 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23420 * predict.c (maybe_hot_count_p): Update.
23421 (probably_never_executed): Update.
23422 (dump_prediction): Update.
23423 (combine_predictions_for_bb): Update.
23424 (propagate_freq): Update.
23425 (handle_missing_profiles): Update.
23426 (counts_to_freqs): Update.
23427 (rebuild_frequencies): Update.
23428 (force_edge_cold): Update.
23429 * predict.h: Include profile-count.h
23430 (maybe_hot_count_p, counts_to_freqs): UPdate.
23431 * print-rtl-function.c: Do not include cfg.h
23432 * print-rtl.c: Include basic-block.h
23433 * profile-count.c: New file.
23434 * profile-count.h: New file.
23435 * profile.c (is_edge_inconsistent): Update.
23436 (correct_negative_edge_counts): Update.
23437 (is_inconsistent): Update.
23438 (set_bb_counts): Update.
23439 (read_profile_edge_counts): Update.
23440 (compute_frequency_overlap): Update.
23441 (compute_branch_probabilities): Update; Initialize and deinitialize
23443 (branch_prob): Update.
23444 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23445 (edge_gcov_count): New.
23446 (bb_gcov_count): New.
23447 * shrink-wrap.c (try_shrink_wrapping): Update.
23448 * tracer.c (better_p): Update.
23449 * trans-mem.c (expand_transaction): Update.
23450 (ipa_tm_insert_irr_call): Update.
23451 (ipa_tm_insert_gettmclone_call): Update.
23452 * tree-call-cdce.c: Update.
23453 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23454 (gimple_duplicate_sese_tail): Update.
23455 (gimple_account_profile_record): Update.
23456 (execute_fixup_cfg): Update.
23457 * tree-inline.c (copy_bb): Update.
23458 (copy_edges_for_bb): Update.
23459 (initialize_cfun): Update.
23460 (freqs_to_counts): Update.
23461 (copy_cfg_body): Update.
23462 (expand_call_inline): Update.
23463 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23464 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23465 (try_unroll_loop_completely): Update.
23466 (try_peel_loop): Update.
23467 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23468 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23469 * tree-ssa-loop-split.c (connect_loops): Update.
23470 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23471 * tree-ssa-reassoc.c (branch_fixup): Update.
23472 * tree-ssa-tail-merge.c (replace_block_by): Update.
23473 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23474 (compute_path_counts): Update.
23475 (update_profile): Update.
23476 (recompute_probabilities): Update.
23477 (update_joiner_offpath_counts): Update.
23478 (estimated_freqs_path): Update.
23479 (freqs_to_counts_path): Update.
23480 (clear_counts_path): Update.
23481 (ssa_fix_duplicate_block_edges): Update.
23482 (duplicate_thread_path): Update.
23483 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23484 (struct switch_conv_info): Update.
23485 * tree-tailcall.c (decrease_profile): Update.
23486 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23487 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23488 * value-prof.c (check_counter): Update.
23489 (gimple_divmod_fixed_value): Update.
23490 (gimple_mod_pow2): Update.
23491 (gimple_mod_subtract): Update.
23492 (gimple_ic_transform): Update.
23493 (gimple_stringop_fixed_value): Update.
23494 * value-prof.h (gimple_ic): Update.
23496 2017-06-02 Carl Love <cel@us.ibm.com>
23498 * config/rs6000/rs6000-c: Add support for built-in functions
23499 vector double vec_doublee (vector signed int);
23500 vector double vec_doublee (vector unsigned int);
23501 vector double vec_doublee (vector float);
23502 vector double vec_doubleh (vector signed int);
23503 vector double vec_doubleh (vector unsigned int);
23504 vector double vec_doubleh (vector float);
23505 vector double vec_doublel (vector signed int);
23506 vector double vec_doublel (vector unsigned int);
23507 vector double vec_doublel (vector float);
23508 vector double vec_doubleo (vector signed int);
23509 vector double vec_doubleo (vector unsigned int);
23510 vector double vec_doubleo (vector float);.
23511 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23512 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23514 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23515 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23516 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23518 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23519 vec_doublel, vec_doubleh.
23520 * doc/extend.texi: Update the built-in documentation file for the
23521 new built-in functions.
23523 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23526 * ipa-inline-analysis.c (free_growth_caches): Set
23527 edge_removal_hook_holder to NULL after removing it.
23529 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23531 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23532 comparision with zero.
23534 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23535 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23536 for early expansion of vec_min and vec_max builtins.
23537 (builtin_function_type): Add min/max unsigned variants to those
23538 identified as having unsigned arguments.
23540 2017-06-02 Olivier Hainque <hainque@adacore.com>
23542 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23544 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23546 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23547 Use VALL_F16 iterator rather than VALL.
23549 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23551 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23552 Emit CBNZ inside loop when doing a strong exchange and comparing
23553 against zero. Generate the CC flags after the loop.
23555 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23557 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23558 (dl_section_ref): New.
23559 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23560 On AIX, append an expression to subtract the size of the
23561 section length to dl_section_ref.
23563 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23565 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23566 for early expansion of vector absolute builtins.
23568 2017-06-02 Richard Biener <rguenther@suse.de>
23570 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23571 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23573 2017-06-02 Richard Biener <rguenther@suse.de>
23575 PR tree-optimization/80948
23576 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23577 stmt order as well.
23579 2017-06-02 Richard Biener <rguenther@suse.de>
23581 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23583 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23584 for inductions as relevant.
23586 2017-06-02 Richard Biener <rguenther@suse.de>
23588 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23589 (vectorizable_induction): ... this. Remove dead code.
23591 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23593 * builtins. (expand_builtin_alloca): Remove second parameter and
23594 infer its value from the first parameter instead.
23595 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23597 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23599 PR rtl-optimization/80903
23600 * loop-doloop.c (add_test): Unshare sequence.
23602 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23604 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23606 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23608 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23610 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23611 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23612 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23613 (xlogue_layout::get_stub_rtx): Make static.
23614 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23615 (xlogue_layout::compute_stub_managed_regs): Rename to...
23616 (xlogue_layout::count_stub_managed_regs): ...this.
23617 (xlogue_layout::is_stub_managed_reg): New function.
23618 (xlogue_layout::m_stub_names): Rename to...
23619 (xlogue_layout::s_stub_names): ...this, make static.
23620 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23621 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23622 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23623 xlogue_layout::s_stub_names): Instantiate statics.
23624 (stub_managed_regs): Remove.
23625 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23626 (disable_call_ms2sysv_xlogues): Rename to...
23627 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23628 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23630 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23631 change after reload_completed.
23632 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23634 (ix86_expand_prologue): Likewise.
23635 (ix86_expand_epilogue): Likewise.
23636 (ix86_expand_split_stack_prologue): Likewise.
23637 (ix86_compute_frame_layout): Remove frame parameter ...
23638 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23639 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23641 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23642 (ix86_frame): Move from here ...
23643 * config/i386/i386.h (ix86_frame): ... to here.
23644 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23645 complete ix86_frame data structure instead. Remove some_ld_name.
23647 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23649 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23650 symbols that hold a DECL_VALUE_EXPR.
23652 2017-06-01 Martin Jambor <mjambor@suse.cz>
23654 PR tree-optimization/80898
23655 * tree-sra.c (process_subtree_disqualification): Removed.
23656 (disqualify_candidate): Do not acll
23657 process_subtree_disqualification.
23658 (subtree_mark_written_and_enqueue): New function.
23659 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23660 RHS has been disqualified and re-queue LHS if necessary. Apart
23661 from that, ignore disqualified RHS.
23663 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23665 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23666 address fetch for z10 or later.
23668 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23670 * config/arc/arc.md (tst_movb): Add guard when splitting.
23672 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23674 * config/arc/arc.c (arc_can_eliminate): Test against
23675 arc_frame_pointer_needed.
23677 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23679 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23680 to prevent store reordering.
23681 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23682 (type): Add block type.
23683 (stack_tie): Define special instruction to be used in
23686 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23688 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23689 constraint. It is not valid for the pattern.
23690 (noncommutative_binary_comparison): Likewise.
23692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23694 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23697 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23699 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23700 be used by the reg-alloc.
23702 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23704 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23705 reg-alloc when having mul64 or mul32x16 instructions.
23706 (mulsidi3): Likewise.
23707 (umulsidi3): Likewise.
23708 (mulsi32x16): New pattern.
23709 (mulsi64): Likewise.
23710 (mulsidi64): Likewise.
23711 (umulsidi64): Likewise.
23712 (MUL32x16_REG): Define.
23713 (mul64_600): Use MUL32x16_REG.
23714 (mac64_600): Likewise.
23715 (umul64_600): Likewise.
23716 (umac64_600): Likewise.
23718 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23720 * config/arc/arc.md (mulsi3_700): Make it commutative.
23722 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23724 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23726 (*sign_extendsidi2_insn): Likewise for movstosw.
23728 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23730 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23731 the type of the input discriminant value. Convert the
23732 discriminant value of signedness vary.
23734 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23736 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23737 Add to -Wall section.
23739 2017-06-01 Richard Biener <rguenther@suse.de>
23741 PR middle-end/66313
23742 * fold-const.c (fold_plusminus_mult_expr): If the factored
23743 factor may be zero use a wrapping type for the inner operation.
23744 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23745 and handle moved defs.
23746 (process_assignment): Properly guard the unary op case. Return a
23747 tri-state indicating that moving the stmt before the call may allow
23748 to continue. Pass through to_move.
23749 (find_tail_calls): Handle moving unrelated defs before
23752 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23755 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23756 splitter result in the canonical way.
23758 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23760 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23761 also for 32bit target. Update insn attributes.
23762 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23764 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23766 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23767 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23768 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23769 (_mm_maskz_min_ss): New intrinsics.
23771 2017-05-31 Martin Liska <mliska@suse.cz>
23773 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23774 Change comment style to one we normally use.
23775 (vectorizable_reduction): Likewise.
23776 (vectorizable_induction): Likewise.
23777 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23778 (vectorizable_call): Likewise.
23779 (vectorizable_simd_clone_call): Likewise.
23780 (vectorizable_conversion): Likewise.
23781 (vectorizable_assignment): Likewise.
23782 (vectorizable_shift): Likewise.
23783 (vectorizable_operation): Likewise.
23784 (vectorizable_store): Likewise.
23785 (vectorizable_load): Likewise.
23786 * tree-vectorizer.h: Likewise.
23788 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23790 * passes.c (emergency_dump_function): New.
23791 * tree-pass.h (emergency_dump_function): Declare.
23792 * plugin.c (plugins_internal_error_function): Remove.
23793 * plugin.h (plugins_internal_error_function): Remove declaration.
23794 * toplev.c (internal_error_function): New static function. Use it...
23795 (general_init): ...here.
23797 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23799 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23800 (arc_rtx_costs): Add costs for new patterns.
23801 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23802 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23804 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23806 * tree-ssa-strlen.c (get_next_strinfo): New function.
23807 (get_stridx_plus_constant): Use it.
23808 (zero_length_string): Likewise.
23809 (adjust_related_strinfos): Likewise.
23810 (adjust_last_stmt): Likewise.
23812 2017-05-31 Richard Biener <rguenther@suse.de>
23815 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23816 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23818 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23820 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23821 loop_vinfo argument and use of dependence distance vectors.
23822 Check instead whether the two references differ only in their
23823 initial value and assume that they have the same alignment if the
23824 difference is a multiple of the vector alignment.
23825 (vect_analyze_data_refs_alignment): Update call accordingly.
23827 2017-05-31 Martin Liska <mliska@suse.cz>
23830 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23832 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23834 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23835 (create_intersect_range_checks): Move from ...
23836 * tree-data-ref.c (create_intersect_range_checks_index)
23837 (create_intersect_range_checks): ... to here.
23838 (create_runtime_alias_checks): New function factored from ...
23839 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23840 here. Call above function.
23841 * tree-data-ref.h (create_runtime_alias_checks): New function.
23843 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23845 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23846 segment length for dr_b and compute it in wide_int.
23848 2017-05-31 Richard Biener <rguenther@suse.de>
23850 PR tree-optimization/80906
23851 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23852 and pass through iv_map.
23853 (copy_bb_and_scalar_dependences): Adjust.
23854 (translate_pending_phi_nodes): Likewise.
23855 (copy_loop_close_phi_args): Handle code-generating IVs instead
23858 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23860 * diagnostic-color.c (color_dict): Add "type-diff".
23861 (parse_gcc_colors): Update comment.
23862 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23863 -fdiagnostics-show-template-tree and -fno-elide-type.
23864 (GCC_COLORS): Add type-diff to example.
23866 (-fdiagnostics-show-template-tree): New.
23867 (-fno-elide-type): New.
23868 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23869 the pp_format_decoder callback. Call any m_format_postprocessor's
23871 (pretty_printer::pretty_printer): Initialize
23872 m_format_postprocessor.
23873 (pretty_printer::~pretty_printer): Delete any
23874 m_format_postprocessor.
23875 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23876 (class format_postprocessor): New class.
23877 (struct pretty_printer::format_decoder): Document the new parameters.
23878 (struct pretty_printer::m_format_postprocessor): New field.
23879 * tree-diagnostic.c (default_tree_printer): Update for new
23880 bool and const char ** params.
23881 * tree-diagnostic.h (default_tree_printer): Likewise.
23883 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23885 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23886 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23888 rs6000_gen_cell_microcode code.
23889 (rs6000_final_prescan_insn): Delete.
23890 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23891 "warn-cell-microcode" entries.
23892 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23893 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23894 throughout. Change cc_reg_not_micro_cr0_operand to
23895 cc_reg_not_cr0_operand throughout.
23896 (*extendhi<mode>2_noload): Delete.
23897 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23898 (mwarn-cell-microcode): Delete.
23899 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23900 -mgen-cell-microcode and -mwarn-cell-microcode.
23902 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23905 * config/i386/constraints.md (Yd): New constraint.
23907 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23908 and (?Yd, r) alternatives. Update insn attributes.
23909 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23910 and (?*Yd, r) alternatives. Update insn attributes.
23911 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23913 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23915 * gimplify.c (gimplify_modify_expr): Don't create a
23916 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23919 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23921 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23923 2017-05-30 Richard Biener <rguenther@suse.de>
23925 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23926 and reduc_def fields.
23927 (STMT_VINFO_REDUC_TYPE): New define.
23928 (STMT_VINFO_REDUC_DEF): Likewise.
23929 (vect_force_simple_reduction): Adjust prototype.
23930 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23931 (vect_is_simple_reduction): Remove check_reduction argument.
23932 (vect_force_simple_reduction): Adjust and set
23933 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23934 (vectorizable_reduction): Do not re-do reduction analysis
23935 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23936 * tree-parloops.c (gather_scalar_reductions): Adjust.
23938 2017-05-30 Richard Biener <rguenther@suse.de>
23940 PR middle-end/80901
23941 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23944 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23946 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23947 Introduce unknown_misalignment parameter and remove vf.
23948 (vect_peeling_hash_get_lowest_cost):
23949 Pass unknown_misalignment parameter.
23950 (vect_enhance_data_refs_alignment):
23951 Fix unsupportable data ref treatment.
23953 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23955 * tree-vect-data-refs.c (vect_get_data_access_cost):
23956 Workaround for SLP handling.
23957 (vect_enhance_data_refs_alignment):
23958 Compute costs for doing no peeling at all, compare to the best
23959 peeling costs so far and avoid peeling if cheaper.
23961 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23963 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23964 Return peeling info and set costs to zero for unlimited cost
23966 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23967 with unknown misalignment. Compute and costs for unknown
23968 misalignment, compare them to the costs for known misalignment
23969 and choose the cheapest for peeling.
23971 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23973 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23974 (vect_get_peeling_costs_all_drs): Create function.
23975 (vect_peeling_hash_get_lowest_cost):
23976 Use vect_get_peeling_costs_all_drs.
23977 (vect_peeling_supportable): Create function.
23978 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23980 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23982 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23983 DR_HAS_NEGATIVE_STEP.
23984 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23985 (vect_enhance_data_refs_alignment): Use.
23986 (vect_duplicate_ssa_name_ptr_info): Use.
23987 * tree-vectorizer.h (dr_misalignment): Use.
23988 (known_alignment_for_access_p): Use.
23990 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23993 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23995 (has_section_name): New function.
23997 2017-05-30 Martin Liska <mliska@suse.cz>
24000 * auto-profile.c (get_function_decl_from_block): Fix
24003 2017-05-30 Richard Biener <rguenther@suse.de>
24005 PR middle-end/80876
24006 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24008 2017-05-30 Martin Liska <mliska@suse.cz>
24010 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24011 * dumpfile.h (struct dump_file_info): Remove ctors.
24013 2017-05-30 Martin Liska <mliska@suse.cz>
24015 * predict.def: Fix GNU coding style.
24017 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24019 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24020 Mark 'to' argument with ATTRIBUTE_UNUSED.
24022 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24024 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24025 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24026 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24029 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24031 * doc/install.texi (Options specification): Restore entry of
24032 --enable-sjlj-exceptions.
24034 2017-05-27 Michael Eager <eager@eagercon.com>
24037 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24039 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24041 * config/microblaze/microblaze.h
24042 (FIXED_REGISTERS): Update in macro.
24043 (CALL_USED_REGISTERS): Update in macro.
24045 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24047 * doc/install.texi: Add links to macOS binary distributions.
24049 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24053 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24055 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24057 2017-05-26 Martin Liska <mliska@suse.cz>
24059 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24061 2017-05-26 Martin Liska <mliska@suse.cz>
24063 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24064 always leading ';; '.
24065 (dump_bb_info): Likewise.
24066 (brief_dump_cfg): Likewise.
24067 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24068 * dumpfile.c: Remove usage of TDF_VERBOSE.
24069 * dumpfile.h (enum dump_kind): Likewise.
24070 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24071 * print-tree.c (debug_verbose): Remove.
24072 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24073 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24074 * tree-diagnostic.c (default_tree_printer): Replace
24075 TDF_DIAGNOSTIC with TDF_SLIM.
24077 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24079 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24080 in parameter loop, rather than loop_vinfo.
24081 (create_intersect_range_checks): Ditto.
24082 (vect_create_cond_for_alias_checks): Update call to above functions.
24084 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24086 PR tree-optimization/80815
24087 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24088 for merging runtime alias checks. Handle negative DR_STEPs.
24090 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24092 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24094 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24095 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24096 out code pruning runtime alias checks.
24097 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24098 factored out from above.
24099 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24101 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24103 (prune_runtime_alias_test_list): New decalaration.
24105 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24107 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24108 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24109 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24110 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24111 (operator==, comp_dr_with_seg_len_pair): Ditto.
24112 (vect_prune_runtime_alias_test_list): Ditto.
24114 2017-05-26 Martin Liska <mliska@suse.cz>
24117 * params.def: Bound partial-inlining-entry-probability param.
24119 2017-05-26 Marek Polacek <polacek@redhat.com>
24122 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24125 2017-05-26 Richard Biener <rguenther@suse.de>
24127 PR tree-optimization/80842
24128 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24131 2017-05-26 Richard Biener <rguenther@suse.de>
24133 PR tree-optimization/80844
24134 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24136 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24138 * doc/md.texi (Machine Constraints): Update x86 family
24139 machine constraints section to match 'config/i386/constraints.md'.
24141 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24143 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24145 2017-05-25 Nathan Sidwell <nathan@acm.org>
24147 * doc/invoke.texi (--enable-languages): Update documentation.
24149 2017-05-25 Martin Liska <mliska@suse.cz>
24151 * dumpfile.c: Add TDF_FOLDING.
24152 * dumpfile.h (enum dump_kind): Likewise.
24153 * genmatch.c (dt_simplify::gen_1): Use it.
24155 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24157 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24159 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24161 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24162 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24164 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24166 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24167 * match.pd (X == C): Rewrite it here.
24168 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24169 with_certain_nonzero_bits2): New predicates.
24170 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24172 2017-05-24 Nathan Sidwell <nathan@acm.org>
24174 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24177 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24180 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24182 * config/powerpcspe: New port. Files are copied from the rs6000
24183 port, with "rs6000" in filenames replaced by "powerpcspe".
24185 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24187 PR rtl-optimization/80754
24188 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24190 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24192 * config/sparc/sparc.md (length): Return the correct value for -mflat
24193 sibcalls to match output_sibcall.
24195 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24199 * config/rs6000/rs6000.c (struct machine_function): Add new field
24201 (rs6000_get_separate_components): Init that field, use it.
24202 (rs6000_components_for_bb): Use the field.
24204 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24206 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24208 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24210 PR middle-end/80823
24211 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24213 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24216 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24217 address against address_operand predicate.
24218 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24220 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24222 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24223 parameters passed indirectly.
24225 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24227 * config/i386/i386.md (*movdi_internal): Remove SSE4
24228 alternative 18 (?r, *v). Update insn attributes.
24229 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24230 Update insn attributes.
24231 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24232 Update insn attributes.
24233 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24234 alternative 1 (r, v). Remove isa attribute.
24235 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24236 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24237 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24239 2017-05-23 Tom de Vries <tom@codesourcery.com>
24241 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24244 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24246 * cgraphunit.c (symbol_table::process_new_functions): Update.
24247 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24248 (inline_generate_summary): Rename to ...
24249 (ipa_fn_summary_generate): ... this one.
24250 (inline_read_summary): Rename to ...
24251 (ipa_fn_summary_read): ... this one.
24252 (inline_write_summary): Rename to ...
24253 (ipa_fn_summary_write): ... this one.
24254 (inline_free_summary): Rename to ...
24255 (ipa_free_fn_summary): ... this one.
24256 (pass_data_local_fn_summary, pass_local_fn_summary,
24257 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24258 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24259 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24260 make_pass_ipa_fn_summary): New.
24261 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24262 inline_write_summary, inline_free_summary): Remove.
24263 (ipa_free_fn_summary) : New.
24264 * ipa-inline.c (ipa_inline): Update.
24265 (pass_ipa_inline): Do not generate summaries.
24266 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24268 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
24269 and add pass_ipa_fn_summary.
24270 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
24272 (make_pass_inline_parameters): Remove.
24274 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
24276 * omp-low.c (struct omp_context): Remove "default_kind" member.
24279 * omp-offload.c (execute_oacc_device_lower): Remove the
24280 parallelism dimensions function attributes for unparallelized
24281 OpenACC kernels constructs.
24283 2017-05-23 Martin Liska <mliska@suse.cz>
24285 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
24287 (cgraph_edge::make_speculative): Likewise.
24288 (cgraph_edge::resolve_speculation): Likewise.
24289 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24290 (cgraph_node::dump): Likewise.
24291 * cgraph.h: Likewise.
24292 * cgraphunit.c (analyze_functions): Likewise.
24293 (symbol_table::compile): Likewise.
24294 * ipa-cp.c (print_all_lattices): Likewise.
24295 (determine_versionability): Likewise.
24296 (initialize_node_lattices): Likewise.
24297 (ipcp_verify_propagated_values): Likewise.
24298 (estimate_local_effects): Likewise.
24299 (update_profiling_info): Likewise.
24300 (create_specialized_node): Likewise.
24301 (perhaps_add_new_callers): Likewise.
24302 (decide_about_value): Likewise.
24303 (decide_whether_version_node): Likewise.
24304 (identify_dead_nodes): Likewise.
24305 (ipcp_store_bits_results): Likewise.
24306 * ipa-devirt.c (dump_targets): Likewise.
24307 (ipa_devirt): Likewise.
24308 * ipa-icf.c (sem_item::dump): Likewise.
24309 (sem_function::equals): Likewise.
24310 (sem_variable::equals): Likewise.
24311 (sem_item_optimizer::read_section): Likewise.
24312 (sem_item_optimizer::execute): Likewise.
24313 (congruence_class::dump): Likewise.
24314 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24315 (dump_inline_summary): Likewise.
24316 (estimate_node_size_and_time): Likewise.
24317 (inline_analyze_function): Likewise.
24318 * ipa-inline-transform.c (inline_call): Likewise.
24319 * ipa-inline.c (report_inline_failed_reason): Likewise.
24320 (want_early_inline_function_p): Likewise.
24321 (edge_badness): Likewise.
24322 (update_edge_key): Likewise.
24323 (inline_small_functions): Likewise.
24324 * ipa-profile.c (ipa_profile): Likewise.
24325 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24326 (ipa_make_edge_direct_to_target): Likewise.
24327 (remove_described_reference): Likewise.
24328 (ipa_impossible_devirt_target): Likewise.
24329 (propagate_controlled_uses): Likewise.
24330 (ipa_print_node_params): Likewise.
24331 (ipcp_transform_function): Likewise.
24332 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24333 (propagate_pure_const): Likewise.
24334 * ipa-reference.c (generate_summary): Likewise.
24335 (read_write_all_from_decl): Likewise.
24336 (propagate): Likewise.
24337 (ipa_reference_read_optimization_summary): Likewise.
24338 * ipa-utils.c (ipa_merge_profiles): Likewise.
24339 * ipa.c (walk_polymorphic_call_targets): Likewise.
24340 (symbol_table::remove_unreachable_nodes): Likewise.
24341 (ipa_single_use): Likewise.
24342 * passes.c (execute_todo): Likewise.
24343 * predict.c (drop_profile): Likewise.
24344 * symtab.c (symtab_node::get_dump_name): New function.
24345 (symtab_node::dump_name): Likewise.
24346 (symtab_node::dump_asm_name): Likewise.
24347 (symtab_node::dump_references): Likewise.
24348 (symtab_node::dump_referring): Likewise.
24349 (symtab_node::dump_base): Likewise.
24350 (symtab_node::debug_symtab): Likewise.
24351 * tree-sra.c (convert_callers_for_node): Likewise.
24352 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24353 * value-prof.c (init_node_map): Likewise.
24355 2017-05-23 Martin Liska <mliska@suse.cz>
24357 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24358 and symtab_node::debug_symtab to symbol_table::debug.
24359 * cgraphunit.c (analyze_functions): Use the renamed function.
24360 (symbol_table::compile): Likewise.
24361 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24362 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24363 * passes.c (execute_todo): Likewise.
24364 * symtab.c (symbol_table::dump): New function.
24365 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24367 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24369 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24370 that nonconst implies exec.
24372 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24374 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24375 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24376 (inline_edge_summary_vec): Turn into ...
24377 (ipa_call_summaries): ... this one.
24378 (redirect_to_unreachable, edge_set_predicate,
24379 evaluate_properties_for_edge, inline_summary_alloc,
24380 reset_ipa_call_summary, reset_inline_summary,
24381 inline_summary_t::duplicate): Update.
24382 (inline_edge_duplication_hook): Turn to ...
24383 (ipa_call_summary_t::duplicate): ... this one.
24384 (inline_edge_removal_hook): Turn to ...
24385 (ipa_call_summary_t::remove): ... this one.
24386 (dump_inline_edge_summary): Turn to ...
24387 (dump_ipa_call_summary): ... this one.
24388 (estimate_function_body_sizes): Update.
24389 (inline_update_callee_summaries): Update.
24390 (remap_edge_change_prob): Update.
24391 (remap_edge_summaries): Update.
24392 (inline_merge_summary): Update.
24393 (do_estimate_edge_time): Update.
24394 (inline_generate_summary): Update.
24395 (inline_read_section): Update.
24396 (inline_read_summary): Update.
24397 (inline_free_summary): Update.
24398 * ipa-inline.c (can_inline_edge_p): Update.
24399 (compute_inlined_call_time): Update.
24400 (want_inline_small_function_p): Update.
24401 (edge_badness): Update.
24402 (early_inliner): Update.
24403 * ipa-inline.h (inline_edge_summary): Turn to ...
24404 (ipa_call_summary): ... this one.
24405 (ipa_call_summary_t): New class.
24406 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24407 (ipa_call_summaries): New.
24408 (inline_edge_summary): Remove.
24409 (estimate_edge_growth): Update.
24410 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24411 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24412 * ipa-split.c (execute_split_functions): Update.
24413 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24415 2017-05-23 Tom de Vries <tom@codesourcery.com>
24417 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24418 attributes): Document rdrand effective target.
24420 2017-05-23 Tom de Vries <tom@codesourcery.com>
24422 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24423 attributes): Sort alphabetically.
24425 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24427 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24429 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24432 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24433 V2DF/V2DI splat into two separate patterns, one that handles
24434 registers, and the other that only handles memory. Drop support
24435 for splatting from a GPR on ISA 2.07 and then splitting the
24436 splat into direct move and splat.
24437 (vsx_splat_<mode>_reg): Likewise.
24438 (vsx_splat_<mode>_mem): Likewise.
24440 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24442 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24444 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24446 PR middle-end/80809
24447 * omp-low.c (finish_taskreg_remap): New function.
24448 (finish_taskreg_scan): If unit size of ctx->record_type
24449 is non-constant, unshare the size expression and replace
24450 decls in it with possible outer var refs.
24452 PR middle-end/80809
24453 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24454 GOVD_SHARED rather than GOVD_PRIVATE with it.
24455 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24456 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24458 PR middle-end/80853
24459 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24460 as last argument to build_outer_var_ref for pointer bases of array
24461 section reductions.
24463 2017-05-19 Martin Sebor <msebor@redhat.com>
24465 * print-tree.c (print_node): Print DECL_READ_P flag.
24467 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24469 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24470 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24471 * cgraph.c: Likewise.
24472 * cgraphunit.c: Likewise.
24473 * gengtype.c: Likewise.
24474 * ipa-cp.c: Likewise.
24475 * ipa-devirt.c: Likewise.
24476 * ipa-icf.c: Likewise.
24477 * ipa-predicate.c: Likewise.
24478 * ipa-profile.c: Likewise.
24479 * ipa-prop.c: Likewise.
24480 * ipa-split.c: Likewise.
24482 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24483 edge_predicate_pool, dump_inline_hints,
24484 inline_summary::account_size_time, redirect_to_unreachable,
24485 edge_set_predicate, set_hint_predicate,
24486 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24487 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24488 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24489 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24490 ipa_call_summary_t::remove, initialize_growth_caches,
24491 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24492 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24493 mark_modified, unmodified_parm_1, unmodified_parm,
24494 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24495 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24496 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24497 will_be_nonconstant_predicate, record_modified_bb_info,
24498 get_minimal_bb, record_modified, param_change_prob,
24499 phi_result_unknown_predicate, predicate_for_phi_result,
24500 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24501 estimate_function_body_sizes, compute_inline_parameters,
24502 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24503 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24504 inline_update_callee_summaries, remap_edge_change_prob,
24505 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24506 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24507 inline_analyze_function, inline_summary_t::insert,
24508 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24509 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24510 inline_free_summary): Move to ipa-fnsummary.h
24511 (predicate_t): Remove.
24512 * ipa-fnsummary.c: New file.
24513 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24514 (enum inline_hints_vals, inline_hints, agg_position_info,
24515 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24516 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24517 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24518 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24519 inline_read_summary, inline_write_summary, inline_free_summary,
24520 inline_analyze_function, initialize_inline_failed,
24521 inline_merge_summary, inline_update_overall_summary,
24522 compute_inline_parameters): Move to ipa-fnsummary.h
24523 * ipa-fnsummary.h: New file.
24524 * ipa-inline-transform.h: Include ipa-inline.h.
24525 * ipa-inline.c: LIkewise.
24527 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24529 * ipa-inline.c (edge_badness): Use inlined_time instead of
24530 inline_summaries->get.
24532 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24534 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24536 2017-05-22 Nathan Sidwell <nathan@acm.org>
24538 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24539 (fdump-lang): Document 'raw' option.
24540 * dumpfile.h (TDI_tu): Delete.
24541 * dumpfile.c (dump_files): Remove translation-unit.
24542 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24544 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24546 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24547 command option from $(AWK) call.
24548 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24549 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24550 [FORMAT]: Remove handling of variable.
24551 * config/avr/t-multilib: Regenerate.
24553 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24555 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24557 (dump_inline_summary): Do not print self_time.
24558 (estimate_function_body_sizes): Do not set self_time.
24559 (compute_inline_parameters): Likewise.
24560 (inline_read_section, inline_write_summary): Do not stream self_time.
24561 * ipa-inline.h (inline_summary): Drop self_time.
24563 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24565 * ipa-inline-analysis.c (account_size_time): Rename to ...
24566 (inline_summary::account_size_time): ... this one.
24567 (reset_ipa_call_summary): Turn to ...
24568 (ipa_call_summary::reset): ... this one.
24569 (reset_inline_summary): Turn to ...
24570 (inline_summary::reset): ... this one.
24571 (inline_summary_t::remove): Update.
24572 (inline_summary_t::duplicate): Update.
24573 (ipa_call_summary_t::remove): Update.
24574 (dump_inline_summary): Update.
24575 (estimate_function_body_sizes): Update.
24576 (compute_inline_parameters): Update.
24577 (estimate_node_size_and_time): Update.
24578 (inline_merge_summary): Update.
24579 (inline_update_overall_summary): Update.
24580 (inline_read_section): Update.
24581 (inline_write_summary): Update.
24582 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24583 add account_size_time and reset member functions.
24584 (ipa_call_summary): Add reset function.
24585 * ipa-predicate.h (predicate::operator &): Constify.
24587 2017-05-22 Richard Biener <rguenther@suse.de>
24589 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24591 2017-05-19 Jason Merrill <jason@redhat.com>
24593 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24595 2017-05-19 Marek Polacek <polacek@redhat.com>
24598 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24599 TYPE_OVERFLOW_WRAPS checks.
24601 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24603 * tree-core.h (enum omp_clause_default_kind): Add
24604 "OMP_CLAUSE_DEFAULT_PRESENT".
24605 * tree-pretty-print.c (dump_omp_clause): Handle it.
24606 * gimplify.c (enum gimplify_omp_var_data): Add
24607 "GOVD_MAP_FORCE_PRESENT".
24608 (gimplify_adjust_omp_clauses_1): Map it to
24609 "GOMP_MAP_FORCE_PRESENT".
24610 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24612 * gimplify.c (oacc_default_clause): Clarify.
24614 2017-05-19 Nathan Sidwell <nathan@acm.org>
24616 LANG_HOOK_REGISTER_DUMPS
24617 * toplev.c (general_init): Call register dump lang hook.
24618 * doc/invoke.texi: Document -fdump-lang option family.
24619 * dumpfile.c (dump_files): Remove class dump here.
24620 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24621 * dumpfile.h (tree_dump_index): Remove TDI_class.
24622 * langhooks-def.h (lhd_register_dumps): Declare.
24623 (LANG_HOOKS_REGISTER_DUMPS): Define.
24624 (LANG_HOOKS_INITIALIZER): Add it.
24625 * langhooks.c (lhd_register_dumps): Define.
24626 * langhooks.h (struct lang_hooks): Add register_dumps.
24628 2017-05-19 Nathan Sidwell <nathan@acm.org>
24630 * context.h (context::set_passes): New.
24631 * context.c (context::context): Do not create pass manager.
24632 * toplev.c (general_init): Create pass manager here.
24634 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24636 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24637 use this splitter if two add or or instructions would also work for
24638 the constant we want to generate.
24640 2017-05-19 Richard Biener <rguenther@suse.de>
24643 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24644 predicate evaluation.
24646 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24648 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24650 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24653 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24655 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24656 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24657 (inline_edge_summary_vec): Turn into ...
24658 (ipa_call_summaries): ... this one.
24659 (redirect_to_unreachable, edge_set_predicate,
24660 evaluate_properties_for_edge, inline_summary_alloc,
24661 reset_ipa_call_summary, reset_inline_summary,
24662 inline_summary_t::duplicate): Update.
24663 (inline_edge_duplication_hook): Turn to ...
24664 (ipa_call_summary_t::duplicate): ... this one.
24665 (inline_edge_removal_hook): Turn to ...
24666 (ipa_call_summary_t::remove): ... this one.
24667 (dump_inline_edge_summary): Turn to ...
24668 (dump_ipa_call_summary): ... this one.
24669 (estimate_function_body_sizes): Update.
24670 (inline_update_callee_summaries): Update.
24671 (remap_edge_change_prob): Update.
24672 (remap_edge_summaries): Update.
24673 (inline_merge_summary): Update.
24674 (do_estimate_edge_time): Update.
24675 (inline_generate_summary): Update.
24676 (inline_read_section): Update.
24677 (inline_read_summary): Update.
24678 (inline_free_summary): Update.
24679 * ipa-inline.c (can_inline_edge_p): Update.
24680 (compute_inlined_call_time): Update.
24681 (want_inline_small_function_p): Update.
24682 (edge_badness): Update.
24683 (early_inliner): Update.
24684 * ipa-inline.h (inline_edge_summary): Turn to ...
24685 (ipa_call_summary): ... this one.
24686 (ipa_call_summary_t): New class.
24687 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24688 (ipa_call_summaries): New.
24689 (inline_edge_summary): Remove.
24690 (estimate_edge_growth): Update.
24691 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24692 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24693 * ipa-split.c (execute_split_functions): Update.
24694 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24696 2017-05-19 Richard Biener <rguenther@suse.de>
24698 PR middle-end/80764
24699 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24701 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24703 * config/rs6000/rs6000.c (struct machine_function): Add field
24704 fpr_is_wrapped_separately.
24705 (rs6000_get_separate_components): Use 64 components. Handle the
24706 new FPR components.
24707 (rs6000_components_for_bb): Handle the FPR components.
24708 (rs6000_emit_prologue_components): Handle the FPR components.
24709 (rs6000_emit_epilogue_components): Handle the FPR components.
24710 (rs6000_set_handled_components): Handle the FPR components.
24711 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24712 that are already separately shrink-wrapped.
24713 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24714 that are already separately shrink-wrapped.
24716 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24719 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24722 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24723 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24724 cases where the register allocator uses a move and an offsettable
24725 memory operation to/from a FPR register on ISA 2.06/2.07.
24726 (define_peephole2 for Altivec d-form store): Likewise.
24728 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24731 * config/i386/mmx.md (*mov<mode>_internal): Enable
24732 alternatives 11, 12, 13 and 14 also for 32bit targets.
24733 Remove alternatives 15, 16, 17 and 18.
24734 * config/i386/sse.md (vec_concatv2di): Change
24735 alternative (!x, *y) to (x, ?!*Yn).
24737 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24739 * dumpfile.h (enum dump_kind): Remove stray comma.
24741 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24743 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24744 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24745 predicate::num_conditions
24746 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24747 (CHANGED): turn into predicate::changed.
24748 (agg_position_info): Move to ipa-predicate.h
24749 (add_condition, predicate::add_clause, predicate::operator &=,
24750 predicate::or_with, predicate::evaluate, predicate::probability,
24751 dump_condition, dump_clause, predicate::dump,
24752 predicate::remap_after_duplication, predicate::remap_after_inlining,
24753 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24754 (evaluate_conditions_for_known_args): Update.
24755 (set_cond_stmt_execution_predicate): Update.
24756 * ipa-inline.h: Include ipa-predicate.h
24757 (condition, inline_param_summary, conditions, agg_position_info,
24758 predicate): Move to ipa-predicate.h
24759 * ipa-predicate.c: New file.
24760 * ipa-predicate.h: New file.
24762 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24764 * final.c (leaf_function_p): Check we are not in a sequence.
24766 2017-05-18 Martin Liska <mliska@suse.cz>
24768 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24769 * dumpfile.c (dump_register): Use new enum dump_kind.
24770 (get_dump_file_name): Likewise.
24771 (dump_enable_all): Likewise.
24772 (dump_switch_p_1): Likewise.
24773 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24774 * dumpfile.h (enum dump_kind): New enum type.
24775 (struct dump_file_info): Create constructor and
24776 format fields and comments.
24777 * passes.c (pass_manager::register_one_dump_file):
24779 * statistics.c (statistics_early_init): Likewise.
24780 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24781 TDF_TREE with TDF_SLIM.
24782 (gather_memory_references_ref): Likewise.
24784 2017-05-18 Martin Liska <mliska@suse.cz>
24786 * vec.h (struct vnull): Use it.
24788 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24790 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24791 (true_predicate, false_predicate, true_predicate_p,
24792 false_predicate_p): Remove.
24793 (single_cond_predicate, not_inlined_predicate): Turn to member function
24795 (add_condition): Update.
24796 (add_clause): Turn to...
24797 (predicate::add_clause): ... this one; update; allow passing NULL
24799 (and_predicates): Turn to ...
24800 (predicate::operator &=): ... this one.
24801 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24802 (or_predicates): Turn to ...
24803 (predicate::or_with): ... this one.
24804 (evaluate_predicate): Turn to ...
24805 (predicate::evaluate): ... this one.
24806 (predicate_probability): Turn to ...
24807 (predicate::probability): ... this one.
24808 (dump_condition): Update.
24809 (dump_predicate): Turn to ...
24810 (predicate::dump): ... this one.
24811 (account_size_time): Update.
24812 (edge_set_predicate): Update.
24813 (set_hint_predicate): UPdate.
24814 (evaluate_conditions_for_known_args): Update.
24815 (evaluate_properties_for_edge): Update.
24816 (remap_predicate_after_duplication): Turn to...
24817 (predicate::remap_after_duplication): ... this one.
24818 (remap_hint_predicate_after_duplication): Update.
24819 (inline_summary_t::duplicate): UPdate.
24820 (dump_inline_edge_summary): Update.
24821 (dump_inline_summary): Update.
24822 (set_cond_stmt_execution_predicate): Update.
24823 (set_switch_stmt_execution_predicate): Update.
24824 (compute_bb_predicates): Update.
24825 (will_be_nonconstant_expr_predicate): Update.
24826 (will_be_nonconstant_predicate): Update.
24827 (phi_result_unknown_predicate): Update.
24828 (predicate_for_phi_result): Update.
24829 (array_index_predicate): Update.
24830 (estimate_function_body_sizes): Update.
24831 (estimate_node_size_and_time): Update.
24832 (estimate_ipcp_clone_size_and_time): Update.
24833 (remap_predicate): Rename to ...
24834 (predicate::remap_after_inlining): ... this one.
24835 (remap_hint_predicate): Update.
24836 (inline_merge_summary): Update.
24837 (inline_update_overall_summary): Update.
24838 (estimate_size_after_inlining): Update.
24839 (read_predicate): Rename to ...
24840 (predicate::stream_in): ... this one.
24841 (read_inline_edge_summary): Update.
24842 (write_predicate): Rename to ...
24843 (predicate::stream_out): ... this one.
24844 (write_inline_edge_summary): Update.
24845 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24846 (clause_t): Turn to uint32_t
24847 (predicate): Turn to class; implement constructor and operators
24849 (size_time_entry): Update.
24850 (inline_summary): Update.
24851 (inline_edge_summary): Update.
24853 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24855 * fold-const.c (fold_binary_loc): Move transformation...
24856 * match.pd (C - X CMP X): ... here.
24858 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24860 * config/sparc/sparc.c (sparc_option_override): Set function
24861 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24862 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24864 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24866 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24868 2017-05-18 Marek Polacek <polacek@redhat.com>
24871 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24872 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24873 gimple_assign_load_p.
24875 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24877 PR middle-end/80692
24878 * real.c (do_compare): Give decimal_do_compare preference over
24879 comparing just the signs.
24881 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24883 * doc/md.texi (Canonicalization of Instructions): Describe the
24884 canonical form of instructions that inherently set a condition
24887 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24889 PR middle-end/80775
24890 * tree-cfg.c: Move deletion of unreachable case statements to after
24891 the merging of consecutive case labels.
24893 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24895 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24896 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24897 restoring of callee-saved registers.
24899 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24901 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24902 * config/visium/visium.c (single_set_and_flags): Likewise.
24903 * config/visium/visium.md (Substitutions): Likewise.
24905 2017-05-17 Martin Liska <mliska@suse.cz>
24907 * cfg.c: Introduce dump_flags_t type and
24908 use it instead of int type.
24910 * cfghooks.c: Likewise.
24911 * cfghooks.h (struct cfg_hooks): Likewise.
24912 * cfgrtl.c: Likewise.
24913 * cfgrtl.h: Likewise.
24914 * cgraph.c (cgraph_node::get_body): Likewise.
24915 * coretypes.h: Likewise.
24916 * domwalk.c: Likewise.
24917 * domwalk.h: Likewise.
24918 * dumpfile.c (struct dump_option_value_info): Likewise.
24919 (dump_enable_all): Likewise.
24920 (dump_switch_p_1): Likewise.
24921 (opt_info_switch_p): Likewise.
24922 * dumpfile.h (enum tree_dump_index): Likewise.
24923 (struct dump_file_info): Likewise.
24924 * genemit.c: Likewise.
24925 * generic-match-head.c: Likewise.
24926 * gengtype.c (open_base_files): Likewise.
24927 * gimple-pretty-print.c: Likewise.
24928 * gimple-pretty-print.h: Likewise.
24929 * graph.c (print_graph_cfg): Likewise.
24930 * graphite-scop-detection.c (dot_all_sese): Likewise.
24931 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24932 * loop-unroll.c (report_unroll): Likewise.
24933 * passes.c (pass_manager::register_one_dump_file): Likewise.
24934 * print-tree.c: Likewise.
24935 * statistics.c: Likewise.
24936 * tree-cfg.c: Likewise.
24937 * tree-cfg.h: Likewise.
24938 * tree-dfa.c: Likewise.
24939 * tree-dfa.h: Likewise.
24940 * tree-dump.c (dump_function): Likewise.
24941 * tree-dump.h (struct dump_info): Likewise.
24942 * tree-pretty-print.c: Likewise.
24943 * tree-pretty-print.h: Likewise.
24944 * tree-ssa-live.c: Likewise.
24945 * tree-ssa-live.h: Likewise.
24946 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24947 * tree-vect-loop.c: Likewise.
24948 * tree-vect-slp.c: Likewise.
24950 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24951 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24953 PR tree-optimization/80457
24954 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24955 of all arguments to a statement as scalar_to_vec operations.
24956 (vectorizable_call): Adjust call to vect_model_simple_cost for
24958 (vectorizable_conversion): Likewise.
24959 (vectorizable_assignment): Likewise.
24960 (vectorizable_shift): Likewise.
24961 (vectorizable_operation): Likewise.
24962 (vectorizable_comparison): Likewise.
24963 (vect_is_simple_cond): Record the def types for operands.
24964 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24965 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24966 for statement argument count.
24968 2017-05-16 Carl Love <cel@us.ibm.com>
24970 * config/rs6000/rs6000-c: Add support for built-in functions
24971 vector unsigned long long vec_bperm (vector unsigned long long,
24972 vector unsigned char)
24973 vector signed long long vec_mule (vector signed int,
24975 vector unsigned long long vec_mule (vector unsigned int,
24976 vector unsigned int)
24977 vector signed long long vec_mulo (vector signed int,
24979 vector unsigned long long vec_mulo (vector unsigned int,
24980 vector unsigned int)
24981 vector signed char vec_sldw (vector signed char,
24982 vector signed char,
24984 vector unsigned char vec_sldw (vector unsigned char,
24985 vector unsigned char,
24987 vector signed short vec_sldw (vector signed short,
24988 vector signed short,
24990 vector unsigned short vec_sldw (vector unsigned short,
24991 vector unsigned short,
24993 vector signed int vec_sldw (vector signed int,
24996 vector unsigned int vec_sldw (vector unsigned int,
24997 vector unsigned int,
24999 vector signed long long vec_sldw (vector signed long long,
25000 vector signed long long,
25002 vector unsigned long long vec_sldw (vector unsigned long long,
25003 vector unsigned long long,
25005 * config/rs6000/rs6000-c: Add support for built-in functions
25006 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25007 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25008 * doc/extend.texi: Update the built-in documentation for the
25009 new built-in functions.
25011 2017-05-16 Marek Polacek <polacek@redhat.com>
25015 * tree.c (save_expr): Don't fold the expression.
25017 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25019 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25020 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25021 and (?*y,m). Update insn attributes.
25023 2017-05-16 Martin Liska <mliska@suse.cz>
25025 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25026 flags argument of print_gimple_stmt, print_gimple_expr,
25027 print_generic_stmt and print_generic_expr.
25028 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25029 * coretypes.h: Likewise.
25030 * except.c (dump_eh_tree): Likewise.
25031 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25032 * gimple-pretty-print.h: Likewise.
25033 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25034 (backprop::push_to_worklist): Likewise.
25035 (backprop::pop_from_worklist): Likewise.
25036 (backprop::process_use): Likewise.
25037 (backprop::intersect_uses): Likewise.
25038 (note_replacement): Likewise.
25039 * gimple-ssa-store-merging.c
25040 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25041 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25042 (pass_store_merging::execute): Likewise.
25043 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25044 (ssa_base_cand_dump_callback): Likewise.
25045 (dump_incr_vec): Likewise.
25046 (replace_refs): Likewise.
25047 (replace_mult_candidate): Likewise.
25048 (create_add_on_incoming_edge): Likewise.
25049 (create_phi_basis): Likewise.
25050 (insert_initializers): Likewise.
25051 (all_phi_incrs_profitable): Likewise.
25052 (introduce_cast_before_cand): Likewise.
25053 (replace_one_candidate): Likewise.
25054 * gimplify.c (gimplify_expr): Likewise.
25055 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25056 (set_rename): Likewise.
25057 (rename_uses): Likewise.
25058 (copy_loop_phi_nodes): Likewise.
25059 (add_close_phis_to_merge_points): Likewise.
25060 (copy_loop_close_phi_args): Likewise.
25061 (copy_cond_phi_args): Likewise.
25062 (graphite_copy_stmts_from_block): Likewise.
25063 (translate_pending_phi_nodes): Likewise.
25064 * graphite-poly.c (print_pdr): Likewise.
25065 (dump_gbb_cases): Likewise.
25066 (dump_gbb_conditions): Likewise.
25067 (print_scop_params): Likewise.
25068 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25069 (build_cross_bb_scalars_use): Likewise.
25070 (gather_bbs::before_dom_children): Likewise.
25071 * hsa-dump.c (dump_hsa_immed): Likewise.
25072 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25073 (get_replacement_map): Likewise.
25074 * ipa-inline-analysis.c (dump_condition): Likewise.
25075 (estimate_function_body_sizes): Likewise.
25076 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25077 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25078 * ipa-prop.c (ipa_dump_param): Likewise.
25079 (ipa_print_node_jump_functions_for_edge): Likewise.
25080 (ipa_modify_call_arguments): Likewise.
25081 (ipa_modify_expr): Likewise.
25082 (ipa_dump_param_adjustments): Likewise.
25083 (ipa_dump_agg_replacement_values): Likewise.
25084 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25085 * ipa-pure-const.c (check_stmt): Likewise.
25086 (pass_nothrow::execute): Likewise.
25087 * ipa-split.c (execute_split_functions): Likewise.
25088 * omp-offload.c (dump_oacc_loop_part): Likewise.
25089 (dump_oacc_loop): Likewise.
25090 * trans-mem.c (tm_log_emit): Likewise.
25091 (tm_memopt_accumulate_memops): Likewise.
25092 (dump_tm_memopt_set): Likewise.
25093 (dump_tm_memopt_transform): Likewise.
25094 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25095 (print_loop): Likewise.
25096 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25097 (chkp_gather_checks_info): Likewise.
25098 (chkp_get_check_result): Likewise.
25099 (chkp_remove_check_if_pass): Likewise.
25100 (chkp_use_outer_bounds_if_possible): Likewise.
25101 (chkp_reduce_bounds_lifetime): Likewise.
25102 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25103 (chkp_mark_completed_bounds): Likewise.
25104 (chkp_register_incomplete_bounds): Likewise.
25105 (chkp_mark_invalid_bounds): Likewise.
25106 (chkp_maybe_copy_and_register_bounds): Likewise.
25107 (chkp_build_returned_bound): Likewise.
25108 (chkp_get_bound_for_parm): Likewise.
25109 (chkp_build_bndldx): Likewise.
25110 (chkp_get_bounds_by_definition): Likewise.
25111 (chkp_generate_extern_var_bounds): Likewise.
25112 (chkp_get_bounds_for_decl_addr): Likewise.
25113 * tree-chrec.c (chrec_apply): Likewise.
25114 * tree-data-ref.c (dump_data_reference): Likewise.
25115 (dump_subscript): Likewise.
25116 (dump_data_dependence_relation): Likewise.
25117 (analyze_overlapping_iterations): Likewise.
25118 * tree-inline.c (expand_call_inline): Likewise.
25119 (tree_function_versioning): Likewise.
25120 * tree-into-ssa.c (dump_defs_stack): Likewise.
25121 (dump_currdefs): Likewise.
25122 (dump_names_replaced_by): Likewise.
25123 (dump_update_ssa): Likewise.
25124 (update_ssa): Likewise.
25125 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25126 * tree-parloops.c (build_new_reduction): Likewise.
25127 (try_create_reduction_list): Likewise.
25128 (ref_conflicts_with_region): Likewise.
25129 (oacc_entry_exit_ok_1): Likewise.
25130 (oacc_entry_exit_single_gang): Likewise.
25131 * tree-pretty-print.h: Likewise.
25132 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25133 (get_scalar_evolution): Likewise.
25134 (add_to_evolution): Likewise.
25135 (get_loop_exit_condition): Likewise.
25136 (analyze_evolution_in_loop): Likewise.
25137 (analyze_initial_condition): Likewise.
25138 (analyze_scalar_evolution): Likewise.
25139 (instantiate_scev): Likewise.
25140 (number_of_latch_executions): Likewise.
25141 (gather_chrec_stats): Likewise.
25142 (final_value_replacement_loop): Likewise.
25143 (scev_const_prop): Likewise.
25144 * tree-sra.c (dump_access): Likewise.
25145 (disqualify_candidate): Likewise.
25146 (create_access): Likewise.
25147 (reject): Likewise.
25148 (maybe_add_sra_candidate): Likewise.
25149 (create_access_replacement): Likewise.
25150 (analyze_access_subtree): Likewise.
25151 (analyze_all_variable_accesses): Likewise.
25152 (sra_modify_assign): Likewise.
25153 (initialize_constant_pool_replacements): Likewise.
25154 (find_param_candidates): Likewise.
25155 (decide_one_param_reduction): Likewise.
25156 (replace_removed_params_ssa_names): Likewise.
25157 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25158 * tree-ssa-copy.c (dump_copy_of): Likewise.
25159 (copy_prop_visit_cond_stmt): Likewise.
25160 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25161 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25162 (record_equivalences_from_stmt): Likewise.
25163 * tree-ssa-dse.c (compute_trims): Likewise.
25164 (delete_dead_call): Likewise.
25165 (delete_dead_assignment): Likewise.
25166 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25167 (forward_propagate_into_cond): Likewise.
25168 (pass_forwprop::execute): Likewise.
25169 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25170 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25172 (move_computations_worker): Likewise.
25173 (execute_sm): Likewise.
25174 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25175 (remove_exits_and_undefined_stmts): Likewise.
25176 (remove_redundant_iv_tests): Likewise.
25177 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25178 (adjust_iv_update_pos): Likewise.
25179 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25180 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25181 (value_replacement): Likewise.
25182 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25183 * tree-ssa-pre.c (print_pre_expr): Likewise.
25184 (get_representative_for): Likewise.
25185 (create_expression_by_pieces): Likewise.
25186 (insert_into_preds_of_block): Likewise.
25187 (eliminate_insert): Likewise.
25188 (eliminate_dom_walker::before_dom_children): Likewise.
25189 (eliminate): Likewise.
25190 (remove_dead_inserted_code): Likewise.
25191 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25192 * tree-ssa-reassoc.c (get_rank): Likewise.
25193 (eliminate_duplicate_pair): Likewise.
25194 (eliminate_plus_minus_pair): Likewise.
25195 (eliminate_not_pairs): Likewise.
25196 (undistribute_ops_list): Likewise.
25197 (eliminate_redundant_comparison): Likewise.
25198 (update_range_test): Likewise.
25199 (optimize_range_tests_var_bound): Likewise.
25200 (optimize_vec_cond_expr): Likewise.
25201 (rewrite_expr_tree): Likewise.
25202 (rewrite_expr_tree_parallel): Likewise.
25203 (linearize_expr): Likewise.
25204 (break_up_subtract): Likewise.
25205 (linearize_expr_tree): Likewise.
25206 (attempt_builtin_powi): Likewise.
25207 (attempt_builtin_copysign): Likewise.
25208 (transform_stmt_to_copy): Likewise.
25209 (transform_stmt_to_multiply): Likewise.
25210 (dump_ops_vector): Likewise.
25211 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25212 (print_scc): Likewise.
25213 (set_ssa_val_to): Likewise.
25214 (visit_reference_op_store): Likewise.
25215 (visit_use): Likewise.
25216 (sccvn_dom_walker::before_dom_children): Likewise.
25217 (run_scc_vn): Likewise.
25218 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25220 (expr_hash_elt::print): Likewise.
25221 (const_and_copies::pop_to_marker): Likewise.
25222 (const_and_copies::record_const_or_copy_raw): Likewise.
25223 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25224 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25225 (dump_predicates): Likewise.
25226 (find_uninit_use): Likewise.
25227 (warn_uninitialized_phi): Likewise.
25228 (pass_late_warn_uninitialized::execute): Likewise.
25229 * tree-ssa.c (verify_vssa): Likewise.
25230 (verify_ssa): Likewise.
25231 (maybe_optimize_var): Likewise.
25232 * tree-vrp.c (dump_value_range): Likewise.
25233 (dump_all_value_ranges): Likewise.
25234 (dump_asserts_for): Likewise.
25235 (register_edge_assert_for_2): Likewise.
25236 (vrp_visit_cond_stmt): Likewise.
25237 (vrp_visit_switch_stmt): Likewise.
25238 (vrp_visit_stmt): Likewise.
25239 (vrp_visit_phi_node): Likewise.
25240 (simplify_cond_using_ranges_1): Likewise.
25241 (fold_predicate_in): Likewise.
25242 (evrp_dom_walker::before_dom_children): Likewise.
25243 (evrp_dom_walker::push_value_range): Likewise.
25244 (evrp_dom_walker::pop_value_range): Likewise.
25245 (execute_early_vrp): Likewise.
25247 2017-05-16 Richard Biener <rguenther@suse.de>
25249 * dwarf2out.c (loc_list_from_tree_1): Do not create
25250 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25252 2017-05-16 Richard Biener <rguenther@suse.de>
25254 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25256 (note_variable_value_in_expr): If we resolved the decl ref
25257 do not push to the stack.
25259 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25261 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25262 operations in fast-math mode.
25263 (vaddq_f16): Likewise.
25264 (vmul_f16): Likewise.
25265 (vmulq_f16): Likewise.
25266 (vsub_f16): Likewise.
25267 (vsubq_f16): Likewise.
25268 * config/arm/neon.md (add<mode>3): New.
25270 (fma:<VH:mode>3): New. Also remove outdated comment.
25273 2017-05-16 Martin Liska <mliska@suse.cz>
25277 * ipa-devirt.c (warn_types_mismatch): Fix typo.
25278 (odr_types_equivalent_p): Likewise.
25280 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
25282 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
25284 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
25287 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
25288 non-interunit SSE move alternatives with '?'.
25289 (zero-extendsidi peephole2): New peephole to skip intermediate
25290 general register in SSE zero-extend sequence.
25292 2017-05-15 Jeff Law <law@redhat.com>
25294 * reorg.c (relax_delay_slots): Create a new variable to hold
25295 the temporary target rather than clobbering TARGET_LABEL.
25297 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
25298 missing argument to extract_bit_field call.
25299 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
25301 2017-05-15 Martin Liska <mliska@suse.cz>
25304 * gcc.c (process_command): Do not allow empty argument of -o option.
25306 2017-05-15 Renlin Li <renlin.li@arm.com>
25308 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25309 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25310 * config/aarch64/constraints.md (Usf): Add long call check.
25311 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25312 (call_value): Likewise.
25313 (sibcall): Likewise.
25314 (sibcall_value): Likewise.
25316 (call_value_insn): New.
25317 (sibcall_insn): Update rtx pattern.
25318 (sibcall_value_insn): Likewise.
25319 (call_internal): Remove.
25320 (call_value_internal): Likewise.
25321 (sibcall_internal): Likewise.
25322 (sibcall_value_internal): Likewise.
25323 (call_reg): Likewise.
25324 (call_symbol): Likewise.
25325 (call_value_reg): Likewise.
25326 (call_value_symbol): Likewise.
25328 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25331 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25333 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25335 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25336 compatible with CCGOCmode and with CCZmode.
25338 2017-05-14 Martin Sebor <msebor@redhat.com>
25340 PR middle-end/77671
25341 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25342 (gimple_fold_builtin_snprintf): Same.
25343 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25344 (gimple_fold_builtin_snprintf): Same.
25345 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25346 of character types.
25347 (is_call_safe): New function.
25348 (try_substitute_return_value): Call it.
25349 (try_simplify_call): New function.
25350 (pass_sprintf_length::handle_gimple_call): Call it.
25352 2017-05-14 Martin Sebor <msebor@redhat.com>
25354 PR middle-end/80669
25355 * builtins.c (expand_builtin_stpncpy): Simplify.
25357 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25359 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25360 * config/i386/i386.h
25361 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25362 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25363 (struct machine_function): Add new members call_ms2sysv,
25364 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25365 (struct machine_frame_state): New fields sp_realigned and
25366 sp_realigned_offset.
25367 * config/i386/i386.c
25368 (enum xlogue_stub): New enum.
25369 (enum xlogue_stub_sets): New enum.
25370 (class xlogue_layout): New class.
25371 (struct ix86_frame): New fields stack_realign_allocate_offset,
25372 stack_realign_offset and outlined_save_offset. Modify comments to
25373 detail stack layout when using out-of-line stubs.
25374 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25375 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25376 -mcall-ms2sysv-xlogues.
25377 (stub_managed_regs): New static variable.
25378 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25379 registers managed by out-of-line stub.
25380 (disable_call_ms2sysv_xlogues): New function.
25381 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25382 m->call_ms2sysv when appropriate and compute frame layout for
25384 (sp_valid_at, fp_valid_at): New inline functions.
25385 (choose_basereg): New function.
25386 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25388 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25389 Use align parameter of choose_baseaddr to generated aligned SSE movs
25391 (pro_epilogue_adjust_stack): Modify to track
25392 machine_frame_state::sp_realigned.
25393 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25394 (ix86_nsaved_sseregs): Likewise.
25395 (ix86_emit_save_regs): Likewise.
25396 (ix86_emit_save_regs_using_mov): Likewise.
25397 (ix86_emit_save_sse_regs_using_mov): Likewise.
25398 (get_scratch_register_on_entry): Likewise.
25399 (gen_frame_set): New function.
25400 (gen_frame_load): Likewise.
25401 (gen_frame_store): Likewise.
25402 (emit_outlined_ms2sysv_save): Likewise.
25403 (emit_outlined_ms2sysv_restore): Likewise.
25404 (ix86_expand_prologue): Modify stack re-alignment code and call
25405 emit_outlined_ms2sysv_save when appropriate.
25406 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25407 parameter rtx_insn *insn, which allows the function to be used to only
25408 generate the notes.
25409 (ix86_expand_epilogue): Modify validity checks of frame and stack
25410 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25411 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25412 * config/i386/predicates.md
25413 (save_multiple): New predicate.
25414 (restore_multiple): Likewise.
25415 * config/i386/sse.md
25416 (save_multiple<mode>): New pattern.
25417 (save_multiple_realign<mode>): Likewise.
25418 (restore_multiple<mode>): Likewise.
25419 (restore_multiple_and_return<mode>): Likewise.
25420 (restore_multiple_leave_return<mode>): Likewise.
25421 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25423 2017-05-14 Julia Koval <julia.koval@intel.com>
25425 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25426 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25427 (__builtin_ia32_xsetbv): New builtins.
25428 * config/i386/i386.c (ix86_expand_special_args_builtin):
25430 (ix86_expand_builtin): Special expand for new intrinsics.
25431 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25432 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25433 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25435 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25437 * cfganal.c (inverted_post_order_compute): Change argument type
25439 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25440 * df-core.c (rest_of_handle_df_initialize): Adjust.
25441 (rest_of_handle_df_finish): Likewise.
25442 (df_analyze_1): Likewise.
25443 (df_analyze): Likewise.
25444 (loop_inverted_post_order_compute): Change argument to be a vec *.
25445 (df_analyze_loop): Adjust.
25446 (df_get_n_blocks): Likewise.
25447 (df_get_postorder): Likewise.
25448 * df.h (struct df_d): Change field to be a vec.
25449 * lcm.c (compute_laterin): Adjust.
25450 (compute_available): Likewise.
25451 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25452 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25453 * tree-ssa-pre.c (compute_antic): Likewise.
25455 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25457 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25458 (depth_first_search::depth_first_search): Change structure init
25459 function to this constructor.
25460 (depth_first_search::add_bb): Rename function to this member.
25461 (depth_first_search::execute): Likewise.
25462 (flow_dfs_compute_reverse_finish): Adjust.
25464 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25466 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25467 (longest_simple_path): Likewise.
25468 * shrink-wrap.c (spread_components): Likewise.
25469 (disqualify_problematic_components): Likewise.
25470 (emit_common_heads_for_components): Likewise.
25471 (emit_common_tails_for_components): Likewise.
25472 (insert_prologue_epilogue_for_components): Likewise.
25474 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25476 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25479 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25481 * df-core.c (df_set_blocks): Start using auto_bitmap.
25482 (df_compact_blocks): Likewise.
25483 * df-problems.c (df_rd_confluence_n): Likewise.
25484 * df-scan.c (df_insn_rescan_all): Likewise.
25485 (df_process_deferred_rescans): Likewise.
25486 (df_update_entry_block_defs): Likewise.
25487 (df_update_exit_block_uses): Likewise.
25488 (df_entry_block_bitmap_verify): Likewise.
25489 (df_exit_block_bitmap_verify): Likewise.
25490 (df_scan_verify): Likewise.
25491 * lra-constraints.c (lra_constraints): Likewise.
25492 (undo_optional_reloads): Likewise.
25493 (lra_undo_inheritance): Likewise.
25494 * lra-remat.c (calculate_gen_cands): Likewise.
25495 (do_remat): Likewise.
25496 * lra-spills.c (assign_spill_hard_regs): Likewise.
25497 (spill_pseudos): Likewise.
25498 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25499 (bitmap_set_subtract_values): Likewise.
25501 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25503 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25504 management with auto_bitmap.
25505 (fix_inter_tick): Likewise.
25506 (fix_recovery_deps): Likewise.
25507 * ira.c (add_store_equivs): Likewise.
25508 (find_moveable_pseudos): Likewise.
25509 (split_live_ranges_for_shrink_wrap): Likewise.
25510 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25511 (rtx_reuse_manager::seen_def_p): Likewise.
25512 (rtx_reuse_manager::set_seen_def): Likewise.
25513 * print-rtl.h (class rtx_reuse_manager): Likewise.
25515 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25517 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25519 (migrate_btr_def): Likewise.
25520 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25521 * df-core.c (loop_post_order_compute): Likewise.
25522 (loop_inverted_post_order_compute): Likewise.
25523 * hsa-common.h: Likewise.
25524 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25525 * init-regs.c (initialize_uninitialized_regs): Likewise.
25526 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25527 (inline_small_functions): Likewise.
25528 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25529 * ira.c (combine_and_move_insns): Likewise.
25530 (build_insn_chain): Likewise.
25531 * loop-invariant.c (find_invariants): Likewise.
25532 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25533 * predict.c (tree_predict_by_opcode): Likewise.
25534 (predict_paths_leading_to): Likewise.
25535 (predict_paths_leading_to_edge): Likewise.
25536 (estimate_loops_at_level): Likewise.
25537 (estimate_loops): Likewise.
25538 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25539 (spread_components): Likewise.
25540 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25541 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25542 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25543 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25544 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25545 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25546 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25547 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25548 (mark_threaded_blocks): Likewise.
25549 (thread_through_all_blocks): Likewise.
25550 * tree-ssa.c (verify_ssa): Likewise.
25551 (execute_update_addresses_taken): Likewise.
25552 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25554 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25556 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25558 (post_order_compute): Likewise.
25559 (inverted_post_order_compute): Likewise.
25560 (pre_and_rev_post_order_compute_fn): Likewise.
25562 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25564 * genrecog.c (int_set::int_set): Explicitly construct our
25565 auto_vec base class.
25566 * vec.h (auto_vec::auto_vec): New constructor.
25568 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25570 * bitmap.h (class auto_bitmap): New constructor taking
25571 bitmap_obstack * argument.
25573 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25575 * bitmap.h (class auto_bitmap): Change type of m_bits to
25576 bitmap_head, and adjust ctor / dtor and member operators.
25578 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25580 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25581 when returned register mode doesn't match original mode.
25583 2017-05-12 Jeff Law <law@redhat.com>
25584 Jakub Jelinek <jakub@redhat.com>
25586 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25587 we look for cc setter after the compare-elim changes.
25588 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25589 within the vector to match what compare-elim now expects.
25590 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25591 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25593 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25594 after the compare-elim changes.
25595 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25596 the vector to match what compare-elim now expects.
25597 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25598 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25599 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25600 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25601 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25603 * config/visium/visium.c (single_set_and_flags): Fix where
25604 we look for cc setter after the compare-elim changes.
25605 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25606 with the vector to match what compare-elim now expects.
25607 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25608 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25609 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25610 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25611 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25612 (neg<mode>2_insn_set_overflow): Likewise.
25614 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25616 PR middle-end/79794
25617 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25618 maybe_expand_insn call, set ops[0].target. If still set after call,
25619 set alt_rtl. Add extra arg to recursive calls.
25620 (extract_bit_field): Add alt_rtl argument. Pass to
25622 * expmed.h (extract_bit_field): Fix prototype.
25623 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25624 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25625 to extract_bit_field_calls.
25626 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25627 Pass alt_rtl to extract_bit_field calls.
25628 * calls.c (store_unaligned_arguments_into_psuedos)
25629 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25630 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25632 * optabs.h (struct expand_operand): Add target bitfield.
25634 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25636 * compare-elim.c (try_eliminate_compare): Canonicalize
25637 operation with embedded compare to
25638 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25639 (set (reg) (operation)].
25641 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25643 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25646 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25647 cost of adding a carry flag for ADC instruction.
25648 [case MINUS]: Ignore the cost of subtracting a carry flag
25649 for SBB instruction.
25651 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25653 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25655 * config/rs6000/bmiintrin.h: New file.
25656 * config/rs6000/bmi2intrin.h: New file.
25657 * config/rs6000/x86intrin.h: New file.
25659 2017-05-12 Jeff Law <law@redhat.com>
25661 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25664 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25666 PR middle-end/80707
25667 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25669 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25671 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25672 early expansion of vector divide builtins.
25673 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25674 builtins identified as having unsigned arguments.
25676 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25678 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25679 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25680 expansion of vector logical operations (and, andc, or, xor,
25683 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25685 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25686 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25688 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25690 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25691 early GIMPLE expansion of vector multiplies.
25693 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25695 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25696 TARGET_HAVE_MOVT conditional.
25697 (movt splitter): Likewise.
25699 2017-05-12 Richard Biener <rguenther@suse.de>
25701 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25702 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25703 Fold all stmts not inplace.
25705 2017-05-12 Richard Biener <rguenther@suse.de>
25707 PR tree-optimization/80713
25708 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25709 inserted_exprs bit for not removed stmts.
25711 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25713 PR middle-end/69921
25714 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25715 parallelized" attribute for parallelized OpenACC kernels.
25716 * omp-offload.c (execute_oacc_device_lower): Use it.
25718 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25719 Set "oacc kernels" attribute.
25720 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25721 parameter. Adjust all users.
25722 (oacc_fn_attrib_kernels_p): Remove function.
25723 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25724 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25725 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25726 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25727 assert "oacc kernels" attribute is set.
25729 2017-05-11 Carl Love <cel@us.ibm.com>
25731 * config/rs6000/rs6000-c: Add support for built-in functions
25732 vector unsigned char vec_popcnt (vector signed char)
25733 vector unsigned char vec_popcnt (vector unsigned char)
25734 vector unsigned short vec_popcnt (vector signed short)
25735 vector unsigned short vec_popcnt (vector unsigned short)
25736 vector unsigned int vec_popcnt (vector signed int)
25737 vector unsigned int vec_popcnt (vector unsigned int)
25738 vector unsigned long long vec_popcnt (vector signed long long)
25739 vector unsigned long long vec_popcnt (vector unsigned long long)
25740 vector signed long long vec_slo (vector signed long long,
25741 vector signed char)
25742 vector signed long long vec_slo (vector signed long long,
25743 vector unsigned char)
25744 vector unsigned long long vec_slo (vector unsigned long long,
25745 vector signed char)
25746 vector unsigned long long vec_slo (vector unsigned long long,
25747 vector unsigned char)
25748 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25749 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25750 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25751 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25752 * doc/extend.texi: Update the built-in documentation file for the
25753 new built-in functions.
25755 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25757 * attribs.h (sorted_attr_string): Move machine independent
25758 functions for target clone support from the i386 port to common
25759 code. Rename ix86_function_versions to common_function_versions.
25760 Rename make_name to make_unique_name.
25761 (common_function_versions): Likewise.
25762 (make_unique_name): Likewise.
25763 (make_dispatcher_decl): Likewise.
25764 (is_function_default_version): Likewise.
25765 * attribs.c (attr_strcmp): Likewise.
25766 (sorted_attr_string): Likewise.
25767 (common_function_versions): Likewise.
25768 (make_unique_name): Likewise.
25769 (make_dispatcher_decl): Likewise.
25770 (is_function_default_version): Likewise.
25771 * config/i386/i386.c (attr_strcmp): Likewise.
25772 (sorted_attr_string): Likewise.
25773 (ix86_function_versions): Likewise.
25774 (make_name): Likewise.
25775 (make_dispatcher_decl): Likewise.
25776 (is_function_default_version): Likewise.
25777 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25779 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25782 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25783 Account for direct move costs for vec_construct of integer
25786 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25789 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25790 (UNSPEC_STX_ATOMIC): Ditto.
25791 (loaddi_via_sse): New insn.
25792 (storedi_via_sse): Ditto.
25793 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25794 Update corresponding peephole2 patterns.
25795 (atomic_storedi_fpu): Ditto.
25797 2017-05-11 Julia Koval <julia.koval@intel.com>
25799 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25800 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25802 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25803 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25804 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25806 2017-05-11 Nathan Sidwell <nathan@acm.org>
25808 * graphite-poly.c: Include dumpfile.h.
25810 * dumpfle.h (dump_function): Declare here ...
25811 * tree-dump.h (dump_function): ... not here.
25812 * dumpfile.c: #include tree-cfg.h.
25813 (dump_function): Move here from ...
25814 * tree-dump.c (dump_function): ... here.
25815 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25816 * graphite-poly.c: Don't include tree-dump.h.
25817 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25818 * print-tree.c: Likewise.
25819 * stor-layout.c: Likewise.
25820 * tree-nested.c: Likewise.
25822 * dumpfile.c (dump_start): Use TDF_FLAGS.
25823 (dump_enable_all): Fix TDF_KIND check thinko.
25825 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25827 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25828 array entries to represent two legal parameterizations of the
25829 overloaded __builtin_cmpb function, as represented by the
25830 P6_OV_BUILTIN_CMPB constant.
25831 (altivec_resolve_overloaded_builtin): Add special case handling
25832 for the __builtin_cmpb function, as represented by the
25833 P6_OV_BUILTIN_CMPB constant.
25834 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25835 (BU_P6_64BIT_2): New macro.
25836 (BU_P6_OVERLOAD_2): New macro
25837 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25838 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25839 (CMPB): Add overload support to represent both 32-bit and 64-bit
25840 compare-bytes function.
25841 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25842 support for TARGET_CMPB.
25843 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25844 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25845 documentation of the __builtin_cmpb overloaded built-in function.
25847 2017-05-11 Richard Biener <rguenther@suse.de>
25849 PR tree-optimization/80705
25850 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25851 bases are not vectorizable.
25853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25855 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25856 when counting register pressure.
25858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25860 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25861 (iv_ca_more_deps): Renamed to ...
25862 (iv_ca_compare_deps): ... this.
25863 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25865 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25867 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25869 (determine_group_iv_costs): ... here.
25870 (find_inv_vars_cb): Record inv var if it's not recorded before.
25872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25874 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25875 (get_shiftadd_cost): Ditto.
25877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25879 * tree-ssa-address.c: Include header file.
25880 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25882 (add_to_parts): Refactor.
25883 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25884 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25889 PR tree-optimization/53090
25890 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25892 (extract_cond_operands): Detect condition with IV on both sides
25893 and return COMP_IV_EXPR_2.
25894 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25895 (rewrite_use_compare): Simplify by removing call to function
25896 extract_cond_operands.
25898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25900 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25901 (extract_cond_operands): Detect condition comparing against non-
25902 invariant bound and return appropriate enum value.
25903 (find_interesting_uses_cond): Update use of extract_cond_operands.
25904 Handle its return value accordingly.
25905 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25907 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25909 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25910 nonlinear iv_use computation in loop invariant sensitive way.
25912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25914 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25915 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25917 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25919 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25920 (dump_cand): Support iv_cand.inv_exprs.
25921 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25923 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25926 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25928 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25930 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25931 as local function. Include necessary header files.
25932 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25934 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25936 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25938 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25940 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25941 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25942 RSHIFT_EXPR and BIT_NOT_EXPR.
25944 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25946 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25947 (adjust_setup_cost): New parameter supporting round up adjustment.
25948 (struct address_cost_data): Delete.
25949 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25950 (split_address_cost, ptr_difference_cost): Delete.
25951 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25952 (struct ainc_cost_data): New struct.
25953 (get_address_cost_ainc): New function.
25954 (get_address_cost, get_computation_cost): Reimplement.
25955 (determine_group_iv_cost_address): Record inv_expr for all uses of
25957 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25958 (iv_ca_has_deps): Reimplemented to ...
25959 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25961 (iv_ca_extend): Call iv_ca_more_deps.
25963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25965 * tree-ssa-address.c (struct mem_address): Move to header file.
25966 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25967 * tree-ssa-address.h (struct mem_address): Move from C file.
25968 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25970 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25972 * tree-affine.h (aff_combination_type): New interface.
25973 (aff_combination_zero_p): Remove static.
25974 (aff_combination_const_p): New interface.
25975 (aff_combination_singleton_var_p): New interfaces.
25977 2017-05-11 Richard Biener <rguenther@suse.de>
25979 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25980 Skip unreachable blocks and destinations.
25981 (eliminate): Move stmt removal and fixup ...
25982 (fini_eliminate): ... here. Skip inserted exprs.
25983 (pass_pre::execute): Move fini_pre after fini_eliminate.
25984 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25985 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25986 PRE to get rid of dead code that has invalid SSA form and
25987 split critical edges again.
25989 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25991 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25993 2017-05-11 Richard Biener <rguenther@suse.de>
25995 * passes.c (execute_function_todo): Verify loops if they are
25996 said to be up-to-date.
25997 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25998 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26000 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26003 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26004 handle calling assemble_external ourself.
26007 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26008 modes with zero size. Enhance comment.
26010 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26012 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26013 built-ins for vec_xl and vec_xst with short and char pointer
26016 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26018 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26019 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26020 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26021 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26022 (_mm_maskz_min_round_ss): New intrinsics.
26023 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26024 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26025 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26026 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26027 (__builtin_ia32_minss_mask_round): New builtins.
26028 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26029 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26030 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26032 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26033 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26035 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26038 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26040 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26041 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26042 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26043 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26044 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26045 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26046 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26047 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26048 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26049 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26050 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26051 (__builtin_ia32_mulss_mask_round): New builtins.
26052 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26053 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26054 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26056 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26057 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26059 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26062 2017-05-10 Julia Koval <julia.koval@intel.com>
26064 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26065 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26066 (_mm256_setr_m128i): New intrinsics.
26068 2017-05-10 Julia Koval <julia.koval@intel.com>
26070 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26071 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26072 (_mm_maskz_rcp14_ss): New intrinsics.
26073 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26074 (__builtin_ia32_rcp14ss_mask): New builtins.
26075 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26077 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26079 PR tree-optimization/51513
26080 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26081 (assert_unreachable_fallthru_edge_p): Use it.
26082 (group_case_labels_stmt): Likewise.
26083 * tree-cfg.h: Prototype it.
26084 * stmt.c: Include cfghooks.h and tree-cfg.h.
26085 (emit_case_dispatch_table) <gap_label>: New local variable.
26086 Use it to fill dispatch table gaps.
26087 Test for default_label before updating probabilities.
26088 (expand_case) <default_label>: Remove unneeded initialization.
26089 Test for unreachable default case statement and remove its edge.
26090 Set default_label accordingly.
26091 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26093 2017-05-10 Carl Love <cel@us.ibm.com>
26095 * config/rs6000/rs6000-c: Add support for built-in functions
26096 vector signed char vec_neg (vector signed char)
26097 vector signed short int vec_neg (vector short int)
26098 vector signed int vec_neg (vector signed int)
26099 vector signed long long vec_neg (vector signed long long)
26100 vector float vec_neg (vector float)
26101 vector double vec_neg (vector double)
26102 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26104 * config/rs6000/altivec.h: Add define for vec_neg
26105 * doc/extend.texi: Update the built-in documentation for the
26106 new built-in functions.
26108 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26110 PR tree-optimization/77644
26111 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26113 2017-05-10 Nathan Sidwell <nathan@acm.org>
26115 * dumpfile.h (TDI_lang_all): New.
26116 (TDF_KIND): New. Renumber others
26117 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26119 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26121 (get_dump_file_name): Adjust suffix generation.
26122 (dump_enable_all): Use TDF_KIND.
26123 * doc/invoke.texi (-fdump-lang-all): Document.
26125 * dumpfile.h: Tabify.
26127 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26130 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26131 Move member access before delete.
26133 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26135 * tree-inline.c (expand_call_inline): Split block at stmt
26138 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26141 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26142 are now unused after splitting mov{sf,sd}_hardfloat.
26143 (f32_lr2): Likewise.
26144 (f32_lm): Likewise.
26145 (f32_lm2): Likewise.
26146 (f32_li): Likewise.
26147 (f32_li2): Likewise.
26148 (f32_lv): Likewise.
26149 (f32_sr): Likewise.
26150 (f32_sr2): Likewise.
26151 (f32_sm): Likewise.
26152 (f32_sm2): Likewise.
26153 (f32_si): Likewise.
26154 (f32_si2): Likewise.
26155 (f32_sv): Likewise.
26156 (f32_dm): Likewise.
26157 (f32_vsx): Likewise.
26158 (f32_av): Likewise.
26159 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26160 For movsf, order stores so the VSX stores occur before the GPR
26161 store which encourages the register allocator to use a traditional
26162 FPR instead of a GPR. For movsd, order the stores so that the GPR
26163 store comes before the VSX stores to allow the power6 to work.
26164 This is due to the power6 not having a 32-bit integer store
26165 instruction from a FPR.
26166 (movsf_hardfloat): Likewise.
26167 (movsd_hardfloat): Likewise.
26169 2017-05-09 Martin Sebor <msebor@redhat.com>
26171 PR translation/80280
26172 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26175 PR translation/80280
26176 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26177 data member added in r247778.
26178 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26180 2017-05-09 Nathan Sidwell <nathan@acm.org>
26182 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26184 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26187 2017-05-09 Marek Polacek <polacek@redhat.com>
26189 * doc/invoke.texi: Fix typo.
26191 2017-05-09 Richard Biener <rguenther@suse.de>
26193 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26194 (vrp_val_is_min): Likewise.
26195 (set_value_range_to_value): Likewise.
26196 (set_value_range_to_nonnegative): Likewise.
26197 (gimple_assign_nonzero_p): Likewise.
26198 (gimple_stmt_nonzero_p): Likewise.
26199 (vrp_int_const_binop): Likewise. Remove unreachable case.
26200 (adjust_range_with_scev): Adjust comments.
26201 (compare_range_with_value): Likewise.
26202 (extract_range_from_phi_node): Likewise.
26203 (test_for_singularity): Likewise.
26205 2017-05-09 Richard Biener <rguenther@suse.de>
26207 * tree-vrp.c (get_single_symbol): Add assert that we don't
26208 get overflowed constants as invariant part.
26209 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26210 checks. Use wi::cmp instead of recursing for integer constants.
26211 (compare_values): Just ignore whether we assumed undefined
26212 overflow instead of failing the compare.
26213 (extract_range_for_var_from_comparison_expr): Add comment before the
26214 TREE_NO_WARNING sets.
26215 (test_for_singularity): Likewise.
26216 (extract_range_from_comparison): Do not disable optimization
26217 when we assumed undefined overflow.
26218 (extract_range_basic): Remove init of unused var.
26220 2017-05-09 Richard Biener <rguenther@suse.de>
26222 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26223 (extract_range_from_multiplicative_op_1): Adjust.
26224 (extract_range_from_binary_expr_1): Use int_const_binop.
26226 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26229 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26230 rs6000_store_data_bypass_p in seven define_bypass directives and
26231 in several comments.
26232 * config/rs6000/rs6000-protos.h: Add prototype for
26233 rs6000_store_data_bypass_p function.
26234 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26235 function implements slightly different (rs6000-specific) semantics
26236 than store_data_bypass_p, returning false rather than aborting
26237 with assertion error when arguments do not satisfy the
26238 requirements of store data bypass.
26239 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26240 rs6000_store_data_bypass_p.
26242 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26244 * config/xtensa/xtensa-protos.h
26245 (xtensa_initial_elimination_offset): New declaration.
26246 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26247 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26248 macro definition, add case for FRAME_POINTER_REGNUM when
26249 FRAME_GROWS_DOWNWARD.
26250 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26251 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26252 xtensa_initial_elimination_offset.
26254 2017-05-08 Nathan Sidwell <nathan@acm.org>
26256 * doc/invoke.texi: Alphabetize -fdump options.
26258 2017-05-08 Martin Sebor <msebor@redhat.com>
26260 PR translation/80280
26261 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26263 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26265 * target.def (compute_frame_layout): New optional target hook.
26266 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26267 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26268 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
26270 * reload1.c (verify_initial_elim_offsets): Likewise.
26271 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
26272 (use_simple_return_p): Call arm_compute_frame_layout if needed.
26273 (arm_get_frame_offsets): Split up into this ...
26274 (arm_compute_frame_layout): ... and this function.
26276 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
26278 * config/aarch64/constraints.md (Usa): New constraint.
26279 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
26281 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
26283 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
26284 with_multilib_list after it has been checked.
26286 2017-05-08 Richard Biener <rguenther@suse.de>
26288 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
26289 (bitmap_set_subtract_values): Likewise.
26291 2017-05-08 Richard Biener <rguenther@suse.de>
26293 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
26294 (gimple_assign_nonzero): ... this and remove strict_overflow_p
26296 (gimple_stmt_nonzero_warnv_p): Rename to ...
26297 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
26299 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
26300 (extract_range_basic): Adjust, do not disable propagation on
26301 strict overflow sensitive simplification.
26302 (vrp_visit_cond_stmt): Likewise.
26304 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26306 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26307 body size unconditionally.
26309 2017-05-07 Jeff Law <law@redhat.com>
26312 2017-05-06 Jeff Law <law@redhat.com>
26313 PR tree-optimization/78496
26314 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26317 PR tree-optimization/78496
26318 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26319 (simplify_stmt_using_ranges): Call it.
26320 (vrp_dom_walker::before_dom_children): Extract equivalences
26321 from an ASSERT_EXPR with an equality comparison against a
26324 2017-05-06 Jeff Law <law@redhat.com>
26326 PR tree-optimization/78496
26327 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26330 PR tree-optimization/78496
26331 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26332 (simplify_stmt_using_ranges): Call it.
26333 (vrp_dom_walker::before_dom_children): Extract equivalences
26334 from an ASSERT_EXPR with an equality comparison against a
26337 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26339 * lra-constraints.c (lra_copy_reg_equiv): New function.
26340 (split_reg): Use it to copy equivalence information from the
26341 original register to the spill register.
26343 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26345 PR rtl-optimization/75964
26346 * simplify-rtx.c (simplify_const_relational_operation): Remove
26347 invalid handling of comparisons of integer ABS.
26349 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26351 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26352 initialize to zero.
26353 (init_regs): Remove declaration.
26354 (function_arg_advance_32): Initialize error_p as boolean variable.
26356 2017-05-05 Nathan Sidwell <nathan@acm.org>
26358 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26359 lines. Use for (;;).
26361 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26363 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26364 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26365 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26366 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26367 VF=2 that require versioning.
26369 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26371 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26374 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26376 * diagnostic.h (diagnostic_override_option_index): Convert from
26377 macro to inline function.
26379 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26381 * diagnostic.c (last_module_changed_p): New function.
26382 (set_last_module): New function.
26383 (diagnostic_report_current_module): Convert macro usage to
26384 the above functions.
26385 * diagnostic.h (diagnostic_context::last_module): Strengthen
26386 from const line_map * to const line_map_ordinary *.
26387 (diagnostic_last_module_changed): Delete macro.
26388 (diagnostic_set_last_module): Delete macro.
26390 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26392 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26393 with diagnostic_report_diagnostic.
26394 (diagnostic_n_impl_richloc): Likewise.
26395 * diagnostic.h (report_diagnostic): Delete macro.
26396 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26397 with diagnostic_report_diagnostic.
26398 * substring-locations.c (format_warning_va): Likewise.
26400 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26402 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26403 save/restor of format_spec. Move option-printing code to...
26404 (print_option_information): ...this new function, and
26405 reimplement by simply printing to the pretty_printer,
26406 rather than appending to the format string.
26408 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26410 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26411 handling logic into...
26412 (update_effective_level_from_pragmas): ...this new function.
26414 2017-05-04 Andrew Waterman <andrew@sifive.com>
26416 * config/riscv/riscv.opt (mstrict-align): New option.
26417 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26418 (SLOW_UNALIGNED_ACCESS): Define.
26419 (riscv_slow_unaligned_access): Declare.
26420 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26422 (riscv_slow_unaligned_access): New variable.
26423 (rocket_tune_info): Set slow_unaligned_access to true.
26424 (optimize_size_tune_info): Set slow_unaligned_access to false.
26425 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26426 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26427 (riscv_option_override): Set riscv_slow_unaligned_access.
26428 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26430 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26432 * config/riscv/riscv.md: Unify indentation.
26434 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26439 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26441 (extendsi<mode>2): Add support for doing sign extension via
26442 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26443 don't have ISA 3.0 instructions.
26444 (extendsi<mode>2 splitter): Likewise.
26445 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26446 generate the normal insns since SImode can now go in vector
26447 registers. Disallow the special UNSPECs needed for previous
26448 machines to hide SImode being used. Add new insns
26449 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26450 (fix_trunc<mode>si2_stfiwx): Likewise.
26451 (fix_trunc<mode>si2_internal): Likewise.
26452 (fixuns_trunc<mode>si2): Likewise.
26453 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26454 (fctiw<u>z_<mode>_smallint): Likewise.
26455 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26456 of floating point to 32-bit integer from doing a direct move to
26457 the GPR registers to do a store.
26458 (fctiwz_<mode>): Break long line.
26460 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26462 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26463 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26464 (addr_list, addr_offset_valid_p): New.
26465 (split_address_groups): Check offset validity with above function.
26466 (gt-tree-ssa-loop-ivopts.h): Include header file.
26468 2017-05-05 Nathan Sidwell <nathan@acm.org>
26470 * config.gcc (arm*-*-*): Add missing 'fi'.
26472 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26474 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26475 included in -fopt-info does not matter.
26476 * doc/optinfo.texi (-fopt-info): Fix description of default
26477 behavour. Explicitly say order of options included in -fopt-info
26480 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26482 * config.gcc: Allow combinations of aprofile and rmprofile values for
26483 --with-multilib-list.
26484 * config/arm/t-multilib: New file.
26485 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26486 variables. Remove setting of ISA and floating-point ABI in
26487 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26488 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26489 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26490 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26492 * config/arm/t-rmprofile: Likewise except for the matches changes.
26493 * doc/install.texi (--with-multilib-list): Document the combination of
26494 aprofile and rmprofile values and warn about pitfalls in doing that.
26496 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26498 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26499 (movdi_aarch64): Likewise.
26501 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26503 PR tree-optimization/80632
26504 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26506 (build_arrays): Initialize it for virtual phis.
26507 (fix_phi_nodes): Use it for virtual phis.
26509 PR tree-optimization/80558
26510 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26511 [x, y] op z into [x op, y op z] for op & or | if conditions
26514 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26515 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26518 * config/arm/arm.md (use_literal_pool): Remove.
26519 (64-bit immediate split): No longer takes cost into consideration
26520 if arm_disable_literal_pool is enabled.
26521 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26522 used when arm_disable_literal_pool is enabled.
26523 (arm_max_const_double_inline_cost): Remove use of
26524 arm_disable_literal_pool.
26525 (push_minipool_fix): Add assert.
26526 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26527 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26528 (no_literal_pool_sf_immediate): New.
26530 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26532 PR tree-optimization/80613
26533 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26534 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26536 2017-05-05 Richard Biener <rguenther@suse.de>
26538 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26540 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26542 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26543 of this flag from insn conditions due to removal from r247495.
26545 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26547 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26549 (arm_early_store_addr_dep_ptr): Likewise.
26550 * config/arm/aarch-common-protos.h
26551 (arm_early_load_addr_dep_ptr): Add prototype.
26552 (arm_early_store_addr_dep_ptr): Likewise.
26553 * config/arm/cortex-a53.md: Add new bypasses.
26555 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26557 * tree.c (next_type_uid): Change type to unsigned.
26558 (type_hash_canon): Decrement back next_type_uid if
26559 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26560 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26563 2017-05-04 Martin Sebor <msebor@redhat.com>
26565 * builtins.c: Fix a trivial typo in a comment.
26567 PR middle-end/79234
26568 * builtins.c (check_sizes): Adjust to handle reading past the end.
26569 Avoid printing excessive upper bound of ranges. Use %E to print
26570 tree nodes instead of converting them to %wu.
26571 (expand_builtin_memchr): New function.
26572 (compute_dest_size): Rename...
26573 (compute_objsize): ...to this.
26574 (expand_builtin_memcpy): Adjust.
26575 (expand_builtin_mempcpy): Adjust.
26576 (expand_builtin_strcat): Adjust.
26577 (expand_builtin_strcpy): Adjust.
26578 (check_strncat_sizes): Adjust.
26579 (expand_builtin_strncat): Adjust.
26580 (expand_builtin_strncpy): Adjust and simplify.
26581 (expand_builtin_memset): Adjust.
26582 (expand_builtin_bzero): Adjust.
26583 (expand_builtin_memcmp): Adjust.
26584 (expand_builtin): Handle memcmp.
26585 (maybe_emit_chk_warning): Check strncat just once.
26587 2017-05-04 Martin Sebor <msebor@redhat.com>
26589 PR preprocessor/79214
26590 PR middle-end/79222
26591 PR middle-end/79223
26592 * builtins.c (check_sizes): Add inlining context and issue
26593 warnings even when -Wno-system-headers is set.
26594 (check_strncat_sizes): Same.
26595 (expand_builtin_strncat): Same.
26596 (expand_builtin_memmove): New function.
26597 (expand_builtin_stpncpy): Same.
26598 (expand_builtin): Handle memmove and stpncpy.
26600 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26602 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26603 which is not used any more.
26605 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26607 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26609 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26611 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26612 (cortexa53_tunings): Likewise.
26613 (cortexa57_tunings): Likewise.
26614 (cortexa72_tunings): Likewise.
26615 (cortexa73_tunings): Likewise.
26617 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26619 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26620 Set loop alignment to 8.
26622 2017-05-04 Martin Sebor <msebor@redhat.com>
26624 PR translation/80280
26625 * builtins.c (expand_builtin_object_size): Add missing quoting to
26626 %D and like directives.
26627 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26628 (hsa_type_for_tree_type): Same.
26629 (verify_function_arguments): Same.
26630 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26631 * varasm.c (get_section): Same.
26634 2017-05-04 Martin Sebor <msebor@redhat.com>
26636 PR translation/80280
26637 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26639 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26641 * config/aarch64/aarch64.c (generic_addrcost_table):
26642 Change HI/TI mode setting.
26644 2017-05-04 Martin Jambor <mjambor@suse.cz>
26646 PR tree-optimization/80622
26647 * tree-sra.c (comes_initialized_p): New function.
26648 (build_accesses_from_assign): Only set write lazily when
26649 comes_initialized_p is false.
26650 (analyze_access_subtree): Use comes_initialized_p.
26651 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26652 instead of testing for PARM_DECL.
26654 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26656 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26657 constraint on operand 0 to allow more general addressing modes.
26658 Adjust output template.
26659 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26661 * config/aarch64/aarch64-protos.h
26662 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26663 * config/aarch64/constraints.md (Dp): New address constraint.
26664 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26667 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26669 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26670 update use of estimate_ipcp_clone_size_and_time.
26671 (estimate_local_effects): Update use of
26672 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26673 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26674 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26675 Return nonspecialized time.
26677 2017-05-04 Richard Biener <rguenther@suse.de>
26679 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26680 for the last VUSE which def dominates the PHI. Directly call
26682 (get_continuation_for_phi_1): Remove.
26684 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26686 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26687 to explain the use of truncating division. Cap the number of
26688 iterations to the maximum given by nb_iterations_upper_bound,
26691 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26693 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26694 * configure: Regenerate.
26695 * config.in: Regenerate.
26696 * config/i386/driver-mingw32.c: new file.
26697 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26698 * config.host: Link driver-mingw32.o on MinGW host.
26699 * doc/install.texi: Document new --enable-mingw-wildcard configure
26702 2017-05-04 Marek Polacek <polacek@redhat.com>
26704 PR tree-optimization/80612
26705 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26707 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26708 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26710 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26711 (movt splitter): Likewise.
26712 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26713 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26714 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26715 block for Thumb-1 with MOVT.
26716 (thumb2_legitimate_address_p): Move code block ...
26717 (can_avoid_literal_pool_for_label_p): ... into this new function.
26718 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26720 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26721 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26722 "M-profile targets with the MOVT instruction".
26724 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26726 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26727 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26728 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26730 2017-05-04 Martin Liska <mliska@suse.cz>
26732 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26733 variable cond_code.
26735 2017-05-04 Richard Biener <rguenther@suse.de>
26737 * tree.c (array_at_struct_end_p): Handle arrays at struct
26738 end with flexarrays more conservatively. Refactor and treat
26739 arrays of arrays or aggregates more strict. Fix
26740 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26741 * tree.c (array_at_struct_end_p): Adjust prototype.
26742 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26743 * gimple-fold.c (get_range_strlen): Likewise.
26744 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26746 2017-05-04 Richard Biener <rguenther@suse.de>
26748 PR tree-optimization/31130
26749 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26751 (supports_overflow_infinity): Likewise.
26752 (is_negative_overflow_infinity): Likewise.
26753 (is_positive_overflow_infinity): Likewise.
26754 (is_overflow_infinity): Likewise.
26755 (stmt_overflow_infinity): Likewise.
26756 (overflow_infinity_range_p): Likewise.
26757 (usable_range_p): Remove as always returning true.
26758 (make_overflow_infinity): Remove.
26759 (negative_overflow_infinity): Likewise.
26760 (positive_overflow_infinity): Likewise.
26761 (avoid_overflow_infinity): Likewise.
26762 (set_value_range): Adjust accordingly.
26763 (set_value_range_to_nonnegative): Likewise, remove now unused
26764 overflow_infinity arg.
26765 (vrp_operand_equal_p): Adjust.
26766 (update_value_range): Likewise.
26767 (range_int_cst_singleton_p): Likewise.
26768 (operand_less_p): Likewise.
26769 (compare_values_warnv): Likewise.
26770 (extract_range_for_var_from_comparison_expr): Likewise.
26771 (vrp_int_const_binop): Likewise.
26772 (zero_nonzero_bits_from_vr): Likewise.
26773 (extract_range_from_multiplicative_op_1): Likewise.
26774 (extract_range_from_binary_expr_1): Likewise.
26775 (extract_range_from_unary_expr): Likewise.
26776 (extract_range_from_comparison): Likewise.
26777 (extract_range_basic): Likewise.
26778 (adjust_range_with_scev): Likewise.
26779 (compare_ranges): Likewise.
26780 (compare_range_with_value): Likewise.
26781 (dump_value_range): Likewise.
26782 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26784 (simplify_cond_using_ranges): Adjust.
26786 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26788 * brig-builtins.def: Added a builtin for class_f64.
26789 * builtin-types.def: Added a builtin type needed by class_f64.
26791 2017-05-03 Jason Merrill <jason@redhat.com>
26793 * timevar.def: Add TV_CONSTEXPR.
26795 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26797 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26799 2017-05-03 Martin Jambor <mjambor@suse.cz>
26801 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26802 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26803 * ipa-cp.c (ipcp_propagate_stage): Do not call
26804 ipa_update_after_lto_read.
26805 * ipa-inline.c (ipa_inline): Likewise.
26807 2017-05-03 Martin Jambor <mjambor@suse.cz>
26809 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26810 tag. Added a default constructor and a destructor.
26811 (ipa_edge_args_sum_t): New class;
26812 (ipa_edge_args_sum): Declare.
26813 (ipa_edge_args_vector): Remove declaration.
26814 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26815 (ipa_free_edge_args_substructures): Remove declaration.
26816 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26817 (ipa_edge_args_info_available_for_edge_p): Likewise.
26818 * ipa-prop.c (ipa_edge_args_vector): Removed.
26819 (edge_removal_hook_holder): Likewise.
26820 (edge_duplication_hook_holder): Likewise.
26821 (ipa_edge_args_sum): New variable.
26822 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26823 ipa_edge_args_vector.
26824 (ipa_free_edge_args_substructures): Likewise.
26825 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26826 ipa_edge_args_vector.
26827 (ipa_edge_removal_hook): Turned into method
26828 ipa_edge_args_sum_t::remove.
26829 (ipa_edge_duplication_hook): Turned into method
26830 ipa_edge_args_sum_t::duplicate.
26831 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26832 registering edge hooks.
26833 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26834 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26835 ipa_edge_args_sum instead of ipa_edge_args_vector.
26836 * ipa-profile.c (ipa_profile): Likewise.
26838 2017-05-03 Martin Jambor <mjambor@suse.cz>
26840 * symbol-summary.h (function_summary): New method exists.
26841 (function_summary::symtab_removal): Deallocate through release.
26842 (call_summary): New class.
26843 (gt_ggc_mx): New overload.
26844 (gt_pch_nx): Likewise.
26845 (gt_pch_nx): Likewise.
26847 2017-05-03 Jeff Law <law@redhat.com>
26849 PR tree-optimization/78496
26850 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26851 from simplify_cond_using_ranges. Split off code to walk
26852 backwards through casts into ...
26853 (simplify_cond_using_ranges_2): New function.
26854 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26855 (execute_vrp): After identifying jump threads, call
26856 simplify_cond_using_ranges_2.
26858 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26861 * ipa-inline.h (inline_summary): Add ctor.
26862 (create_ggc): Do not use ggc_cleared_alloc.
26864 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26865 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26867 * gcc.c (handle_braces): Support escaping in switch matching
26869 * doc/invoke.texi (Spec Files): Document it.
26870 Remove superfluous @code markup in items.
26872 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26874 * diagnostic-show-locus.c (struct column_range): New struct.
26875 (get_affected_columns): New function.
26876 (get_printed_columns): New function.
26877 (struct correction): New struct.
26878 (correction::ensure_capacity): New function.
26879 (correction::ensure_terminated): New function.
26880 (struct line_corrections): New struct.
26881 (line_corrections::~line_corrections): New dtor.
26882 (line_corrections::add_hint): New function.
26883 (layout::print_trailing_fixits): Reimplement in terms of the new
26885 (selftest::test_overlapped_fixit_printing): New function.
26886 (selftest::diagnostic_show_locus_c_tests): Call it.
26888 2017-05-03 Nathan Sidwell <nathan@acm.org>
26890 Canonicalize canonical type hashing
26891 * tree.h (type_hash_canon_hash): Declare.
26892 * tree.c (type_hash_list, attribute_hash_list): Move into
26893 type_hash_canon_hash.
26894 (build_type_attribute_qual_variant): Break out hash code calc into
26895 type_hash_canon_hash.
26896 (type_hash_canon_hash): New. Generic type hash computation.
26897 (build_range_type_1, build_array_type_1, build_function_type,
26898 build_method_type_directly, build_offset_type, build_complex_type,
26899 make_vector_type): Call it.
26901 2017-05-03 Richard Biener <rguenther@suse.de>
26903 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26904 When all DRs have unknown misaligned do not always peel
26905 when there is a store but apply the same costing model as if
26906 there were only loads.
26908 2017-05-03 Richard Biener <rguenther@suse.de>
26911 PR tree-optimization/80492
26912 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26913 compare_base_decls returning dont-know properly.
26915 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26917 * config/arm/iterators.md (CCSI): New mode iterator.
26918 (arch): New mode attribute.
26919 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26920 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26921 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26922 code iterator for success result mode.
26923 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26924 the corresponding new insn generators.
26926 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26929 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26930 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26932 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26934 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26935 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26936 (DDR_A): Wrap DDR argument in brackets.
26937 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26938 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26939 (DDR_REVERSED_P): Likewise.
26941 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26943 PR tree-optimization/79472
26944 * tree-switch-conversion.c (struct switch_conv_info): Add
26945 contiguous_range and default_case_nonstandard fields.
26946 (collect_switch_conv_info): Compute contiguous_range and
26947 default_case_nonstandard fields, don't clear final_bb if
26948 contiguous_range and only the default case doesn't have the required
26950 (check_all_empty_except_final): Set default_case_nonstandard instead
26951 of failing if contiguous_range and the default case doesn't have empty
26953 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26954 and only the default case doesn't have the required constants. Skip
26956 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26957 if default_case_nonstandard.
26958 (build_constructors): Build constant 1 just once. Assert that default
26959 values aren't inserted in between cases if contiguous_range. Skip
26961 (build_arrays): Skip virtual phis.
26962 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26963 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26964 Handle virtual phis.
26965 (gen_inbound_check): Handle default_case_nonstandard case.
26966 (process_switch): Adjust check_final_bb caller. Call
26967 gather_default_values with the first non-default case instead of
26968 default case if default_case_nonstandard.
26970 2017-05-02 Nathan Sidwell <nathan@acm.org>
26972 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26973 check. Fix formatting.
26975 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26977 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26978 errors when comparing specialized and unspecialized times.
26980 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26982 * diagnostic-show-locus.c
26983 (layout::should_print_annotation_line_p): Make private.
26984 (layout::print_annotation_line): Make private.
26985 (layout::annotation_line_showed_range_p): Make private.
26986 (layout::show_ruler): Make private.
26987 (layout::print_source_line): Make private. Pass in line and
26988 line_width, rather than calling location_get_source_line. Drop
26990 (layout::print_leading_fixits): New method.
26991 (layout::print_any_fixits): Rename to...
26992 (layout::print_trailing_fixits): ...this, and make private.
26993 Don't print newline fixits.
26994 (diagnostic_show_locus): Move logic for printing one row into...
26995 (layout::print_line): ...this new function. Move the
26996 location_get_source_line call and error-handling from
26997 print_source_line to here. Call print_leading_fixits, and rename
26998 print_any_fixits to print_trailing_fixits.
26999 (selftest::test_fixit_insert_containing_newline): Update now that
27000 newlines are partially supported.
27001 (selftest::test_fixit_insert_containing_newline_2): New test.
27002 (selftest::test_fixit_replace_containing_newline): Update comments.
27003 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27004 * edit-context.c (class added_line): New class.
27005 (class edited_line): Describe newline handling in comment.
27006 (edited_line::actually_edited_p): New method.
27007 (edited_line::print_content): Delete redundant decl.
27008 (edited_line::m_predecessors): New field.
27009 (edited_file::print_content): Call edited_line::print_content.
27010 (edited_file::print_diff): Update to support newlines.
27011 (edited_file::print_diff_hunk): Likewise.
27012 (edited_file::print_run_of_changed_lines): New function.
27013 (edited_file::print_diff_line): Convert to...
27014 (print_diff_line): ...this.
27015 (edited_file::get_effective_line_count): New function.
27016 (edited_line::edited_line): Initialize new field m_predecessors.
27017 (edited_line::~edited_line): Clean up m_predecessors.
27018 (edited_line::apply_fixit): Handle newlines.
27019 (edited_line::get_effective_line_count): New function.
27020 (edited_line::print_content): New function.
27021 (edited_line::print_diff_lines): New function.
27022 (selftest::test_applying_fixits_insert_containing_newline): New
27024 (selftest::test_applying_fixits_replace_containing_newline): New
27026 (selftest::insert_line): New function.
27027 (selftest::test_applying_fixits_multiple_lines): Add example of
27029 (selftest::edit_context_c_tests): Call the new tests.
27031 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27033 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27034 parameter cand. Update dump information.
27035 (get_computation_cost): Update uses.
27037 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27039 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27040 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27041 (get_computation_at, rewrite_use_address): Update use of
27042 get_computation_aff.
27044 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27046 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27047 (get_computation): Delete.
27048 (get_computation_cost): Implement like get_computation_cost_at.
27049 Use get_computation_at.
27050 (get_computation_cost_at): Delete.
27051 (rewrite_use_nonlinear_expr): Use get_computation_at.
27052 (rewrite_use_compare, remove_unused_ivs): Ditto.
27054 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27056 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27058 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27060 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27061 (ivopts_global_cost_for_size): Rename parameter and update uses.
27062 (iv_ca_recount_cost): Update uses.
27063 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27064 candidates seperately in n_invs and n_cands.
27065 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27069 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27070 (find_inv_vars_cb): New.
27071 (find_depends): Renamed to ...
27072 (find_inv_vars): ... this.
27073 (add_candidate_1, force_var_cost): Call find_inv_vars.
27074 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27076 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27078 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27079 inv_vars. Add inv_exprs.
27080 (struct iv_cand): Rename depends_on to inv_vars.
27081 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27082 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27083 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27084 (dump_cand): Dump inv_vars.
27085 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27086 (record_invariant, find_depends, add_candidate_1): Ditto.
27087 (set_group_iv_cost, force_var_cost): Ditto.
27088 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27089 (get_computation_cost_at, get_computation_cost): Ditto.
27090 (determine_group_iv_cost_generic): Ditto.
27091 (determine_group_iv_cost_address): Ditto.
27092 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27093 (determine_group_iv_costs): Ditto.
27094 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27095 (iv_ca_set_remove_invariants): Renamed to ...
27096 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27097 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27098 (iv_ca_set_add_invariants): Renamed to ...
27099 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27100 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27101 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27102 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27103 (create_new_ivs): Remove useless dump.
27105 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27107 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27109 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27110 (iv_ca_set_no_cp, create_new_iv): Ditto.
27112 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27114 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27116 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27118 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27119 function tree_check2.
27121 2017-05-02 Martin Liska <mliska@suse.cz>
27123 * doc/gcov.texi: Add missing preposition.
27124 * gcov.c (function_info::function_info): Properly fill up
27125 all member variables.
27127 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27129 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27131 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27133 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27135 2017-05-02 Martin Liska <mliska@suse.cz>
27138 * lto-streamer-in.c (lto_read_tree_1): Remove
27139 LTO_STREAMER_DEBUG.
27140 * lto-streamer.c (struct tree_hash_entry): Likewise.
27141 (struct tree_entry_hasher): Likewise.
27142 (tree_entry_hasher::hash): Likewise.
27143 (tree_entry_hasher::equal): Likewise.
27144 (lto_streamer_init): Likewise.
27145 (lto_orig_address_map): Likewise.
27146 (lto_orig_address_get): Likewise.
27147 (lto_orig_address_remove): Likewise.
27148 * lto-streamer.h: Likewise.
27149 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27150 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27152 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27154 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27155 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27156 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27157 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27158 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27159 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27160 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27161 (mm_maskz_sub_ss): New intrinsics.
27162 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27163 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27164 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27165 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27166 (__builtin_ia32_subss_mask_round): New builtins.
27167 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27168 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27169 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27171 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27172 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27174 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27177 2017-05-02 Martin Jambor <mjambor@suse.cz>
27179 PR tree-optimization/78687
27180 * tree-sra.c (access): New field parent.
27181 (process_subtree_disqualification): New function.
27182 (disqualify_candidate): Call it.
27183 (build_accesses_from_assign): Reset write flag if creating an
27185 (build_access_subtree): Fill in parent field and also prpagate
27186 down grp_write flag.
27187 (create_artificial_child_access): New parameter set_grp_write, set
27188 grp_write to its value.
27189 (propagate_subaccesses_across_link): Also propagate grp_write flag
27191 (propagate_all_subaccesses): Push the closest parent back to work
27192 queue if add_access_to_work_queue returned true.
27194 2017-05-02 Richard Biener <rguenther@suse.de>
27196 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27197 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27198 -fstrict-overflow documentation.
27199 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27200 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27201 flag_strict_overflow.
27202 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27203 * lto-opts.c (lto_write_options): Do not stream it.
27204 * lto-wrapper.c (merge_and_complain): Do not handle it.
27205 * opts.c (default_options_table): Do not set -fstrict-overflow.
27206 (finish_options): Likewise do not clear it when sanitizing.
27207 * simplify-rtx.c (simplify_const_relational_operation): Do not
27208 test flag_strict_overflow.
27210 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27212 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27213 using enabled attribute.
27214 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27215 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27216 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27217 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27218 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27219 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27220 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27221 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27222 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27223 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27225 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27227 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27229 2017-05-02 Richard Biener <rguenther@suse.de>
27231 PR tree-optimization/80591
27233 2017-04-10 Richard Biener <rguenther@suse.de>
27235 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27238 2017-05-02 Richard Biener <rguenther@suse.de>
27240 PR tree-optimization/80549
27241 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27242 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27243 headers if they do not have a preheader.
27245 2017-05-02 Martin Liska <mliska@suse.cz>
27248 * common.opt: Fix typo.
27249 * doc/invoke.texi: Likewise.
27251 2017-05-01 Jan Beulich <jbeulich@suse.com>
27253 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27254 swapping, add (x,x,m,x,n) alternative.
27256 2017-05-01 Nathan Sidwell <nathan@acm.org>
27258 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27259 unnecessary unadjusted_alignment check.
27261 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27264 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27266 * gimplify.c (gimplify_cilk_detach): New function.
27267 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27268 * tree-core.h: Document EXPR_CILK_SPAWN.
27269 * tree.h (EXPR_CILK_SPAWN): Define.
27271 2017-05-01 David Malcolm <dmalcolm@redhat.com>
27273 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
27274 to use new fixit_hint representation, using the "replace" logic.
27275 (get_line_span_for_fixit_hint): Likewise.
27276 (layout::print_any_fixits): Likewise.
27277 (selftest::test_one_liner_many_fixits): Rename to...
27278 (selftest::test_one_liner_many_fixits_1): ...this, and update
27279 comment and expected output to reflect that the multiple fix-it
27280 hints are now consolidated into one insertion.
27281 (selftest::test_one_liner_many_fixits_2): New test.
27282 (selftest::test_diagnostic_show_locus_one_liner): Update for
27284 (selftest::test_fixit_consolidation): Update for fix-it API
27286 * diagnostic.c (print_parseable_fixits): Likewise.
27287 * edit-context.c (edited_line::m_line_events): Convert from
27288 auto_vec <line_event *> to auto_vec <line_event>.
27289 (class line_event): Convert from abstract base class to a concrete
27290 class, taking over the role of replace_event.
27291 (class insert_event): Delete.
27292 (class replace_event): Rename to class line_event. Convert to
27294 (edit_context::add_fixits): Reimplement.
27295 (edit_context::apply_insert): Delete.
27296 (edit_context::apply_replace): Rename to...
27297 (edit_context::apply_fixit): ...this. Convert to half-open range.
27298 (edited_file::apply_insert): Delete.
27299 (edited_file::apply_replace): Rename to...
27300 (edited_file::apply_fixit): ...this.
27301 (edited_line::~edited_line): Drop deletion of events.
27302 (edited_line::apply_insert): Delete.
27303 (edited_line::apply_replace): Rename to...
27304 (edited_line::apply_fixit): ...this. Convert to half-open range.
27305 Update for change to type of m_line_events.
27306 * edit-context.h (edit_context::apply_insert): Delete.
27307 (edit_context::apply_replace): Rename to...
27308 (edit_context::apply_fixit): ...this.
27310 2017-05-01 Martin Sebor <msebor@redhat.com>
27312 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27315 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27318 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27319 __get_cpuid_max returns 0.
27320 (__get_cpuid_count): Ditto.
27322 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27324 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27325 replacement expression is another instance of one of its arguments.
27327 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27330 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27331 check for stack push/pop autoinc.
27332 * config/i386/i386.c (ix86_agi_dependent): Return false
27333 if the only reason why modified_in_p returned true is that
27334 addr is SP based and set_insn is a push or pop.
27336 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27338 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27341 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27344 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27345 (account_size_time): Use two predicates - exec_pred and
27347 (evaluate_conditions_for_known_args): Compute both clause and
27349 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27350 (inline_summary_t::duplicate): Update.
27351 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27353 (compute_inline_parameters): Likewise.
27354 (estimate_edge_size_and_time): Update caluclation of time.
27355 (estimate_node_size_and_time): Compute both time and nonspecialized
27357 (estimate_ipcp_clone_size_and_time): Update.
27358 (inline_merge_summary): Update.
27359 (do_estimate_edge_time): Update.
27360 (do_estimate_edge_size): Update.
27361 (do_estimate_edge_hints): Update.
27362 (inline_read_section, inline_write_summary): Stream both new predicates.
27363 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27365 (compute_inlined_call_time): Cleanup.
27366 (big_speedup_p): Update.
27367 (edge_badness): Update.
27368 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27369 (size_time_entry): Replace predicate by exec_predicate and
27370 nonconst_predicate.
27371 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27372 (estimate_edge_time): Return also nonspec_time.
27373 (reset_edge_growth_cache): Update.
27375 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27377 PR rtl-optimization/80491
27378 * ifcvt.c (noce_process_if_block): When looking for x setter
27379 with missing else_bb, don't check only the insn right before
27380 cond_earliest, but look for the last insn that x is modified in
27381 within the same bb.
27383 PR rtl-optimization/80491
27384 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27386 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27388 PR tree-optimization/80487
27389 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27391 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27393 PR tree-optimization/79697
27394 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27395 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27396 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27398 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27399 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27401 2017-04-28 Martin Sebor <msebor@redhat.com>
27403 PR tree-optimization/80523
27404 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27405 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27407 (maybe_warn, format_directive, parse_directive): Use new functions.
27408 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27410 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27412 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27414 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27416 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27417 target_header_dir): Set correctly.
27418 * configure: Regenerated.
27419 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27420 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27421 instead of SYSTEM_HEADER_DIR.
27423 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27425 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27426 (estimate_local_effects): Likewise.
27427 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27428 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27429 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27430 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27431 do_estimate_edge_time, estimate_edge_time): Likewise.
27432 * ipa-inline-analysis.c (estimate_node_size_and_time,
27433 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27434 (estimate_time_after_inlining): Remove.
27436 2017-04-28 Martin Liska <mliska@suse.cz>
27438 * doc/gcov.texi: Enhance documentation of gcov.
27440 2017-04-28 Martin Liska <mliska@suse.cz>
27442 * doc/gcov.texi: Sort options in alphabetic order.
27443 * doc/gcov-dump.texi: Likewise.
27444 * doc/gcov-tool.texi: Likewise.
27445 * gcov.c (print_usage): Likewise.
27446 * gcov-dump.c (print_usage): Likewise.
27447 * gcov-tool.c (print_merge_usage_message): Likewise.
27448 (print_rewrite_usage_message): Likewise.
27449 (print_overlap_usage_message): Likewise.
27451 2017-04-28 Martin Liska <mliska@suse.cz>
27453 PR gcov-profile/53915
27454 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27456 2017-04-28 Martin Liska <mliska@suse.cz>
27458 PR gcov-profile/79891
27459 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27460 is marked by compiler as living on a line.
27461 (get_cycles_count): Remove usage of the union.
27462 (output_intermediate_file): Likewise.
27463 (find_source): Fix GNU coding style.
27464 (accumulate_line_counts): Remove old non-all block mode.
27465 (output_lines): Remove usage of the union.
27466 * profile.c (output_location): Include all BBs, even if
27467 belonging to a same line (and file) as a previous BB.
27469 2017-04-28 Martin Liska <mliska@suse.cz>
27471 * gcov.c (process_args): Handle new argument 'w'.
27472 (read_graph_file): Assign ID to BBs.
27473 (output_branch_count): Display BB # if verbose flag is set.
27474 (output_lines): Likewise for arcs.
27475 (print_usage): Add '--verbose' option help.
27476 * doc/gcov.texi: Document --verbose (-w) option.
27478 2017-04-28 Martin Liska <mliska@suse.cz>
27480 * gcov.c (struct block_location_info): New struct.
27481 (process_file): Fill up the new structure.
27482 (read_graph_file): Replace usage of encoding by the newly added
27484 (add_line_counts): Likewise.
27485 (accumulate_line_counts): Remove usage of the union.
27486 (function_info::function_info): New function.
27487 (function_info::~function_info): Likewise.
27488 (process_file): Call delete instead of release_function.
27489 (release_function): Release the function.
27490 (release_structures): Call delete instead of release_function.
27491 (solve_flow_graph): Replace usage of num_blocks.
27492 (find_exception_blocks): Likewise.
27493 (output_lines): Fix GNU coding style.
27495 2017-04-28 Martin Liska <mliska@suse.cz>
27498 * coverage.c (coverage_remove_note_file): New function.
27499 * coverage.h: Declare the function.
27500 * toplev.c (finalize): Clean if an error has been seen.
27502 2017-04-28 Martin Liska <mliska@suse.cz>
27504 PR gcov-profile/80031
27505 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27506 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27507 * gcov.c (read_graph_file): Read just number of blocks.
27508 * profile.c (branch_prob): Do not stream 0 flags per a basic
27511 2017-04-28 Martin Liska <mliska@suse.cz>
27513 * gcov-dump.c (tag_*): Add new argument to declarations.
27514 (dump_gcov_file): Likewise.
27515 (tag_blocks): Add and use new argument depth.
27516 (tag_arcs): Likewise.
27517 (tag_lines): Likewise.
27518 (tag_counters): Likewise.
27519 (tag_summary): Likewise.
27520 (dump_working_sets): Use depth to do a proper indentation.
27522 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27525 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27526 * symtab.c (symtab_node::debug_symtab): Move definition here.
27528 2017-04-28 Richard Biener <rguenther@suse.de>
27530 * lto-streamer.h (LTO_major_version): Bump to 7.
27532 2017-04-28 Richard Biener <rguenther@suse.de>
27534 * tree-vrp.c (assert_info): New struct.
27535 (add_assert_info): New helper.
27536 (register_edge_assert_for_2): Refactor to add asserts to a vector
27538 (register_edge_assert_for_1): Likewise.
27539 (register_edge_assert_for): Likewise.
27540 (finish_register_edge_assert_for): New helper actually registering
27541 asserts where live on edge.
27542 (find_conditional_asserts): Adjust.
27543 (find_switch_asserts): Likewise.
27544 (evrp_dom_walker::try_find_new_range): Generalize.
27545 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27547 2017-04-27 Marek Polacek <polacek@redhat.com>
27550 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27551 arg10 and arg11 to itype.
27553 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27555 * doc/extend.texi (Object Size Checking): Improve grammar.
27557 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27560 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27561 that the logic for permitting reciprocal estimates matches that
27564 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27567 * tree.c (type_cache_hasher::equal): Only compare
27568 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27569 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27570 non-aggregate element types.
27571 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27572 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27574 2017-04-27 Richard Biener <rguenther@suse.de>
27576 PR middle-end/80533
27577 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27578 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27579 keeping a reference to a trailing array.
27581 2017-04-27 Richard Biener <rguenther@suse.de>
27583 PR middle-end/80539
27584 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27585 being in loop-closed SSA form conservatively.
27586 (chrec_fold_multiply_poly_poly): Likewise.
27588 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27590 PR middle-end/79665
27591 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27592 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27594 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27597 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27598 (aarch64_function_arg_alignment): Return unsigned int again, but still
27599 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27600 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27601 Don't emit -Wpsabi note.
27602 (aarch64_function_arg_boundary): Likewise.
27603 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27606 2017-04-26 Nathan Sidwell <nathan@acm.org>
27608 * tree.h (crc32_unsigned_n): Declare.
27609 (crc32_unsigned, crc32_unsigned): Make inline.
27610 * tree.c (crc32_unsigned_bits): Replace with ...
27611 (crc32_unsigned_n): ... this.
27612 (crc32_unsigned, crc32_byte): Remove.
27613 (crc32_string): Remove unnecessary braces.
27615 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27617 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27618 * ipa-inline-analysis.c (MAX_TIME): Remove.
27619 (account_size_time): Use sreal for time.
27620 (dump_inline_summary): Update.
27621 (estimate_function_body_sizes): Update.
27622 (estimate_edge_size_and_time): Update.
27623 (estimate_calls_size_and_time): Update.
27624 (estimate_node_size_and_time): Update.
27625 (inline_merge_summary): Update.
27626 (inline_update_overall_summary): Update.
27627 (estimate_time_after_inlining): Update.
27628 (inline_read_section): Update.
27629 (inline_write_summary): Update.
27630 * ipa-inline.c (compute_uninlined_call_time): Update.
27631 (compute_inlined_call_time): Update.
27632 (recursive_inlining): Update.
27633 (inline_small_functions): Update.
27634 (dump_overall_stats): Update.
27635 * ipa-inline.h: Include sreal.h.
27636 (size_time_entry): Turn time to sreal.
27637 (inline_summary): Turn self_time nad time to sreal.
27639 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27641 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27643 (sreal::stream_out, sreal::stream_in): New.
27644 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27646 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27648 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27651 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27654 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27656 (dimode_scalar_chain::compute_convert_gain): Ditto.
27657 (dimode_scalar_chain::make_vector_copies): Ditto.
27658 (dimode_scalar_chain::convert_reg): Ditto.
27659 (dimode_scalar_chain::convert_insn): Ditto.
27660 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27661 (VI248_AVX512BW_1): New mode iterator.
27662 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27663 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27666 2017-04-25 Martin Sebor <msebor@redhat.com>
27668 PR tree-optimization/80497
27669 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27670 constants are representable in HOST_WIDE_INT.
27671 (parse_directive): Ditto.
27673 2017-04-25 Martin Sebor <msebor@redhat.com>
27676 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27677 (new_zero_array): Adjust signature.
27678 (dom_info::dom_init): Used unsigned rather that size_t.
27679 (dom_info::dom_info): Same.
27681 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27682 Jakub Jelinek <jakub@redhat.com>
27685 * config/arm/arm.c: Include gimple.h.
27686 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27687 returns negative, increment ncrn only if it returned positive.
27688 (arm_needs_doubleword_align): Return int instead of bool,
27689 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27690 members, but if there is any such non-FIELD_DECL
27691 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27692 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27693 returns negative, increment nregs only if it returned positive.
27694 (arm_setup_incoming_varargs): Likewise.
27695 (arm_function_arg_boundary): Emit -Wpsabi note if
27696 arm_needs_doubleword_align returns negative, return
27697 DOUBLEWORD_ALIGNMENT only if it returned positive.
27699 2017-04-25 Marek Polacek <polacek@redhat.com>
27702 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27703 first argument to type.
27705 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27709 type checks to test for compatibility instead of equality.
27711 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27712 Jakub Jelinek <jakub@redhat.com>
27715 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27717 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27718 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27719 the alignment computation, but return their maximum in warn_alignment.
27720 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27721 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27723 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27724 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27727 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27729 * config/arc/simdext.md (dmpyh): Fix typo.
27731 2017-04-25 Richard Biener <rguenther@suse.de>
27733 PR tree-optimization/80492
27734 * alias.c (compare_base_decls): Handle registers with asm
27735 specification conservatively.
27736 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27737 compare_base_decls returning dont-know properly.
27739 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27741 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27742 (legitimate_offset_address_p): New function.
27743 (arc_legitimate_address_p): Use above function.
27745 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27747 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27749 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27751 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27752 ACCH registers whenever they are available.
27754 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27756 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27757 double regs fix when not used.
27759 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27761 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27763 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27764 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27766 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27768 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27769 when emitting short ADD instructions.
27771 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27773 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27775 (cmpsi_cc_c_insn): Likewise.
27776 (cbranchsi4_scratch): Compute proper instruction length using
27777 compact_hreg_operand.
27778 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27780 2017-04-25 Richard Biener <rguenther@suse.de>
27782 PR middle-end/80509
27783 * passes.c (pass_manager::pass_manager): Initialize
27784 m_name_to_pass_map.
27786 2017-04-25 Richard Biener <rguenther@suse.de>
27788 PR tree-optimization/79201
27789 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27791 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27794 * config/s390/vector.md: Split MEM->GPR vector moves for
27795 non-s_operand addresses.
27797 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27800 * config/s390/predicates.md (reload_const_wide_int_operand): New
27802 * config/s390/s390.md ("movti"): Remove d/P alternative.
27803 ("movti_bigconst"): New pattern definition.
27805 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27808 * s390-protos.h (s390_expand_cs_hqi): Removed.
27809 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27810 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27811 modes as well as CCZ1mode and CCZmode.
27812 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27813 signature of s390_emit_compare_and_swap.
27814 (s390_expand_cs_hqi): Likewise, make static.
27815 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27816 compare-and-swap, in some cases.
27817 (s390_expand_cs): Wrapper function.
27818 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27820 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27821 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27822 patterns for small and large integers. Forbid symref memory operands.
27823 Move expander to s390.c. Require cc register.
27824 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27825 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27826 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27827 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27828 symref memory operands. Remove CC mode and call s390_match_ccmode
27830 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27832 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27834 * config/s390/s390.md (define_peephole2): New peephole to help
27835 combining the load-and-test pattern with volatile memory.
27837 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27839 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27840 with CCZmode for TARGET_Z196.
27842 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27844 PR rtl-optimization/80501
27845 * combine.c (make_compound_operation_int): Set subreg_code to SET
27846 even for AND with mask of the sign bit of mode.
27848 PR rtl-optimization/80500
27849 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27850 sum's initial value.
27852 2017-04-25 Julian Brown <julian@codesourcery.com>
27853 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27855 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27857 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27859 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27861 2017-04-25 Julian Brown <julian@codesourcery.com>
27862 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27864 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27865 (thunderx2t99_sha): New Reservation.
27867 2017-04-25 Julian Brown <julian@codesourcery.com>
27868 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27870 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27871 type for 1-element load.
27873 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27875 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27877 2017-04-24 Martin Jambor <mjambor@suse.cz>
27879 PR tree-optimization/80293
27880 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27881 char arrays not totally scalarizable if it is false.
27882 (analyze_all_variable_accesses): Pass correct value in the new
27883 parameter. Add a statistics counter.
27885 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27887 PR middle-end/79931
27888 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27890 2017-04-24 Richard Biener <rguenther@suse.de>
27892 PR tree-optimization/80494
27893 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27894 out for complex types.
27896 2017-04-24 Richard Biener <rguenther@suse.de>
27898 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27899 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27900 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27901 (DFS): Adjust and never fail.
27902 (sccvn_dom_walker::fail): Remove.
27903 (sccvn_dom_walker::before_dom_children): Adjust.
27904 (run_scc_vn): Likewise and never fail.
27905 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27906 (pass_fre::execute): Likewise.
27908 2017-04-24 Richard Biener <rguenther@suse.de>
27910 PR tree-optimization/79725
27911 * tree-ssa-sink.c (statement_sink_location): Return whether
27912 failure reason was zero uses. Move that check later.
27913 (sink_code_in_bb): Deal with zero uses by removing the stmt
27916 2017-04-24 Richard Biener <rguenther@suse.de>
27919 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27920 pointer-based references.
27922 2017-04-24 Richard Biener <rguenther@suse.de>
27925 * pass_manager.h (pass_manager::operator new): Remove.
27926 (pass_manager::operator delete): Likewise.
27927 * passes.c (pass_manager::operator new): Remove.
27928 (pass_manager::operator delete): Likewise.
27929 (pass_manager::pass_manager): Zero individual pass members.
27931 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27934 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27935 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27936 Check "XEXP (src, 1)" operand here.
27937 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27938 Check "XEXP (src, 1)" operand here.
27939 (dimode_scalar_chain::make_vector_copies): Detect count register
27940 of a shift instruction. Zero extend count register from QImode
27941 to DImode to satisfy vector shift pattern count operand predicate.
27942 Substitute vector shift count operand with a DImode copy.
27943 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27946 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27948 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27949 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27950 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27951 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27952 (UNSPEC_NOREX_MEM): Remove definition.
27954 2017-04-21 Richard Biener <rguenther@suse.de>
27956 PR tree-optimization/79547
27957 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27958 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27959 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27960 without any constraints.
27962 2017-04-21 Richard Biener <rguenther@suse.de>
27964 PR tree-optimization/78847
27965 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27967 2017-04-21 Richard Biener <rguenther@suse.de>
27969 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27970 (build_distinct_type_copy): Likewise.
27971 (build_variant_type_copy): Likewise.
27972 * tree.c (build_qualified_type): Pass down mem-stat info.
27973 (build_distinct_type_copy): Likewise.
27974 (build_variant_type_copy): Likewise.
27976 2017-04-21 Richard Biener <rguenther@suse.de>
27978 PR tree-optimization/80237
27979 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27981 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27982 for a simplified result.
27984 2016-04-21 Richard Biener <rguenther@suse.de>
27986 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27987 sth as strict as a simple_iv but a chrec without symbols and an
27988 operand defined in the loop we are peeling (and not some subloop).
27989 (propagate_constants_for_unrolling): Propagate all constants.
27991 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27994 * config/i386/i386.c (print_reg): Remove assert for disalowed
27995 regno values, call output_operand_lossage instead.
27997 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28000 * config/i386/constraints.md (Yc): New register constraint.
28001 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28002 Use Yc constraint for alternative 2 of operand 0. Remove
28003 preferred_for_speed attribute.
28005 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28007 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28008 lastprivate clauses in SIMT case.
28010 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28012 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28014 2017-04-20 Richard Biener <rguenther@suse.de>
28016 PR tree-optimization/57796
28017 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28018 as N scalar stores.
28019 (vect_model_load_cost): Cost gathers as N scalar loads.
28021 2017-04-20 Richard Biener <rguenther@suse.de>
28023 * ggc-page.c (ggc_allocated_p): Rename to ...
28024 (safe_lookup_page_table_entry): ... this and return the lookup
28026 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28028 2017-04-20 Richard Biener <rguenther@suse.de>
28030 PR tree-optimization/80453
28031 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28032 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28033 from the conditions.
28034 (vn_phi_eq): Pass them down.
28035 (vn_phi_lookup): Record them.
28036 (vn_phi_insert): Likewise.
28038 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28040 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28041 uninitialized variable warning to avoid buffer overrun.
28043 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28046 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28047 is suppressed for '{ 0 }' in C.
28049 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28051 * BASE-VER: Set to 8.0.0.
28053 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28055 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28056 priority .init_array and .fini_array section with SECTION_NOTYPE
28059 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28061 PR middle-end/80423
28062 * tree.h (build_array_type): Add typeless_storage default argument.
28063 * tree.c (type_cache_hasher::equal): Also compare
28064 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28065 (build_array_type): Add typeless_storage argument, set
28066 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28068 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28069 (build_array_type): Likewise. Add typeless_storage argument.
28071 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28072 Jakub Jelinek <jakub@redhat.com>
28074 PR tree-optimization/80426
28075 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28076 operation on symbolic operands, also compute the overflow for the
28077 invariant part when the operation degenerates into a negation.
28079 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28082 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28083 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28086 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28088 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28091 * config/avr/avr.c (tree.h): Include it.
28092 (cgraph.h): Include it.
28093 (avr_encode_section_info): Don't warn for uninitialized progmem
28094 variable if it's just an alias.
28096 2017-04-19 Richard Biener <rguenther@suse.de>
28099 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28100 when needed by AutoPGO.
28102 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28105 * doc/lto.texi: Remove an extra 'that'.
28107 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28109 PR rtl-optimization/80429
28110 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28111 are only used in debug insns.
28113 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28114 Vladimir Makarov <vmakarov@redhat.com>
28116 * config/sparc/predicates.md (input_operand): Add comment. Return
28117 true for any memory operand when LRA is in progress.
28118 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28120 2017-04-18 Jeff Law <law@redhat.com>
28123 * mips.md ({return,simple_return}_internal): Do not overwrite
28126 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28128 PR tree-optimization/80443
28129 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28130 instead of adding 1, subtract -1 and similarly instead of subtracting
28133 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28135 PR rtl-optimization/80357
28136 * haifa-sched.c (tmp_bitmap): New variable.
28137 (model_recompute): Handle duplicate use records.
28138 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28139 (free_global_sched_pressure_data): Free it.
28141 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28144 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28145 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28146 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28147 instead of SYSTEM_HEADER_DIR.
28149 2017-04-18 Jeff Law <law@redhat.com>
28151 PR middle-end/80422
28152 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28153 predecessors after walking up the insn chain.
28155 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28158 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28159 sizetype type into debug info.
28161 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28164 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28165 unneeded test for TARGET_UPPER_REGS_SF.
28166 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28168 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28171 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28172 instead of gsi_after_labels.
28174 2017-04-18 Jeff Law <law@redhat.com>
28176 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28180 2017-04-13 Jeff Law <law@redhat.com>
28181 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28182 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28184 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28187 * config/avr/avr.c (intl.h): Include it.
28188 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28190 2017-04-18 Martin Liska <mliska@suse.cz>
28192 PR gcov-profile/78783
28193 * gcov-tool.c (gcov_output_files): Validate that destination
28194 file is either removed by the tool or by a user.
28196 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28197 Guy Benyei <guybe@mellanox.com>
28199 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28200 block, and do not negate it, the stored id is already negative.
28202 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28204 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28206 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28209 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28210 masks of options that should be turned off if the VSX vector
28211 options are turned off.
28212 (OTHER_P8_VECTOR_MASKS): Likewise.
28213 (OTHER_VSX_VECTOR_MASKS): Likewise.
28214 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28215 rs6000_disable_incompatible_switches to validate no type switches
28217 (rs6000_incompatible_switch): New function to disallow turning on
28218 other vector options if -mno-vsx, -mno-power8-vector, or
28219 -mno-power9-vector are specified.
28221 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28223 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28225 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28227 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28228 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28229 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28230 (ARG_POINTER_CFA_OFFSET): Likewise.
28232 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28234 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28235 conditions to take advantage of various optimizations.
28237 2017-04-13 Jeff Law <law@redhat.com>
28239 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28240 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28241 (zero_extendsidi2_dext): Likewise.
28243 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28246 * fold-const.c (fold_ternary_loc): Revert
28247 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28250 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28252 PR rtl-optimization/80343
28253 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28254 new scratch pseudo.
28256 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28259 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28260 to ubsan_encode_value.
28262 2017-04-13 Jeff Law <law@redhat.com>
28264 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28265 appearing in DEBUG_INSNs.
28267 2017-04-13 Martin Liska <mliska@suse.cz>
28269 PR gcov-profile/80413
28270 * gcov-io.c (gcov_write_string): Copy to buffer just when
28271 allocated size is greater than zero.
28273 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28276 * dwarf2out.c (decls_for_scope): Ignore declarations of
28277 current_function_decl in BLOCK_NONLOCALIZED_VARS.
28279 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
28282 * ipa-visibility.c (non_local_p): Fix typos.
28283 (localize_node): When localizing symbol in same comdat group,
28284 dissolve the group only when we know external symbols are going
28286 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
28288 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28290 PR tree-optimization/79390
28291 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
28292 order does not result in usable sequence, retry with reversed operand
28298 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
28299 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
28300 op0 instead of fold_convert_loc (loc, type, arg0).
28302 2017-04-12 Jeff Law <law@redhat.com>
28304 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28305 has a delay slot in the generated code.
28307 * config/cris/cris.md (cris_preferred_reload_class): Return
28308 GENNONACR_REGS rather than GENERAL_REGS.
28310 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28313 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28314 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28315 signedness of the result type.
28317 2017-04-12 Richard Biener <rguenther@suse.de>
28318 Jeff Law <law@redhat.com>
28320 PR tree-optimization/80359
28321 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28322 trim stores to TARGET_MEM_REFs.
28324 2017-04-12 Richard Biener <rguenther@suse.de>
28326 PR tree-optimization/79390
28327 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28328 threading case even more.
28330 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28333 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28334 for quad_address_p for TImode, instead of just not indexed_address.
28336 2017-04-12 Richard Biener <rguenther@suse.de>
28337 Bernd Edlinger <bernd.edlinger@hotmail.de>
28339 PR middle-end/79671
28340 * alias.c (component_uses_parent_alias_set_from): Handle
28341 TYPE_TYPELESS_STORAGE.
28342 (get_alias_set): Likewise.
28343 * tree-core.h (tree_type_common): Add typeless_storage flag.
28344 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28345 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28346 for types containing members with TYPE_TYPELESS_STORAGE.
28347 (place_field): Likewise.
28348 (layout_type): Likewise for ARRAY_TYPE.
28349 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28350 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28351 TYPE_TYPELESS_STORAGE.
28352 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28354 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28357 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28358 first argument to type.
28360 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28364 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28365 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28366 (rs6000_expand_binop_builtin): Likewise.
28367 (rs6000_expand_ternop_builtin): Likewise; also add missing
28368 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28369 vshasigma built-ins.
28370 * doc/extend.texi: Document that vec_xxpermdi's third argument
28371 must be a constant.
28373 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28375 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28376 Use shift_const cost parameter when calculating gain of STV shifts.
28378 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28380 PR rtl-optimization/70478
28381 * lra-constraints.c (process_alt_operands): Check memory for
28382 disfavoring memory insn operand.
28384 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28386 PR middle-end/80100
28387 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28388 left shift in unsigned HOST_WIDE_INT type.
28390 PR rtl-optimization/80385
28391 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28392 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28395 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28396 if they have any depend clauses.
28398 2017-04-11 Martin Liska <mliska@suse.cz>
28401 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28402 * ipa-split.c (split_function): Create a local comdat symbol
28403 if caller is in a comdat group.
28405 2017-04-11 Martin Liska <mliska@suse.cz>
28408 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28411 2017-04-11 Martin Sebor <msebor@redhat.com>
28413 PR middle-end/80364
28414 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28415 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28417 (directive::set_width, directive::set_precision, format_character):
28419 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28422 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28425 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28426 conflict, set target->arch_name instead of target->cpu_name.
28428 2017-04-11 Richard Biener <rguenther@suse.de>
28430 PR tree-optimization/80374
28431 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28432 build_zero_cst, remove fold_convertible_p check again.
28434 2017-04-11 Martin Liska <mliska@suse.cz>
28437 * ubsan.c (instrument_object_size): Do not instrument register
28440 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28443 * config/i386/i386-builtin-types.def
28444 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28445 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28446 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28447 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28448 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28449 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28450 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28451 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28452 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28453 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28454 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28455 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28456 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28457 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28458 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28459 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28460 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28461 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28462 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28463 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28464 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28465 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28466 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28467 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28468 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28469 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28470 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28471 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28472 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28473 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28474 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28475 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28476 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28477 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28478 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28479 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28480 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28481 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28482 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28483 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28484 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28485 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28486 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28487 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28488 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28489 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28491 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28492 flag to second_arg_count, handle 4 argument function type _COUNT
28493 aliases, handle second_arg_count on second argument rather than last.
28495 2017-04-10 Jeff Law <law@redhat.com>
28497 PR tree-optimization/80374
28498 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28499 record anything if we can not convert integer_zero_node to the
28502 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28505 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28506 Enhance special handling given to the TARGET_P9_MINMAX option in
28507 relation to certain other options.
28509 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28511 PR tree-optimization/80153
28512 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28513 remove POINTER_PLUS_EXPR's base part directly, rather than through
28516 2017-04-10 Richard Biener <rguenther@suse.de>
28517 Bin Cheng <bin.cheng@arm.com>
28519 PR tree-optimization/80153
28520 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28521 the first element of pointer type aff_tree. Build result expr in
28523 (add_elt_to_tree): Convert to type unconditionally. Remove other
28524 fold_convert calls.
28525 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28526 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28528 2017-04-10 Richard Biener <rguenther@suse.de>
28530 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28533 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28535 PR rtl-optimization/70478
28536 * lra-constraints.c (curr_small_class_check): New.
28537 (update_and_check_small_class_inputs): New.
28538 (process_alt_operands): Update curr_small_class_check. Disfavor
28539 alternative insn memory operands. Check available regs for small
28542 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28545 * config/mips/mips.opt (-mvirt): Update description.
28546 * doc/invoke.texi (-mvirt): Likewise.
28548 2017-04-10 Richard Biener <rguenther@suse.de>
28550 PR middle-end/80362
28551 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28552 looking for NEGATE_EXPR in -A / -B to A / B folding.
28554 2017-04-10 Martin Liska <mliska@suse.cz>
28556 PR gcov-profile/80224
28557 * gcov.c (print_usage): Fix usage string.
28558 (get_gcov_intermediate_filename): Remove.
28559 (output_gcov_file): Use both for normal and intermediate format.
28560 (generate_results): Do not initialize special file for
28561 intermediate format.
28563 2017-04-10 Richard Biener <rguenther@suse.de>
28565 PR tree-optimization/80304
28566 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28569 2017-04-10 Nathan Sidwell <nathan@acm.org>
28572 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28573 (rs6000_init_builtins): Use it.
28575 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28577 * config/arm/arm.md (<mrc>): Add mode to SET source.
28578 (<mrrc>): Likewise.
28580 2017-04-10 Richard Biener <rguenther@suse.de>
28582 PR middle-end/80344
28583 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28585 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28588 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28589 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28590 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28591 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28592 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28593 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28594 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28595 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28596 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28597 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28598 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28599 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28600 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28601 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28602 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28603 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28604 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28605 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28606 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28607 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28608 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28609 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28610 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28612 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28614 PR rtl-optimization/70478
28615 * lra-constraints.c: Reverse the last patch.
28617 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28619 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28620 Add comment for WCHAR_T.
28622 2017-04-08 Martin Liska <mliska@suse.cz>
28625 2017-04-07 Martin Liska <mliska@suse.cz>
28628 * ipa-split.c (split_function): Add function part to a same comdat
28631 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28634 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28636 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28638 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28639 (vec_pairing): ...to this.
28640 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28641 (rs6000_sched_init): Adjust for name change.
28642 (struct rs6000_sched_context): Likewise.
28643 (rs6000_init_sched_context): Likewise.
28644 (rs6000_set_sched_context): Likewise.
28646 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28652 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28654 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28655 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28656 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28658 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28660 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28662 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28664 PR rtl-optimization/70703
28665 * ira-color.c (update_conflict_hard_regno_costs): Use
28666 int64_t instead of HOST_WIDE_INT.
28668 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28670 PR rtl-optimization/70478
28671 * lra-constraints.c (process_alt_operands): Disfavor alternative
28672 insn memory operands.
28674 2017-04-07 Jeff Law <law@redhat.com>
28676 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28677 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28679 2017-04-07 Martin Liska <mliska@suse.cz>
28682 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28683 Show error message instead of an ICE.
28685 2017-04-07 Martin Liska <mliska@suse.cz>
28688 * ipa-split.c (split_function): Add function part to a same comdat
28691 2017-04-07 Richard Biener <rguenther@suse.de>
28693 PR middle-end/80341
28694 * tree.c (get_unwidened): Also handle ! for_type case for
28696 * convert.c (do_narrow): Split out from ...
28697 (convert_to_integer_1): ... here. Do not pass final truncation
28698 type to get_unwidened for TRUNC_DIV_EXPR.
28700 2017-04-07 Richard Biener <rguenther@suse.de>
28702 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28704 (aff_combination_const): Adjust.
28705 (aff_combination_scale): Likewise.
28706 (aff_combination_add_elt): Likewise.
28707 (aff_combination_add_cst): Likewise.
28708 (aff_combination_convert): Likewise.
28709 (add_elt_to_tree): Likewise. Remove unused argument.
28710 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28712 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28714 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28716 * config/arm/arm.c (arm_default_short_enums): Use
28717 ARM_DEFAULT_SHORT_ENUMS.
28718 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28720 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28723 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28724 members with redundant out-of-class redeclaration.
28726 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28729 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28730 * config/i386/i386.md (*zero_extendsidi2):
28731 Add (?*x,*x) and (?*v,*v) alternatives.
28733 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28736 * config/i386/i386.c (ix86_expand_builtin)
28737 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28738 mode from insn data. Convert operands to insn operand mode.
28739 Copy operands that don't satisfy insn predicate to a register.
28741 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28743 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28746 2017-04-06 Richard Biener <rguenther@suse.de>
28748 PR tree-optimization/80334
28749 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28750 preserve alignment of accesses.
28752 2017-04-06 Richard Biener <rguenther@suse.de>
28754 PR tree-optimization/80262
28755 * tree-sra.c (build_ref_for_offset): Preserve address-space
28757 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28758 Drop useless address-space information on MEM_REF offsets.
28760 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28762 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28764 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28766 PR rtl-optimization/70703
28767 * ira-color.c (update_conflict_hard_regno_costs): Use
28768 HOST_WIDE_INT instead of long.
28770 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28773 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28774 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28776 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28777 for x86_64 target. Handle -m3dnowa option.
28779 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28781 PR rtl-optimization/70703
28782 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28783 (update_conflict_hard_regno_costs): Use long instead of unsigned
28784 arithmetic for cost calculation.
28786 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28787 Bernd Edlinger <bernd.edlinger@hotmail.de>
28790 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28793 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28796 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28797 ptr_mode with Pmode throughout.
28798 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28799 into probe_stack_range and use DImode.
28801 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28804 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28805 call_eh_return is true.
28807 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28809 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28810 Initialize last_match_fntype_index.
28812 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28815 * tree-nvr.c: Include internal-fn.h.
28816 (pass_return_slot::execute): Ignore internal calls without
28819 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28820 Richard Biener <rguenther@suse.de>
28823 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28824 captures used multiple times, except for the last use.
28825 * generic-match-head.c: Include gimplify.h.
28827 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28829 PR tree-optimization/79390
28830 * target.h (struct noce_if_info): Declare.
28831 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28832 * target.def (noce_conversion_profitable_p): New target hook.
28833 * ifcvt.h (struct noce_if_info): New type, moved from ...
28834 * ifcvt.c (struct noce_if_info): ... here.
28835 (noce_conversion_profitable_p): Renamed to ...
28836 (default_noce_conversion_profitable_p): ... this. No longer
28838 (noce_try_store_flag_constants, noce_try_addcc,
28839 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28840 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28841 instead of noce_conversion_profitable_p.
28842 * config/i386/i386.c: Include ifcvt.h.
28843 (ix86_option_override_internal): Don't override
28844 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28845 (ix86_noce_conversion_profitable_p): New function.
28846 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28847 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28848 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28849 * doc/tm.texi: Regenerated.
28851 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28853 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28856 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28859 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28860 instructions for small multiply cores.
28862 2017-04-04 Jeff Law <law@redhat.com>
28864 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28866 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28867 that are not set by the loop over the elements.
28869 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28872 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28873 int mode, convert_modes it to mode as unsigned, otherwise use
28874 lowpart_subreg to mode rather than SImode.
28875 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28876 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28877 Use DImode instead of SImode for the shift count operand.
28878 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28881 2017-04-04 Richard Biener <rguenther@suse.de>
28883 PR middle-end/80281
28884 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28885 arithmetic done for the negate or the plus. Simplify.
28886 (A - (-B) -> A + B): Likewise.
28887 * fold-const.c (split_tree): Make sure to not negate pointers.
28889 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28891 PR rtl-optimization/60818
28892 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28893 a compare of comparisons with the thing compared if this results
28894 in a different machine mode.
28896 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28898 * alias.c (base_alias_check): Fix typo in comment.
28899 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28900 * cgraphunit.c (symbol_table::compile): Likewise.
28901 * collect2.c (maybe_run_lto_and_relink): Likewise.
28902 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28903 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28904 * config/avr/avr.c (avr_map_op_t): Likewise.
28905 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28906 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28907 * config/epiphany/epiphany.md (movcc): Likewise.
28908 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28909 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28911 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28912 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28913 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28914 * config/sh/sh.c (sh_rtx_costs): Likewise.
28915 * fold-const.c (fold_truth_andor): Likewise.
28916 * genautomata.c (collapse_flag): Likewise.
28917 * gengtype.h (struct type::u::s): Likewise.
28918 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28919 * input.c (FORMAT_AMOUNT): Likewise.
28920 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28921 (known_aggs_to_agg_replacement_list): Likewise.
28922 * ipa-inline-analysis.c: Likewise.
28923 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28924 * ipa-polymorphic-call.c
28925 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28926 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28927 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28929 * modulo-sched.c (apply_reg_moves): Likewise.
28930 * omp-expand.c (build_omp_regions_1): Likewise.
28931 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28932 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28933 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28934 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28935 * value-prof.c: Likewise.
28936 * var-tracking.c (val_reset): Likewise.
28938 2017-04-03 Richard Biener <rguenther@suse.de>
28940 PR tree-optimization/80275
28941 * fold-const.c (split_address_to_core_and_offset): Handle
28944 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28946 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28947 descriptors is at least equal to that of functions.
28949 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28951 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28953 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28956 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28957 (mov<IMOD4:mode>): New expander.
28958 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28960 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28962 PR rtl-optimization/79405
28963 * fwprop.c (propagations_left): New variable.
28964 (forward_propagate_into): Decrement it.
28965 (fwprop_init): Initialize it.
28966 (fw_prop): If the variable has reached zero, stop propagating.
28967 (fwprop_addr): Ditto.
28969 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28972 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28973 a FUNCTION_DECL, pass it as decl instead of origin to
28976 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28978 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28981 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28984 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28985 TARGET_VSX_SMALL_INTEGER.
28987 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28989 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28990 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28992 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28994 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28995 extraction from odd-numbered MSA register.
28997 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28999 PR middle-end/80173
29000 * expmed.c (store_bit_field_1): Don't attempt to create
29001 a word subreg out of hard registers wider than word if they
29002 have HARD_REGNO_NREGS of 1 for their mode.
29004 PR middle-end/80163
29005 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29006 conversions to integer types wider than word and pointer.
29009 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29010 (rtx_equal_for_cselib_p): Pass 0 to it.
29011 * cselib.c (cselib_hasher::equal): Likewise.
29012 (rtx_equal_for_cselib_1): Add depth argument. If depth
29013 is 128, don't look up VALUE locs and punt. Increment
29014 depth in recursive calls when walking VALUE locs.
29016 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29018 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29019 (make_gcov_file_name): Use the canonical path name for generating
29021 (read_line): Fix handling of files with ascii null bytes.
29023 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29025 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29026 to initialise a vector register instead
29027 of using a const_int.
29029 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29031 PR translation/80189
29032 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29033 diagnostic messages.
29035 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29038 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29039 (dfp_diex_<mode>): Update mode of operand 1.
29040 * doc/extend.texi (dxex, dxexq): Document change to return type.
29041 (diex, diexq): Document change to argument type.
29043 2017-03-30 Martin Jambor <mjambor@suse.cz>
29046 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29047 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29048 it reflects the signature changes performed at the callee side.
29049 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29050 to cgraph_build_function_type_skip_args.
29051 (build_function_decl_skip_args): Adjust call to the above function.
29053 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29056 * config/i386/sse.md
29057 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29058 register as dest whenever it is a MEM not rtx_equal_p to the
29059 corresponding dup operand, and when forcing into reg move the
29060 reg into the memory afterwards.
29061 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29062 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29063 for the force_reg mode.
29064 (avx512vl_vextractf128<mode>): Use register as dest either
29065 always when a MEM, or when it is a MEM not rtx_equal_p to the
29066 corresponding dup operand, or even not when it is a CONST_VECTOR
29067 depending on the mode and lo vs. hi.
29068 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29070 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29071 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29072 Likewise. Require that operands[2] is even.
29073 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29074 Remove extraneous parens. Require that operands[2] is a multiple
29076 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29077 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29078 disallow memory then.
29080 2017-03-30 Richard Biener <rguenther@suse.de>
29082 PR tree-optimization/77498
29083 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29084 to non-constants over backedges.
29086 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29088 PR rtl-optimization/80233
29089 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29090 as last_combined_insn. Do not test for BARRIER_P separately.
29092 2017-03-29 Andreas Schwab <schwab@suse.de>
29095 * calls.c (prepare_call_address): Convert funexp to Pmode before
29096 copying to temp reg.
29098 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29100 PR tree-optimization/80158
29101 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29102 Handle possible future case of more than one alternate
29104 (replace_rhs_if_not_dup): Likewise.
29105 (replace_one_candidate): Likewise.
29107 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29109 PR rtl-optimization/80193
29110 * ira.c (ira): Do not check allocation for LRA.
29112 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29114 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29115 (nvptx_output_simt_exit): Declare.
29116 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29117 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29118 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29119 (nvptx_declare_function_name): ...here. Emit declaration of local
29120 memory space buffer for omp_simt_enter insn.
29121 (nvptx_output_unisimt_switch): New.
29122 (nvptx_output_softstack_switch): New.
29123 (nvptx_output_simt_enter): New.
29124 (nvptx_output_simt_exit): New.
29125 * config/nvptx/nvptx.h (struct machine_function): New fields
29126 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29127 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29128 (UNSPECV_SIMT_EXIT): Ditto.
29129 (omp_simt_enter_insn): New insn.
29130 (omp_simt_enter): New expansion.
29131 (omp_simt_exit): New insn.
29132 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29134 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29135 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29136 (expand_GOMP_SIMT_EXIT): New.
29137 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29138 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29139 (GOMP_SIMT_EXIT): Ditto.
29140 * target-insns.def (omp_simt_enter): New insn.
29141 (omp_simt_exit): Ditto.
29142 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29144 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29145 (lower_rec_input_clauses): Likewise.
29146 (lower_lastprivate_clauses): Handle SIMT privatization.
29148 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29149 (ompdevlow_adjust_simt_enter): New.
29150 (find_simtpriv_var_op): New.
29151 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29152 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29154 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29155 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29156 (copy_decl_for_dup_finish): Ditto.
29158 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29160 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29163 * config/i386/i386.c (ix86_option_override_internal): Always
29164 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29166 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29168 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29170 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29172 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29173 mark new edge's irreducible flag accordign to it.
29174 (vect_do_peeling): Check loop preheader edge's irreducible flag
29175 and pass it to function slpeel_add_loop_guard.
29177 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29179 PR tree-optimization/80218
29180 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29181 Update block frequencies and counts.
29183 2017-03-28 Richard Biener <rguenther@suse.de>
29185 PR tree-optimization/78644
29186 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29187 of a simplification result we may not use it at all.
29189 2017-03-28 Richard Biener <rguenther@suse.de>
29192 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29193 without arguments, generate default definition of a SSA name.
29195 2017-03-28 Richard Biener <rguenther@suse.de>
29197 PR middle-end/80222
29198 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29199 TYPE_REF_CAN_ALIAS_ALL references.
29200 * fold-const.c (fold_indirect_ref_1): Likewise.
29202 2017-03-28 Martin Liska <mliska@suse.cz>
29205 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29206 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29208 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29209 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29211 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29212 (EXTRA_SPECS): Define.
29213 (SUBTARGET_EXTRA_SPECS): Likewise.
29214 (SUBTARGET_CPP_SPEC): Likewise.
29215 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29216 SUBTARGET_EXTRA_SPECS.
29217 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29219 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29221 * config/arc/simdext.md (vst64_insn): Update pattern.
29222 (vld32wh_insn): Likewise.
29223 (vld32wl_insn): Likewise.
29224 (vld64_insn): Likewise.
29225 (vld32_insn): Likewise.
29227 2017-03-28 Marek Polacek <polacek@redhat.com>
29230 * fold-const.c (fold_comparison): Use protected_set_expr_location
29231 instead of SET_EXPR_LOCATION.
29233 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29235 * tree.c (add_expr): Avoid name lookup warning.
29237 2017-03-27 Jeff Law <law@redhat.com>
29239 PR tree-optimization/80216
29240 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29241 function name. Limit recursion depth.
29242 (record_temporary_equivalences): Corresponding changes.
29244 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29246 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29249 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29252 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29254 * cfgcleanup.c (reg_note_cfa_p): New array.
29255 (insns_have_identical_cfa_notes): New function.
29256 (old_insns_match_p): Don't cross-jump in between /f
29257 and non-/f instructions. If both i1 and i2 are frame related,
29258 verify all CFA notes, their order and content.
29260 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29263 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29264 HImode and SImode with zero extend to DImode to one insn.
29265 (bswap<mode>2_extenddi): Likewise.
29266 (bswapsi2_extenddi): Likewise.
29267 (bswaphi2_extendsi): Likewise.
29268 (bswaphi2): Combine bswap HImode and SImode into one insn.
29269 Separate memory insns from swapping register.
29270 (bswapsi2): Likewise.
29271 (bswap<mode>2): Likewise.
29272 (bswaphi2_internal): Delete, no longer used.
29273 (bswapsi2_internal): Likewise.
29274 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
29275 store, and gpr<-gpr swap insns.
29276 (bswap<mode>2_store): Likewise.
29277 (bswaphi2_reg): Register only splitter, combine with the splitter.
29278 (bswaphi2 splitter): Likewise.
29279 (bswapsi2_reg): Likewise.
29280 (bswapsi2 splitter): Likewise.
29281 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
29282 the insns into load, store, and register/register insns.
29283 (bswapdi2_ldbrx): Likewise.
29284 (bswapdi2_load): Likewise.
29285 (bswapdi2_store): Likewise.
29286 (bswapdi2_reg): Likewise.
29288 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
29290 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
29291 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
29293 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
29296 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
29298 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29299 special handling for target option conflicts between dform
29300 options (-mpower9-dform, -mpower9-dform-vector,
29301 -mpower9-dform-scalar) and -mno-direct-move.
29303 2017-03-27 Richard Biener <rguenther@suse.de>
29305 PR tree-optimization/80181
29306 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29308 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29310 * config/arc/predicates.md (move_double_src_operand): Replace the
29311 call to move_double_src_operand with a call to address_operand.
29313 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29315 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29316 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29317 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29319 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29321 * config/arc/predicates.md (long_immediate_loadstore_operand):
29322 Consider scaled addresses cases.
29324 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29326 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29327 restored when in interrupt.
29328 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29329 doesn't have delay slot.
29331 2017-03-27 Richard Biener <rguenther@suse.de>
29334 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29335 inlined thunk clones.
29337 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29340 * asan.c (instrument_derefs): Copy over last operand from
29341 original COMPONENT_REF to the new COMPONENT_REF with
29342 DECL_BIT_FIELD_REPRESENTATIVE.
29343 * ubsan.c (instrument_object_size): Likewise.
29345 2017-03-27 Richard Biener <rguenther@suse.de>
29347 PR tree-optimization/80170
29348 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29349 sure DR/SCEV didnt fold in constants we do not see when looking
29350 at the reference base alignment.
29352 2017-03-27 Richard Biener <rguenther@suse.de>
29354 PR middle-end/80171
29355 * gimple-fold.c (fold_ctor_reference): Properly guard against
29356 NULL return value from canonicalize_constructor_val.
29358 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29361 * config/i386/i386.c (ix86_expand_builtin)
29362 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29363 flags reg setting and flags reg using instructions.
29364 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29365 clobbering instructions to zero extend op2.
29367 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29369 * doc/install.texi (Configuration) <--with-aix-soname>:
29370 Update link to AIX ld.
29372 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29374 PR rtl-optimization/80160
29375 PR rtl-optimization/80159
29376 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29377 reg_alternate_class into account.
29379 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29382 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29383 to consider in curr_insn_transform.
29385 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29387 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29388 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29389 and emit_mode_inner.
29391 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29393 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29394 argument to the overloaded builtin variants. Use the new flag to
29395 deprecate certain builtin variants.
29396 * config/s390/s390-builtin-types.def: Add new builtin types.
29397 * config/s390/s390-builtins.h: Support new flags field for
29398 overloaded builtins.
29399 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29400 (s390_macro_to_expand): Enable vector float data type.
29401 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29402 builtins by incrementing the __VEC__ version number.
29403 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29405 (s390_resolve_overloaded_builtin): Emit error messages depending
29406 on the builtin flags.
29407 * config/s390/s390.c (s390_expand_builtin): Support additional
29408 flags argument. Change error message to match the messages
29409 emitted in s390-c.c.
29410 * config/s390/s390.md: New UNSPEC_* constants.
29411 (op_type): Add new instruction types.
29412 * config/s390/vecintrin.h: Add new builtins and test data class
29414 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29415 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29416 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29417 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29418 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29419 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29421 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29422 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29423 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29424 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29426 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29427 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29428 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29429 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29431 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29432 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29433 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29434 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29435 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29436 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29437 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29439 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29440 ("vec_scatter_element<V_HW_4:mode>_DI")
29441 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29442 ("vec_fpint<mode>", "vflls")
29443 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29444 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29445 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29446 ("*vec_cmphe<mode>_cc"): ... these.
29448 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29449 mode constant instead of magic value.
29451 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29453 * config/s390/s390.c (s390_expand_vec_compare): Support other
29454 vector floating point modes than just V2DF.
29455 (s390_expand_vcond): Likewise.
29456 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29457 (s390_cannot_change_mode_class): Prevent mode changes between TF
29458 and V1TF in vector registers.
29459 * config/s390/s390.md (DF, SF): New mode attributes.
29460 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29461 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29462 SFmode support for VRs.
29463 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29465 (VFT, VF_HW): New mode iterators.
29466 (vw, sdx): New mode attributes.
29467 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29468 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29469 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29470 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29471 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29472 also the new vector floating point modes. Renaming to ...
29474 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29475 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29476 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29477 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29478 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29479 ("vec_unordered<mode>"): ... these.
29481 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29482 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29483 ("*vec_extendv2df"): New insn definitions.
29485 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29487 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29488 ("mulditi3_2", "*muldi3_sign"): New patterns.
29489 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29490 rename the pattern definition.
29492 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29494 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29496 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29500 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29501 instruction if possible.
29502 * config/s390/vector.md (vec_halfnumelts): New mode
29504 ("*vec_vllezlf<mode>"): New pattern.
29506 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29508 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29509 ("popcountv4si2", "popcountv2di2"): Rename to ...
29510 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29511 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29513 ("popcount<mode>2_vxe"): New pattern.
29515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29517 * common/config/s390/s390-common.c (processor_flags_table): Add
29519 * config.gcc: Add arch12.
29520 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29521 Default to arch12 for unknown CPU model numbers.
29522 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29523 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29524 PROCESSOR_max sanity check.
29525 * config/s390/s390-opts.h (enum processor_type): Add
29527 * config/s390/s390.c (processor_table): Add arch12.
29528 (s390_expand_builtin): Add check for B_VXE flag.
29529 (s390_issue_rate): Add PROCESSOR_ARCH12.
29530 (s390_get_sched_attrmask): Likewise.
29531 (s390_get_unit_mask): Likewise.
29532 (s390_sched_score): Enable z13 scheduling for arch12.
29533 (s390_sched_reorder): Likewise.
29534 (s390_sched_variable_issue): Likewise.
29535 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29537 (s390_tune_attr): Use z13 scheduling also for arch12.
29538 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29539 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29540 (TARGET_VXE_P): New macros.
29541 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29542 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29543 * config/s390/s390.opt: Add arch12 as processor_type.
29545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29547 * config/s390/s390.md
29548 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29549 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29550 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29552 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29553 Rename expanders to ...
29555 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29556 ("fixuns_truncdddi2_emu"): ... these.
29558 ("fixuns_trunc<mode>si2_emu"): New expander.
29560 ("*fixuns_truncdfdi2_z13"): Rename to ...
29561 ("*fixuns_truncdfdi2_vx"): ... this.
29563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29565 * config/s390/2964.md: Remove the single element vector compare
29566 instructions which are no longer used.
29567 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29569 (s390_canonicalize_comparison): Remove handling of DFmode
29571 (s390_expand_vec_compare_scalar): Remove function.
29572 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29573 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29575 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29577 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29579 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29581 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29582 will anyway by matched by mov<mode>_64dfp.
29584 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29586 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29587 vlef/vstef. Add missing operand to vleif.
29589 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29591 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29592 pair for all vector types with 64 bit elements.
29593 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29594 * config/s390/vector.md (V_HW_64): ... here.
29595 (V_128_NOSINGLE): New mode iterator.
29596 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29597 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29598 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29599 ("*vec_load_pairv2di"): Change to ...
29600 ("*vec_load_pair<mode>"): ... this one.
29602 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29604 * config/s390/constraints.md: Add comments.
29605 (jKK): Reject element sizes > 8 bytes.
29606 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29608 * config/s390/s390.md: Add the s_operand checks formerly in
29609 s390_split_ok_p to various splitters where they are still
29611 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29612 for 128 bit vectors. Plus two splitters.
29614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29616 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29619 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29622 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29623 error if the boundary argument is not constant.
29625 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29627 PR rtl-optimization/80112
29628 * loop-doloop.c (doloop_condition_get): Don't check condition
29629 if cmp isn't SET with IF_THEN_ELSE src.
29631 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29633 PR tree-optimization/80158
29634 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29635 replacing a candidate statement, also replace it for the
29636 candidate's alternate interpretation.
29637 (replace_rhs_if_not_dup): Likewise.
29638 (replace_one_candidate): Likewise.
29640 2017-03-24 Richard Biener <rguenther@suse.de>
29642 PR tree-optimization/80167
29643 * graphite-isl-ast-to-gimple.c
29644 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29646 (translate_isl_ast_to_gimple::get_rename): Likewise.
29648 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29651 handling of certain combinations of target options, including the
29652 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29653 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29655 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29658 * config/arm/arm.md (*load_multiple): Add reload_completed to
29659 matching condition.
29661 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29662 Richard Biener <rguenth@suse.de>
29664 PR tree-optimization/79908
29665 PR tree-optimization/80136
29666 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29667 been cast away, gimplify_and_add suffices.
29669 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29671 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29673 2017-03-23 Richard Biener <rguenther@suse.de>
29675 PR tree-optimization/80032
29676 * gimplify.c (gimple_push_cleanup): Forced unconditional
29677 cleanups still have to go to the conditional_cleanups
29680 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29682 PR tree-optimization/80072
29683 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29685 (next_operand_entry_id): Change type to unsigned int.
29686 (sort_by_operand_rank): Make sure to return the right return value
29687 even if unsigned fields are bigger than INT_MAX.
29688 (struct oecount): Change cnt and id type to unsigned int.
29689 (oecount_hasher::equal): Formatting fix.
29690 (oecount_cmp): Make sure to return the right return value
29691 even if unsigned fields are bigger than INT_MAX.
29692 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29695 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29696 TREE_READONLY on result if writing it more than once.
29699 * doc/invoke.texi (-fsanitize=thread): Document that with
29700 -fnon-call-exceptions atomics are not able to throw
29704 * tsan.c: Include tree-eh.h.
29705 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29706 maybe_clean_or_replace_eh_stmt where needed.
29707 (instrument_memory_accesses): Add cfg_changed argument.
29708 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29709 if it returned true.
29710 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29712 PR rtl-optimization/63191
29713 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29714 wrapper function, moved the whole old content into ...
29715 (ix86_delegitimize_address_1): ... this. New inline function.
29716 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29717 true as last argument instead of ix86_delegitimize_address.
29719 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29721 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29722 cortexa57_branch_cost.
29724 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29726 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29728 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29731 * doc/md.texi (Constraints): Document wA constraint.
29732 * config/rs6000/constraints.md (wA): New.
29733 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29734 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29735 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29736 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29738 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29741 * gimplify.c (is_oacc_declared): New function.
29742 (oacc_default_clause): Use it to set default flags for acc declared
29743 variables inside parallel regions.
29744 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29745 declared variables.
29746 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29747 declare attribute to any decl as necessary.
29749 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29752 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29753 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29754 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29755 (arm_arch_lpae): This.
29756 * config/arm/arm.c (arm_arch7ve): Rename into ...
29757 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29758 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29761 2017-03-22 Martin Liska <mliska@suse.cz>
29764 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29765 error message instead of an ICE.
29767 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29769 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29771 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29773 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29775 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29778 2017-03-21 Martin Sebor <msebor@redhat.com>
29780 * doc/extend.texi: Use "cannot" instead of "can't."
29781 * doc/hostconfig.texi: Same.
29782 * doc/install.texi: Same.
29783 * doc/invoke.texi: Same.
29784 * doc/loop.texi: Same.
29785 * doc/md.texi: Same.
29786 * doc/objc.texi: Same.
29787 * doc/rtl.texi: Same.
29788 * doc/tm.texi: Same.
29789 * doc/tm.texi.in: Same.
29790 * doc/trouble.texi: Same.
29792 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29795 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29796 (collect_checksum_attributes): Set it.
29797 (die_checksum_ordered): Use it.
29799 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29801 PR tree-optimization/79908
29802 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29803 change: For a VA_ARG whose LHS has been cast away, use
29804 force_gimple_operand to construct the side effects.
29806 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29808 PR translation/80001
29809 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29810 more amenable to translation.
29811 (oacc_loop_auto_partitions): Likewise.
29813 2017-03-21 Marek Polacek <polacek@redhat.com>
29814 Martin Sebor <msebor@redhat.com>
29816 PR tree-optimization/80109
29817 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29818 on INTEGRAL_TYPE_P.
29820 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29821 Segher Boessenkool <segher@kernel.crashing.org>
29824 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29825 check reg_used_between_p between insn and one of succ or succ2
29826 depending on if succ is artificial insn not inserted into insn
29829 2017-03-21 Martin Liska <mliska@suse.cz>
29831 PR gcov-profile/80081
29832 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29833 * doc/gcc.texi: Include gcov-dump stuff.
29834 * doc/gcov-dump.texi: New file.
29836 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29838 PR rtl-optimization/79150
29839 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29840 conditional jump, if the jump is the last insn of the loop.
29842 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29843 Richard Biener <rguenth@suse.de>
29845 PR tree-optimization/79908
29846 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29847 been cast away, use force_gimple_operand to construct the side
29850 2017-03-21 Martin Liska <mliska@suse.cz>
29852 PR libfortran/79956
29853 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29856 2017-03-21 Brad Spengler <spender@grsecurity.net>
29859 * plugin.c (htab_hash_plugin): New function.
29860 (add_new_plugin): Use it and adjust.
29861 (parse_plugin_arg_opt): Adjust.
29862 (init_one_plugin): Likewise.
29864 2017-03-21 Richard Biener <rguenther@suse.de>
29866 PR tree-optimization/80032
29867 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29868 if set force the cleanup to happen unconditionally.
29869 (gimplify_target_expr): Push inserted clobbers with force_uncond
29870 to avoid them being removed by control-dependent DCE.
29872 2017-03-21 Richard Biener <rguenther@suse.de>
29874 PR tree-optimization/80122
29875 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29876 va_arg_pack_len when the inlined call stmt requires pack
29878 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29880 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29883 * tsan.c (instrument_builtin_call): If the memory model argument
29884 is not a constant, assume it is valid.
29887 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29890 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29892 PR rtl-optimization/79910
29893 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29894 if its dest is used by an insn before I2 (other than the combined
29895 insns themselves, which are properly handled already).
29897 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29900 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29902 * combine.c (record_used_regs): New static function.
29903 (try_combine): Handle situations where there is an additional
29904 instruction between I2 and I3 which needs to have a LOG_LINK
29908 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29910 * combine.c (try_combine): Delete redundant i1 test. Call
29911 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29913 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29916 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29917 alternatives 13/14.
29919 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29921 PR tree-optimization/80054
29922 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29923 the optimization if a PHI or any of its arguments is not dominated
29924 by the candidate's basis. Use gphi* rather than gimple* as
29926 (replace_profitable_candidates): Clean up a gimple* variable that
29927 should be a gphi* variable.
29929 2017-03-20 Martin Sebor <msebor@redhat.com>
29932 * doc/extend.texi (attribute constructor): Document present limitation.
29934 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29937 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29938 __POWER9_VECTOR__ #ifdef control, change template definition to
29939 use Power9-specific built-in function.
29940 (vec_any_eq): Likewise.
29941 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29942 to control outcomes from this test.
29943 (vector_ae_<mode>p): For VEC_F modes, likewise.
29945 2017-03-20 Ian Lance Taylor <iant@google.com>
29947 * config/i386/i386.c (ix86_function_regparm): Save an extra
29948 register for -fsplit-stack with DECL_STATIC_CHAIN.
29950 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29953 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29954 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29956 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29958 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29960 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29963 2017-03-20 Marek Polacek <polacek@redhat.com>
29966 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29968 2017-03-20 Richard Biener <rguenther@suse.de>
29970 PR tree-optimization/80113
29971 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29972 allocate extra SSA name for PHI def.
29973 (add_close_phis_to_outer_loops): Likewise.
29974 (add_close_phis_to_merge_points): Likewise.
29975 (copy_loop_close_phi_args): Likewise.
29976 (copy_cond_phi_nodes): Likewise.
29978 2017-03-20 Martin Liska <mliska@suse.cz>
29980 PR middle-end/79753
29981 * tree-chkp.c (chkp_build_returned_bound): Do not build
29982 returned bounds for a LHS that's not a BOUNDED_P type.
29984 2017-03-20 Martin Liska <mliska@suse.cz>
29988 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29989 COMPLEX_CST and VECTOR_CST.
29991 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29994 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29995 target operand. A new splitter adds the clobber statement in case
29996 the target operand is dead anyway.
29998 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30000 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30001 to age-old versions of binutils and glibc.
30003 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30005 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30007 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30009 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30011 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30013 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30014 requirement for binutils 2.13.
30016 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30018 * combine.c (try_combine): Delete redundant i1 test. Call
30019 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30021 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30023 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30024 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30026 <riscv64-*-elf>: Re-arrange section
30027 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30028 <riscv32-*-linux>: Likewise.
30029 <riscv64-*-elf>: Likewise
30030 <riscv64-*-linux>: Likewise.
30032 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30035 * aarch64.opt(verbose-cost-dump): Fix typo.
30037 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30040 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30041 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30043 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30045 * reload.c (find_reloads): When reloading a nonoffsettable address,
30046 use RELOAD_OTHER for it and its address reloads.
30048 PR rtl-optimization/79910
30049 * combine.c (record_used_regs): New static function.
30050 (try_combine): Handle situations where there is an additional
30051 instruction between I2 and I3 which needs to have a LOG_LINK
30054 2017-03-17 Jeff Law <law@redhat.com>
30056 PR tree-optimization/71437
30057 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30058 conditional in the hash table first.
30059 (vrp_dom_walker::before_dom_children): Extract condition from
30060 ASSERT_EXPR. Record condition, its inverion and any implied
30061 conditions as well.
30063 2017-03-17 Marek Polacek <polacek@redhat.com>
30064 Markus Trippelsdorf <markus@trippelsdorf.de>
30066 PR tree-optimization/80079
30067 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30070 2017-03-17 Richard Biener <rguenther@suse.de>
30072 PR middle-end/80075
30073 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30074 Properly verify the LHS before the RHS possibly claims to be
30076 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30079 2017-03-17 Martin Jambor <mjambor@suse.cz>
30081 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30082 (List of -O2 options): Likewise.
30083 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30086 2017-03-17 Tom de Vries <tom@codesourcery.com>
30088 * gcov-dump.c (print_usage): Print bug_report_url.
30090 2017-03-17 Richard Biener <rguenther@suse.de>
30092 PR middle-end/80050
30093 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30094 (parser::peek): Likewise.
30096 2017-03-17 Richard Biener <rguenther@suse.de>
30098 PR tree-optimization/80048
30099 * sese.c (free_sese_info): Properly release rename_map and
30100 copied_bb_map elements.
30102 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30104 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30105 Add linked-list forward and backlinks. Insert on
30106 construction, remove on destruction.
30107 (class pass_store_merging): Add m_stores_head field.
30108 (pass_store_merging::terminate_and_process_all_chains):
30109 Iterate over m_stores_head list.
30110 (pass_store_merging::terminate_all_aliasing_chains):
30112 (pass_store_merging::execute): Check for debug stmts first.
30113 Push new chains onto the m_stores_head stack.
30115 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30118 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30119 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30120 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30122 2017-03-16 Jeff Law <law@redhat.com>
30124 PR tree-optimization/71437
30125 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30126 member function. Implementation moved into after_dom_children
30127 member function and into the threader's thread_outgoing_edges
30129 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30130 some code into new thread_outgoing_edges.
30131 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30132 definition. Simplify marker handling (do it here). Assume we always
30133 have the available expression and the const/copies tables.
30134 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30136 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30137 * tree-vrp.c (equiv_stack): No longer file scoped.
30138 (vrp_dom_walker): New class.
30139 (vrp_dom_walker::before_dom_children): New member function.
30140 (vrp_dom_walker::after_dom_children): Likewise.
30141 (identify_jump_threads): Setup domwalker. Use it rather than
30142 walking edges in a random order by hand. Simplify setup/finalization.
30143 (finalize_jump_threads): Remove.
30144 (vrp_finalize): Do not call identify_jump_threads here.
30145 (execute_vrp): Do it here instead and call thread_through_all_blocks
30148 PR tree-optimization/71437
30149 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30151 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30153 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30154 (dom_opt_dom_walker::thread_across_edge): Remove
30155 handle_dominating_asserts argument. All callers changed.
30156 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30157 changes. Remove calls to lhs_of_dominating_assert. Other
30158 uses of handle_dominating_asserts turn into unconditional code
30159 (simplify_control_stmt_condition_1): Likewise.
30160 (simplify_control_stmt_condition): Likewise.
30161 (thread_through_normal_block, thread_across_edge): Likewise.
30162 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30163 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30164 object if it is not an SSA_NAME.
30165 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30166 before calling into the VRP specific simplifiers.
30167 (identify_jump_threads): Remove handle_dominating_asserts
30170 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30173 * tree-diagnostic.c (default_tree_printer): No longer static.
30174 * tree-diagnostic.h (default_tree_printer): New prototype.
30176 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30178 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30179 Change ins into fmov.
30181 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30183 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30184 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30185 Use h_con constraint for operand 1.
30186 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30187 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30189 2017-03-15 Jeff Law <law@redhat.com>
30191 PR tree-optimization/71437
30192 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30193 (record_temporary_equivalences): Use it.
30195 PR tree-optimization/71437
30196 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30197 tree-ssa-scopedtables.
30198 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30199 (record_conditions, record_cond, vuse_eq): Likewise.
30200 (record_edge_info): Adjust to API tweak of record_conditions.
30201 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30202 (record_temporary_equivalences, optimize_stmt): Likewise.
30203 (eliminate_redundant_computations): Likewise.
30204 (record_equivalences_from_stmt): Likewise.
30205 * tree-ssa-scopedtables.c: Include options.h and params.h.
30206 (vuse_eq): New function, moved from tree-ssa-dom.c
30207 (build_and_record_new_cond): Likewise.
30208 (record_conditions): Likewise. Accept vector of conditions rather
30209 than edge_equivalence structure for first argument.
30210 for the first argument.
30211 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30212 from tree-ssa-dom.c.
30213 (avail_exprs_stack::record_cond): Likewise.
30214 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30215 from tree-ssa-dom.c.
30216 (avail_exprs_stack): Add new member functions lookup_avail_expr
30218 (record_conditions): Declare.
30220 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30223 * lra-constraints.c (process_alt_operands): Increase reject for
30224 reloading an input/output operand.
30226 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30229 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30230 insns to convert from signed/unsigned char/short to IEEE 128-bit
30232 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30234 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30237 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30238 subreg of inner mode for values already in registers.
30240 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30242 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30243 iteration reg is used after the loop.
30245 2017-03-14 Martin Sebor <msebor@redhat.com>
30247 PR tree-optimization/79800
30248 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30249 precision in negative-positive range.
30250 (format_floating): Call non-const overload with adjusted precision.
30252 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30255 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30258 2017-03-14 Martin Sebor <msebor@redhat.com>
30260 PR middle-end/80020
30261 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30262 * builtins.def (aligned_alloc): Use it.
30265 * Makefile.in (GTFILES): Add calls.c.
30266 * calls.c: Include "gt-calls.h".
30268 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
30270 PR rtl-optimization/79728
30271 * regs.h (struct target_regs): New field
30272 x_contains_allocatable_regs_of_mode.
30273 (contains_allocatable_regs_of_mode): New macro.
30274 * reginfo.c (init_reg_sets_1): Initialize it, and change
30275 contains_reg_of_mode so it includes global regs as well.
30276 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
30277 rather than contains_regs_of_mode.
30279 2017-03-14 Martin Liska <mliska@suse.cz>
30281 * doc/invoke.texi: Document options that can't be combined with
30282 -fcheck-pointer-bounds.
30284 2017-03-14 Martin Liska <mliska@suse.cz>
30286 PR middle-end/79831
30287 * doc/invoke.texi (-Wchkp): Document the option.
30289 2017-03-14 Martin Liska <mliska@suse.cz>
30291 * Makefile.in: Install gcov-dump.
30293 2017-03-14 Martin Liska <mliska@suse.cz>
30295 * multiple_target.c (expand_target_clones): Bail out for
30296 an invalid attribute.
30298 2017-03-14 Richard Biener <rguenther@suse.de>
30300 * alias.c (struct alias_set_entry): Pack properly.
30301 * cfgloop.h (struct loop): Likewise.
30302 * cse.c (struct set): Likewise.
30303 * ipa-utils.c (struct searchc_env): Likewise.
30304 * loop-invariant.c (struct invariant): Likewise.
30305 * lra-remat.c (struct cand): Likewise.
30306 * recog.c (struct change_t): Likewise.
30307 * rtl.h (struct address_info): Likewise.
30308 * symbol-summary.h (function_summary): Likewise.
30309 * tree-loop-distribution.c (struct partition): Likewise.
30310 * tree-object-size.c (struct object_size_info): Likewise.
30311 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30312 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30313 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30314 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30315 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30316 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30317 (struct _stmt_vec_info): Likewise.
30319 2017-03-14 Martin Liska <mliska@suse.cz>
30322 * multiple_target.c (create_dispatcher_calls): Check that
30323 a target can create a function dispatcher.
30325 2017-03-14 Martin Liska <mliska@suse.cz>
30328 * multiple_target.c (expand_target_clones): Drop local.local
30329 flag for default implementation.
30331 2017-03-14 Richard Biener <rguenther@suse.de>
30333 PR tree-optimization/80030
30334 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30336 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30338 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30339 gcc_fallthrough() instead of __attribute__((fallthrough));
30341 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30343 * doc/gcc.texi: Remove "up" link to (DIR).
30344 * doc/gccint.texi: Ditto.
30346 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30348 * doc/install.texi (Specific) <avr>: Remove reference to
30351 2017-03-13 Jeff Law <law@redhat.com>
30353 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30354 attribute rather than comments.
30356 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30357 match_scratch operand is highest.
30359 2017-03-13 Martin Liska <mliska@suse.cz>
30361 PR middle-end/78339
30362 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30363 is a CHKP clone, use original declaration.
30365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30367 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30368 (arc_conditional_register_usage): Use a different allocation order
30369 when optimizing for size.
30370 * common/config/arc/arc-common.c (arc_option_optimization_table):
30371 Section anchors default on when optimizing for size.
30373 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30375 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30377 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30379 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30380 * config/arc/arc.md (cpu_facility): Add cd variant.
30381 (*movqi_insn): Add code density variant.
30382 (*movhi_insn): Likewise.
30383 (*movqi_insn): Likewise.
30384 (*addsi3_mixed): Likewise.
30385 (subsi3_insn): Likewise.
30387 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30389 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30391 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30393 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30394 expressions with MINUS and UNARY ops.
30396 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30399 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30401 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30402 between vec_select and vector argument.
30403 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30404 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30405 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30406 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30407 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30408 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30410 2017-03-13 Richard Biener <rguenther@suse.de>
30413 * params.def (vect-max-peeling-for-alignment): Fix typo.
30415 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30417 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30418 issue that only occurred with binutils below 2.18.
30420 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30422 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30423 refer to binutils 2.11/2.12 minimum.
30425 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30427 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30428 ftp.kernel.org and simplify binutils requirement.
30430 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30432 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30434 (Optimize Options): Ditto. Also remove redundancy.
30436 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30438 PR translation/79848
30439 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30441 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30442 to G_ to avoid double translation.
30444 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30446 PR translation/79923
30447 * auto-profile.c (get_combined_location): Convert leading
30448 character of diagnostics to lower case and remove trailing period.
30449 (read_profile): Likewise for various diagnostics.
30450 * config/arm/arm.c (arm_option_override): Remove trailing period
30451 from various diagnostics.
30452 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30453 (msp430_expand_delay_cycles): Likewise.
30455 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30458 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30459 full command-line argument, rather than just "str".
30460 (aarch64_validate_march): Likewise.
30461 (aarch64_validate_mtune): Likewise.
30463 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30465 PR rtl-optimization/78911
30466 * lra-assigns.c (must_not_spill_p): New function.
30467 (spill_for): Use it.
30469 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30471 PR tree-optimization/79981
30472 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30473 ATOMIC_COMPARE_EXCHANGE ifn result.
30474 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30475 IFN_ATOMIC_COMPARE_EXCHANGE.
30477 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30480 * opts.c (parse_sanitizer_options): Add missing question mark to
30481 "did you mean" message.
30483 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30485 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30487 (VMULEUH_UNS): Likewise.
30488 (VMULOUB_UNS): Likewise.
30489 (VMULOUH_UNS): Likewise.
30490 * config/rs6000/rs6000.c (builtin_function_type): Remove
30491 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30493 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30496 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30497 x with result of extra_parsing_for_operand_code_0.
30498 (function_reader::extra_parsing_for_operand_code_0): Convert
30499 return type from void to rtx, returning x. When reading
30500 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30501 larger size containing struct block_symbol.
30503 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30505 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30506 -mfloat128-hardware without -m64.
30508 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30511 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30512 entries to the case statement that marks unsigned arguments to
30513 overloaded functions.
30515 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30517 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30518 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30520 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30523 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30524 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30526 2017-03-10 Martin Liska <mliska@suse.cz>
30530 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30531 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30534 2017-03-10 Olivier Hainque <hainque@adacore.com>
30536 * tree-switch-conversion (array_value_type): Start by resetting
30537 candidate type to it's main variant.
30539 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30541 PR rtl-optimization/79909
30542 * combine.c (try_combine): Use simplify_replace_rtx on individual
30543 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30544 of the whole CALL_INSN_FUNCTION_USAGE.
30546 PR tree-optimization/79972
30547 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30548 get_range_info on SSA_NAMEs. Formatting fixes.
30550 2017-03-10 Richard Biener <rguenther@suse.de>
30551 Jakub Jelinek <jakub@redhat.com>
30553 PR tree-optimization/77975
30554 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30555 edge to be constant.
30556 (get_val_for): For constant x return it. Formatting fix.
30557 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30558 would use the same bases as the current one.
30560 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30562 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30563 instead of vec_select for V1TImode.
30564 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30566 (VSX_LE_128): Add V1TI to this mode iterator.
30567 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30568 (*vsx_le_perm_store_<mode>): Likewise.
30569 (pre-reload splitter for VSX stores): Likewise.
30570 (post-reload splitter for VSX stores): Likewise.
30571 (*vsx_xxpermdi2_le_<mode>): Likewise.
30572 (*vsx_lxvd2x2_le_<mode>): Likewise.
30573 (*vsx_stxvd2x2_le_<mode>): Likewise.
30575 2017-03-09 Michael Eager <eager@eagercon.com>
30577 Correct failures with --enable-checking=yes,rtl.
30579 * config/microblaze/microblaze.c (microblaze_expand_shift):
30580 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30581 test for const0_rtx.
30582 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30583 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30585 2017-03-09 Richard Biener <rguenther@suse.de>
30587 PR tree-optimization/79977
30588 * graphite-scop-detection.c (scop_detection::merge_sese):
30589 Handle the case of extra exits to blocks dominating the entry.
30591 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30593 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30596 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30598 PR rtl-optimization/79949
30599 * lra-constraints.c (process_alt_operands): Check memory when
30600 trying to predict a cycle. Print about the overall increase.
30602 2017-03-09 Richard Biener <rguenther@suse.de>
30604 PR middle-end/79971
30605 * gimple-expr.c (useless_type_conversion_p): Preserve
30606 TYPE_SATURATING for fixed-point types.
30608 2017-03-09 Richard Biener <rguenther@suse.de>
30611 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30612 alignment of BLKmode params.
30614 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30617 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30618 (VALL_NO_V2Q): Likewise.
30620 * config/aarch64/aarch64-simd.md
30621 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30623 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30624 VALL_NO_V2Q mode iterator.
30625 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30627 2017-03-09 Martin Liska <mliska@suse.cz>
30629 PR tree-optimization/79631
30630 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30631 tree_int_cst_sign_bit just for INTEGER constants.
30633 2017-03-09 Martin Liska <mliska@suse.cz>
30637 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30640 2017-03-09 Marek Polacek <polacek@redhat.com>
30643 * tree.c (inchash::add_expr): Handle TREE_VEC.
30645 2017-03-09 Martin Liska <mliska@suse.cz>
30648 (chkp_narrow_size_and_offset): New function.
30649 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30650 (void chkp_parse_bit_field_ref): New function.
30651 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30652 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30654 2017-03-09 Martin Liska <mliska@suse.cz>
30657 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30658 (chkp_find_bounds_1): Remove gcc_unreachable.
30660 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30663 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30664 BUILT_IN_SYNC*, determine the access type from the size suffix and
30665 always build a MEM_REF with that type. Handle forgotten
30666 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30669 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30670 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30671 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30672 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30673 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30674 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30675 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30676 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30677 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30678 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30679 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30680 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30681 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30682 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30683 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30684 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30685 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30686 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30687 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30688 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30689 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30690 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30691 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30692 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30693 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30694 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30695 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30696 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30697 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30698 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30699 definitions outside of __OPTIMIZE__ guarded section.
30702 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30703 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30704 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30705 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30708 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30710 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30711 ("vfenez<mode>"): Add missing constraints.
30713 2017-03-08 Martin Sebor <msebor@redhat.com>
30716 * config/nds32/nds32.c (nds32_option_override):
30717 Fix misspelled diagnostic.
30719 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30722 * gimplify.c (gimplify_omp_for): Replace index var in outer
30723 taskloop statement with an artificial variable and add
30724 OMP_CLAUSE_PRIVATE clause for it.
30726 2017-03-08 Richard Biener <rguenther@suse.de>
30728 PR tree-optimization/79955
30729 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30730 for accesses that are completely outside of the variable.
30732 2017-03-08 Andrew Haley <aph@redhat.com>
30734 PR tree-optimization/79943
30735 * tree-ssa-loop-split.c (compute_new_first_bound): When
30736 calculating the new upper bound, (END-BEG) should be added, not
30739 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30741 * config/avr/avr.md (setmemhi): Make sure match_dup
30742 operand number comes before match_scratch.
30744 2017-03-08 Richard Biener <rguenther@suse.de>
30746 PR tree-optimization/79920
30747 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30748 with ncopies == 1 to ...
30749 (vect_transform_slp_perm_load): ... here. Properly compute
30750 all element loads by iterating VF times over the group. Do
30751 not handle ncopies (computed in a broken way) in
30752 vect_create_mask_and_perm.
30754 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30757 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30758 is a uniform vector, use uniform_vector_p return value instead of
30759 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30761 2017-03-07 Marek Polacek <polacek@redhat.com>
30763 PR middle-end/79809
30764 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30765 (alloca_call_type): Likewise.
30767 2017-03-07 Martin Liska <mliska@suse.cz>
30769 * gcov.c (process_args): Put comment to correct location.
30771 2017-03-07 Martin Liska <mliska@suse.cz>
30773 PR middle-end/68270
30774 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30775 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30776 (chkp_narrow_bounds_for_field): Likewise.
30777 (chkp_parse_array_and_component_ref): Pass one more argument to
30780 2017-03-07 Richard Biener <rguenther@suse.de>
30782 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30785 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30787 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30788 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30790 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30793 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30794 to end of description.
30795 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30797 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30799 PR rtl-optimization/79901
30800 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30802 (*avx512f_<code><mode>3<mask_name>): ... this.
30803 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30804 iterator instead of VI8_AVX2_AVX512BW.
30806 PR rtl-optimization/79901
30807 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30808 min/max expander, expand it using expand_vec_cond_expr.
30811 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30814 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30817 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30818 to void * for PCH reasons.
30819 * dwarf2out.c (output_loc_operands, output_die): Cast
30820 v.val_vec.array to unsigned char *.
30822 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30825 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30828 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30830 PR rtl-optimization/79571
30831 * lra-constraints.c (process_alt_operands): Calculate static
30832 reject and subtract it from overall when only addresses will be
30835 2017-03-06 Julia Koval <julia.koval@intel.com>
30838 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30839 incoming stack boundary to 128 for 64-bit targets.
30841 2017-03-06 Richard Biener <rguenther@suse.de>
30843 PR tree-optimization/79894
30844 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30845 to NULL after folding it.
30847 2017-03-06 Richard Biener <rguenther@suse.de>
30849 PR tree-optimization/79824
30850 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30851 check disabling peeling for gaps.
30853 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30855 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30856 attributes): Document gettimeofday.
30858 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30860 * config/s390/s390.c (s390_option_override_internal): Set
30861 PARAM_MIN_VECT_LOOP_BOUND
30863 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30865 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30866 * config/s390/s390.md: Likewise.
30868 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30871 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30872 (<avx2_avx512>_perm<mode>): Rename to ...
30873 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30875 (<avx512>_perm<mode>_mask): Rename to ...
30876 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30878 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30879 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30880 instead of VI8F_256_512.
30881 (avx512f_perm<mode>): New define_expand.
30882 (avx512f_perm<mode>_mask): Likewise.
30883 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30884 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30886 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30888 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30889 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30891 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30893 2017-03-06 Martin Liska <mliska@suse.cz>
30896 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30897 when having a SSA NAME w/o VAR_DECL assigned to it.
30899 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30901 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30902 msa_dpsub_<su>_d): Fix MODE for vec_select.
30904 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30906 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30908 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30910 2017-03-06 Richard Biener <rguenther@suse.de>
30912 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30913 * plugin.c (register_plugin_info): Likewise.
30914 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30916 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30918 * config/i386/sse.md (sse_storehps, sse_storelps,
30919 avx_<castmode><avxsizesuffix>_<castmode>,
30920 avx512f_<castmode><avxsizesuffix>_<castmode>,
30921 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30922 in condition that at least one operand is not a MEM.
30924 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30926 PR middle-end/79805
30927 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30928 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30930 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30931 gimple_call_nothrow_p flag based on whether original builtin can throw.
30932 If it can, emit following stmts on the fallthrough edge.
30933 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30934 don't create new bb if inserting just debug stmts on the edge, try to
30935 insert them on the fallthru bb or just reset debug stmts.
30937 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30940 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30941 restore recog_data (including the operand rtxes inside it) around
30942 the call to get_insn_template.
30944 2017-03-03 Martin Sebor <msebor@redhat.com>
30946 PR tree-optimization/79699
30947 * context.c (context::~context): Free MPFR caches to avoid
30948 a memory leak on program exit.
30950 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30952 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30953 Use wide_int::ulow () instead of .elt (0).
30955 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30957 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30958 (*pushxf): Limit oF constraint to 32bit targets and add oC
30959 constraint for 64bit targets.
30960 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30961 (*pushdf): Change rmF constraint to rmC.
30963 2017-03-03 Martin Liska <mliska@suse.cz>
30965 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30966 Remove unused variable.
30968 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30971 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30972 is a memory operand, increase num_memory.
30973 (ix86_expand_args_builtin): Likewise.
30975 2017-03-03 Jan Hubicka <jh@suse.cz>
30978 * ipa-devirt.c (maybe_record_node): Properly handle
30979 __cxa_pure_virtual visibility.
30981 2017-03-03 Martin Liska <mliska@suse.cz>
30983 PR tree-optimization/79803
30984 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30986 (pass_loop_prefetch::execute): Disabled optimization if an
30987 assumption about L1 cache size is not met.
30989 2017-03-03 Martin Liska <mliska@suse.cz>
30991 PR rtl-optimization/79574
30992 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30993 (hash_scan_set): Likewise.
30994 (dump_hash_table): Likewise.
30995 (hoist_code): Likewise.
30997 2017-03-03 Richard Biener <rguenther@suse.de>
30999 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31000 in place of uhigh ().
31001 (fixed_convert_from_real): Likewise.
31003 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31006 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31008 2017-03-03 Richard Biener <rguenther@suse.de>
31010 PR middle-end/79818
31011 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31012 TYPE_OVERFLOW_UNDEFINED check.
31014 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31016 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31018 (vector_ae_<mode>_p): Likewise.
31019 (vector_nez_<mode>_p): Likewise.
31020 (vector_ne_v2di_p): Likewise.
31021 (vector_ae_v2di_p): Likewise.
31022 (vector_ne_<mode>_p): Likewise.
31023 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31025 (vsx_tsqrt<mode>2_fe): Likewise.
31027 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31030 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31032 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31034 PR rtl-optimization/79780
31035 * cprop.c (one_cprop_pass): When second and further conditional trap
31036 in a single basic block is turned into an unconditional trap, turn it
31037 into a deleted note to avoid RTL verification failures.
31039 2017-03-02 Richard Biener <rguenther@suse.de>
31041 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31043 2017-03-02 Richard Biener <rguenther@suse.de>
31045 PR tree-optimization/79345
31047 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31048 param and abort the walk, returning -1 if it is hit.
31049 (walk_aliased_vdefs): Take a limit param and pass it on.
31050 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31051 defaulting to 0 and return a signed int.
31052 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31053 (check_defs): New helper.
31054 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31055 about uninitialized memory.
31056 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31057 bogus uninitialized warning.
31058 (fixed_convert_from_real): Likewise.
31060 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31062 PR tree-optimization/66768
31063 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31064 iv_use if base object can't be determined.
31066 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31068 PR tree-optimization/79345
31069 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31070 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31071 (get_pattern_stats): Initialize it.
31072 * genemit.c (gen_expand): Verify match_scratch numbers come after
31073 match_operand/match_dup numbers.
31074 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31075 match_scratch numbers.
31076 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31078 * config/s390/s390.md (trunctdsd2): Likewise.
31080 2017-03-02 Richard Biener <rguenther@suse.de>
31082 * wide-int.h (wide_int_storage::operator=): Implement in terms
31085 2017-03-02 Richard Biener <rguenther@suse.de>
31087 PR tree-optimization/79777
31088 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31089 the to insert expression to sth existing.
31091 2017-03-01 Martin Sebor <msebor@redhat.com>
31093 PR middle-end/79692
31094 * gimple-ssa-sprintf.c
31095 (directive::known_width_and_precision): New function.
31096 (format_integer): Use it.
31097 (get_mpfr_format_length): Consider the full range of precision
31098 when computing %g output with the # flag. Set the likely byte
31099 count to 3 rather than 1 when precision is indeterminate.
31100 (format_floating): Correct the lower bound of precision.
31102 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31104 * doc/invoke.texi: Document default code model for 64-bit Linux.
31106 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31109 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31110 udiv rather than div since input pattern is unsigned.
31112 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31114 * config/i386/i386.c (print_reg): Warn for values of
31115 unsupported size in integer register.
31117 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31120 * config/rs6000/predicates.md (current_file_function_operand): Do
31121 not allow self calls to be local if the function is replaceable.
31123 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31126 * config/rs6000/altivec.h (vec_ctz and others): Change the
31127 preprocessor macro that controls conditional compilation from
31128 _ARCH_PWR9 to __POWER9_VECTOR__.
31129 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31130 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31131 control (instead of _ARCH_PWR9 control) so that template
31132 definition uses power9-specific function.
31133 (vec_any_eq): Likewise.
31134 (vec_all_ne): Change macro definition to use a power9-specific
31135 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31136 _ARCH_PWR9 control).
31137 (vec_any_eq) Likewise.
31138 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31139 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31140 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31141 support for xvcmpnedp instruction.
31142 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31143 macro expansion so that Power9 implementation of vec_all_ne does
31144 not use the AltiVec predicate framework.
31145 (VCMPNEH_P): Likewise.
31146 (VCMPNEW_P): Likewise.
31147 (VCMPNED_P): Likewise.
31148 (VCMPNEFP_P): Likewise.
31149 (VCMPNEDP_P): Likewise.
31150 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31151 implementation of vec_any_eq to not use AltiVec predicate
31153 (VCMPAEH_P): Likewise.
31154 (VCMPAEW_P): Likewise.
31155 (VCMPAED_P): Likewise.
31156 (VCMPAEFP_P): Likewise.
31157 (VCMPAEDP_P): Likewise.
31158 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31159 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31160 not use the AltiVec predicate framework.
31161 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31162 of vec_any_eq to not use AltiVec predicate framework.
31163 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31164 support for predefined __POWER9_VECTOR__ macro to indicate that
31165 Power9 instruction selection is enabled.
31166 (altivec_overloaded_builtins): Remove extraneous
31167 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31168 function argument types RS6000_BTI_bool_V16QI and
31169 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31170 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31171 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31172 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31173 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31174 Power9 for implementations of vec_cmpne. Change the signature for
31175 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31176 (representing vec_all_ne) to remove the previously described first
31177 argument of type RS6000_BTI_INTSI, as this was an artifact of
31178 reliance on the AltiVec predicate framework, which is no longer
31179 used in the implementation of these functions. Add
31180 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31181 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31182 since, unlike the AltiVec predicate framework implementation, we
31183 do not share function descriptors between vec_alle and vec_anyeq.
31184 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31185 set of modes that receive special treatment even when
31186 TARGET_P9_VECTOR is true. The special treatment emits code that
31187 does not depend on Power9 instructions.
31188 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31189 define_expand to not rely on AltiVec predicate framework.
31190 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31192 (vector_ne_v2di_p): Change this define_expand to not rely on
31193 AltiVec predicate framework.
31194 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31196 (vector_ne_<mode>_p): Change this define_expand to not rely on
31197 AltiVec predicate framework.
31198 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31200 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31201 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31202 define_insn pattern.
31203 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31204 define_insn pattern because the xvcmpne<VSs>. instruction is not
31206 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31207 instruction is not supported.
31209 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31211 * config/nvptx/nvptx.c: Include intl.h.
31213 2017-03-01 Martin Jambor <mjambor@suse.cz>
31216 * ipa-prop.h (ipa_bits): Removed field known.
31217 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31218 to pointers. Adjusted their comments to warn about their sharing.
31219 (ipcp_transformation_summary): Change bits to a vector of pointers.
31220 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31221 (ipa_get_ipa_bits_for_value): Declare.
31222 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31223 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31224 (ipa_bits_hash_table): Likewise.
31225 (ipa_vr_ggc_hash_traits): Likewise.
31226 (ipa_vr_hash_table): Likewise.
31227 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31228 being pointers and vr_known being removed.
31229 (ipa_set_jf_unknown): Likewise.
31230 (ipa_get_ipa_bits_for_value): New function.
31231 (ipa_set_jfunc_bits): Likewise.
31232 (ipa_get_value_range): New overloaded functions.
31233 (ipa_set_jfunc_vr): Likewise.
31234 (ipa_compute_jump_functions_for_edge): Use the above functions to
31235 construct bits and vr parts of jump functions.
31236 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31237 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31239 (ipcp_grow_transformations_if_necessary): Also allocate
31240 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31242 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31243 them. Fix too long lines.
31244 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31245 vr_known being removed.
31246 (ipa_read_jump_function): Use new setter functions to construct bits
31247 and vr parts of jump functions or set them to NULL.
31248 (write_ipcp_transformation_info): Adjust for bits being pointers.
31249 (read_ipcp_transformation_info): Likewise.
31250 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31252 Include gt-ipa-prop.h.
31253 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31255 (ipcp_store_bits_results): Likewise.
31256 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31257 Do not write to existing jump functions but use a temporary instead.
31259 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31262 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31263 attempt to use its first operand as BIT_FIELD_REF base.
31265 2017-03-01 Richard Biener <rguenther@suse.de>
31267 PR middle-end/79721
31268 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
31269 interpolating formula in wrapping arithmetic.
31270 (chrec_apply): Convert chrec_evaluate return value to wanted type.
31272 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31274 PR tree-optimization/79734
31275 * tree-vect-generic.c (expand_vector_condition): Optimize
31276 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
31277 Handle VEC_COND_EXPR where comparison has different inner width from
31278 type's inner width.
31280 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
31282 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
31283 markup, and similar issues. Remove @opindex entries for things
31284 that aren't options. Add missing -mmpy-option entries.
31286 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31288 PR tree-optimization/79737
31289 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
31290 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
31291 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
31292 instead of byte_size. Formatting fix.
31293 (shift_bytes_in_array_right): Formatting fix.
31295 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
31298 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
31299 condition on optimize for the leaf function test.
31301 2017-02-28 Martin Liska <mliska@suse.cz>
31304 * read-rtl-function.c (function_reader::handle_unknown_directive):
31305 Bail out when one uses -flto.
31307 2017-02-28 Martin Liska <mliska@suse.cz>
31309 * common.opt: Replace space with tabular for options of <number>
31311 * config/i386/i386.opt: Show <number> value for
31312 -mlarge-data-threshold.
31313 * opts.c (print_filtered_help): Do not display number in hexadecimal
31316 2017-02-28 Martin Liska <mliska@suse.cz>
31318 * common.opt: Fix --help=option -Q for options which are of
31321 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31323 * config/i386/i386.c (print_reg): Error out for values
31324 of 8-bit size in invalid integer register.
31326 2017-02-28 Martin Sebor <msebor@redhat.com>
31328 PR tree-optimization/79691
31329 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31331 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31334 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31335 gcc_unreachable with output_operand_lossage.
31337 2017-02-28 Richard Biener <rguenther@suse.de>
31339 PR tree-optimization/79740
31340 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31342 (visit_nary_op): Insert the nary into the hashtable if we
31343 pattern-matched sth.
31344 * tree-ssa-pre.c (eliminate_insert): Robustify.
31346 2017-02-28 Richard Biener <rguenther@suse.de>
31348 PR middle-end/79731
31349 * fold-const.c (decode_field_reference): Reject out-of-bound
31352 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31354 * config/i386/i386.c: Include intl.h.
31355 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31356 instead of just cond ? "..." : "...".
31357 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31358 * coverage.c (read_counts_file): Likewise.
31359 * omp-offload.c: Include intl.h.
31360 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31361 of just cond ? "..." : "...".
31362 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31363 of just cond ? "..." : "...".
31365 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31368 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31370 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31371 'tune for' CPU name.
31372 * config/arm/arm-cpu-data.h: Regenerated.
31374 2017-02-28 Richard Biener <rguenther@suse.de>
31376 PR tree-optimization/79732
31377 * tree-inline.c (expand_call_inline): Do not shadow var.
31379 2017-02-28 Richard Biener <rguenther@suse.de>
31381 PR tree-optimization/79723
31382 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31383 address-space properly.
31385 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31387 * doc/optinfo.texi (Optimization groups): Fix option used for
31389 * doc/invoke.texi (-fopt-info): Document "omp".
31390 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31391 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31392 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31393 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31394 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31396 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31398 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31399 OPTGROUP_OMP with "omp".
31401 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31404 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31405 for arithmetic shift of unsigned V2DI.
31407 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31409 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31410 arc/linux.h headers.
31411 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31412 (LINK_SPEC): Likewise.
31413 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31414 (EXTRA_SPECS): Likewise.
31415 (STARTFILE_SPEC): Likewise.
31416 (ENDFILE_SPEC): Likewise.
31417 (LIB_SPEC): Likewise.
31418 (TARGET_SDATA_DEFAULT): Likewise.
31419 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31420 (MULTILIB_DEFAULTS): Likewise.
31421 (DWARF2_UNWIND_INFO): Likewise.
31422 * config/arc/big.h: New file.
31423 * config/arc/elf.h: Likewise.
31424 * config/arc/linux.h: Likewise.
31425 * config/arc/t-uClibc: Remove.
31427 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31429 PR tree-optimization/77536
31430 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31431 (tree_transform_and_unroll_loop): Use above function to compute the
31432 estimated niter of unrolled loop and use it when scaling profile.
31433 Also use count info rather than frequency if it's non-zero.
31434 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31435 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31436 (vect_transform_loop): Call above function.
31438 2017-02-27 Richard Biener <rguenther@suse.de>
31440 PR tree-optimization/45397
31441 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31442 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31443 (visit_nary_op): Add pattern matching for CSEing sign-changed
31444 or truncated operations with wider ones.
31446 2017-02-27 Richard Biener <rguenther@suse.de>
31448 PR tree-optimization/79690
31449 * tree-vect-stmts.c (vectorizable_store): Use vector type
31450 built from the DR with address-space.
31452 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31454 * doc/invoke.texi (Optimize Options): Refine the description
31455 of asan-use-after-return.
31457 2017-02-25 Alan Modra <amodra@gmail.com>
31459 PR rtl-optimization/79584
31460 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31461 base, not ad->base_term, the reg within base. Remove assertion
31462 that ad->base == ad->base_term. Replace gen_int_mode using
31463 bogus mode with const0_rtx.
31465 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31467 PR middle-end/79396
31468 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31469 FMA_EXPR like tcc_binary or tcc_unary.
31471 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31474 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31476 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31477 (output_loc_operands): Handle DW_OP_call_ref and
31478 DW_OP_GNU_variable_value.
31479 (struct variable_value_struct): New type.
31480 (struct variable_value_hasher): Likewise.
31481 (variable_value_hash): New variable.
31482 (string_types): Remove.
31483 (copy_loc_descr): New function.
31484 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31485 (prepend_loc_descr_to_each): New function.
31486 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31487 instead of add_loc_descr_to_each if the first argument is single
31488 location list and the second has multiple.
31489 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31490 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31491 when looking for variable value which doesn't have other location info.
31492 (loc_list_from_tree): Formatting fix.
31493 (gen_array_type_die): Simplify DW_AT_string_length handling.
31494 (adjust_string_types): Remove.
31495 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31496 string_types. Call resolve_variable_values.
31497 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31498 (resolve_addr_in_expr): Likewise. Add A argument.
31499 (copy_deref_exprloc): Remove deref argument. Adjust for the
31500 original expression being DW_OP_GNU_variable_value with optionally
31501 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31502 optionally after it.
31503 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31504 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31505 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31506 (variable_value_hasher::hash, variable_value_hasher::equal): New
31508 (resolve_variable_value_in_expr, resolve_variable_value,
31509 resolve_variable_values, note_variable_value_in_expr,
31510 note_variable_value): New functions.
31511 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31514 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31517 * opts.h (handle_generated_option): Add GENERATED_P argument.
31518 * opts-common.c (handle_option): Adjust function comment.
31519 (handle_generated_option): Add GENERATED_P argument, pass it to
31521 (control_warning_option): Pass false to handle_generated_option
31523 * opts.c (maybe_default_option): Pass true to handle_generated_option
31525 * optc-gen.awk: Likewise.
31527 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31529 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31530 a REG, look at the REG it is a SUBREG of.
31531 (splitter for cmpeqsi_t): Ditto.
31533 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31535 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31536 the special USEs with the pattern of the insn, not the insn itself.
31538 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31541 * doc/invoke.texi: Document -mload-store-pairs.
31543 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31544 Sandra Loosemore <sandra@codesourcery.com>
31546 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31547 argument isn't a CONST_INT.
31548 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31549 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31550 (nios2_validate_compare): Bypass alternate compare logic if *op2
31551 is not a CONST_INT.
31552 (ldstwm_operation_p): Return false if first_base is not a REG or
31553 if first_offset is not a CONST_INT.
31555 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31557 * config/cris/cris.md: Use correct operand in a define_peephole2.
31559 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31561 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31563 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31565 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31566 this_insn if it is an INSN or JUMP_INSN.
31567 (force_offsettable): Look at base, not at addr.
31568 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31569 on things that aren't necessarily CONST_INTs.
31571 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31573 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31574 -mfpmath=sse is the default also for x86-32 targets with SSE2
31575 instruction set when @option{-ffast-math} is enabled
31577 2017-02-24 Jeff Law <law@redhat.com>
31579 PR rtl-optimizatoin/79286
31580 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31583 2017-02-24 Richard Biener <rguenther@suse.de>
31585 PR tree-optimization/79389
31586 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31589 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31591 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31592 function comment to reflect reality.
31593 (loop_exits_before_overflow): Fix typo in function description.
31595 2017-02-24 Richard Biener <rguenther@suse.de>
31597 PR tree-optimization/79389
31598 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31599 properly that a threading opportunity exists. Detect conditional
31600 copy/constant propagation opportunities.
31602 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31604 * config/visium/visium.md (type): Add trap.
31605 (b): New mode attribute.
31606 (*btst): Rename into...
31607 (*btst<mode>): ...this and adjust.
31608 (*cbranchsi4_btst_insn): Rename into...
31609 (*cbranch<mode>4_btst_insn): ...this and adjust.
31610 (trap): New define_insn.
31612 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31614 PR tree-optimization/79389
31615 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31616 (noce_reversed_cond_code): New function.
31617 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31618 reversed_comparison_code. Formatting fix.
31619 (noce_try_store_flag): Test rev_cond != NULL in addition to
31620 reversed_comparison_code.
31621 (noce_try_store_flag_constants): Likewise.
31622 (noce_try_store_flag_mask): Likewise.
31623 (noce_try_addcc): Use rev_cond if non-NULL instead of
31624 reversed_comparison_code.
31625 (noce_try_cmove_arith): Likewise. Formatting fixes.
31626 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31627 (noce_find_if_block): Initialize rev_cond.
31628 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31629 instead of false as last argument never attempt to reverse it
31632 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31634 PR tree-optimization/79663
31635 * tree-predcom.c (combine_chains): Process refs in reverse order
31636 only for ZERO length chains, and add explaining comment.
31638 2017-02-23 Jeff Law <law@redhat.com>
31640 PR tree-optimization/79578
31641 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31642 in call to operand_equal_p.
31644 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31647 * config/i386/cpuid.h: Fix another undefined behavior.
31649 2017-02-23 Richard Biener <rguenther@suse.de>
31651 PR tree-optimization/79683
31652 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31653 vector types for data-refs.
31655 2017-02-23 Martin Liska <mliska@suse.cz>
31657 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31659 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31661 PR middle-end/79665
31662 * internal-fn.c (get_range_pos_neg): Moved to ...
31663 * tree.c (get_range_pos_neg): ... here. No longer static.
31664 * tree.h (get_range_pos_neg): New prototype.
31665 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31666 are known to be in between 0 and signed maximum inclusive, try to
31667 expand both unsigned and signed divmod and use the cheaper one from
31670 2017-02-22 Jeff Law <law@redhat.com>
31672 PR tree-optimization/79578
31673 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31674 to compare base operands.
31676 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31679 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31680 gpc_reg_operand instead of fpr_reg_operand.
31682 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31684 * config/mips/mips.c (mips_return_in_memory): Force FP
31685 vector types to be returned in memory for o32 ABI.
31687 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31689 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31690 instead of DW_TAG_member for static data member declarations and don't
31691 set no_linkage_name for static inline data members.
31692 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31695 2017-02-22 Martin Liska <mliska@suse.cz>
31697 * doc/invoke.texi: Replace inequality signs with square brackets
31700 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31703 * lra-constraints.c (simplify_operand_subreg): Handle
31704 WORD_REGISTER_OPERATIONS targets.
31706 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31709 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31710 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31711 elimination by swapping fld*.
31713 2017-02-22 Richard Biener <rguenther@suse.de>
31715 PR tree-optimization/79673
31716 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31717 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31718 irrelevant address-space qualifiers and avoiding a
31719 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31721 2017-02-22 Richard Biener <rguenther@suse.de>
31723 PR tree-optimization/79666
31724 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31725 to not symbolically negate if that may introduce undefined
31728 2017-02-22 Martin Liska <mliska@suse.cz>
31731 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31732 * data-streamer-out.c (streamer_write_gcov_count_stream):
31734 * value-prof.c (stream_out_histogram_value): Make assert more
31735 precise based on type of counter.
31737 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31740 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31741 Use nonimmediate_operand instead of memory_operand for operand 1.
31742 (float-extend standard_x87sse_constant_load splitter): Ditto.
31744 2017-02-21 Jeff Law <law@redhat.com>
31746 PR tree-optimization/79621
31747 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31748 blocks with edges to themselves.
31750 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31753 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31754 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31755 Use gimple_call_builtin_p.
31758 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31759 on temporarily removed DEBUG_INSNs.
31761 PR tree-optimization/79649
31762 * tree-loop-distribution.c (classify_partition): Give up on
31763 non-generic address space loads/stores.
31765 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31767 * doc/loop.texi (Loop manipulation): Remove nonexistent
31768 tree_ssa_loop_version from the documentation.
31769 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31771 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31774 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31775 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31776 * config/rs6000/rs6000.c: Include except.h.
31777 (rs6000_expand_split_stack_prologue): Call
31778 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31780 2017-02-21 Martin Jambor <mjambor@suse.cz>
31783 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31784 have been analyzed.
31786 2017-02-21 Martin Jambor <mjambor@suse.cz>
31788 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31789 for backward compatibility only.
31790 * doc/invoke.texi (Option Summary): Remove all references to
31791 -fipa-cp-alignment.
31793 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31797 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31799 * lra-constraints.c (curr_insn_transform): Handle
31800 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31802 2017-02-21 Martin Liska <mliska@suse.cz>
31804 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31806 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31808 PR translation/79638
31809 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31811 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31814 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31815 (arm_function_ok_for_sibcall): Return false for an indirect call by
31816 descriptor if all the argument registers are used.
31817 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31818 alignment of the function.
31820 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31822 PR tree-optimization/61441
31823 * simplify-rtx.c (simplify_const_unary_operation): For
31824 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31825 the sNaN unmodified.
31827 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31829 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31830 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31831 instead of SYSTEM_HEADER_DIR.
31833 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31834 Martin Liška <mliska@suse.cz>
31836 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31837 Fix typos and grammar, use active voice, and clarify.
31839 2017-02-20 Marek Polacek <polacek@redhat.com>
31841 PR middle-end/79537
31842 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31845 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31847 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31850 * config/i386/i386.c (ix86_expand_builtin): Handle
31851 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31852 ix86_builtins_isa[fcode].isa as a requirement of those
31853 flags and any other flag in the bitmask.
31854 (ix86_init_mmx_sse_builtins): Use 0 instead of
31855 ~OPTION_MASK_ISA_64BIT as mask.
31856 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31857 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31858 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31859 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31861 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31864 * lra-constraints.c (split_reg): Check requested split mode
31865 is supported by the register.
31867 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31869 * lra-constraints.c (simplify_operand_subreg): Remove early
31872 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31875 * lra-constraints.c (curr_insn_transform): Tighten condition
31876 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31878 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31881 * lra-constraints.c (curr_insn_transform): Handle
31882 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31884 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31887 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31889 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31891 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31894 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31897 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31899 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31902 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31904 * final.c (last_columnnum, override_columnnum): New variables.
31905 (final_start_function): Set last_columnnum, pass it to begin_prologue
31906 hook and pass 0 to dwarf2out_begin_prologue.
31907 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31908 to source_line debug hook.
31909 (notice_source_line): Compute last_columnnum and for debug_column_info
31910 return true on column changes.
31911 * debug.h (struct gcc_debug_hooks): Add column argument to
31912 source_line and begin_prologue hooks.
31913 (debug_nothing_int_charstar_int_bool): Remove prototype.
31914 (debug_nothing_int_int_charstar,
31915 debug_nothing_int_int_charstar_int_bool): New prototypes.
31916 (dwarf2out_begin_prologue): Add column argument.
31917 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31918 begin_prologue hooks.
31919 (debug_nothing_int_charstar_int_bool): Remove.
31920 (debug_nothing_int_int_charstar,
31921 debug_nothing_int_int_charstar_int_bool): New functions.
31922 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31923 through to dwarf2out_source_line.
31924 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31925 (dwarf2out_source_line): Add column argument, emit it if requested.
31926 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31928 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31929 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31930 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31931 through to dwarf2out_begin_prologue.
31932 (vmsdbgout_source_line): Add column argument, pass it through to
31933 dwarf2out_source_line.
31934 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31935 dbxout_source_line caller.
31936 (dbxout_source_line): Add column argument.
31938 * common.opt (gno-column-info, gcolumn-info): New options.
31939 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31940 (check_die): Also test for multiple DW_AT_decl_column attributes.
31941 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31942 DW_AT_decl_column if requested.
31943 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31945 (gen_variable_die): Likewise.
31946 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31947 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31950 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31951 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31952 (ix86_handle_option): Handle OPT_m3dnowa.
31953 * doc/invoke.texi (-m3dnowa): Document.
31954 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31955 -m3dnowa instead of -m3dnow -march=athlon.
31958 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31959 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31961 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31964 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31965 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31966 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31967 generator for vsx_xxpermdi_<mode>_be.
31968 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31969 force big-endian semantics.
31970 (vsx_xxpermdi_<mode>_be): New define_expand with same
31971 implementation as previous version of vsx_xxpermdi_<mode>.
31973 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31975 PR tree-optimization/79327
31976 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31977 variable, its initialization and use.
31979 2017-02-17 Julia Koval <julia.koval@intel.com>
31981 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31982 (OPTION_MASK_ISA_PKU_UNSET): New.
31983 (ix86_handle_option): Handle -mrdpid.
31984 * config/i386/cpuid.h (bit_RDPID): New.
31985 * config/i386/driver-i386.c (host_detect_local_cpu):
31986 Detect RDPID feature.
31987 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31988 * config/i386/i386-c.c (ix86_target_macros_internal):
31990 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31991 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31992 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31993 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31994 * config/i386/i386.md (define_insn "rdpid"): New.
31995 * config/i386/i386.opt Add -mrdpid.
31996 * config/i386/immintrin.h (_rdpid_u32): New.
31998 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32000 PR rtl-optimization/79541
32001 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32002 instead of transforming it into USE.
32004 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32006 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32007 If HONOR_SNANS (SFmode) force the input to a register.
32008 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32009 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32010 an frsp or similar insn.
32012 2017-02-17 Martin Liska <mliska@suse.cz>
32014 PR rtl-optimization/79577
32015 * params.def (selsched-max-sched-times): Increase minimum to 1.
32017 2017-02-17 Martin Liska <mliska@suse.cz>
32019 PR rtl-optimization/79574
32020 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32022 2017-02-17 Martin Liska <mliska@suse.cz>
32024 PR tree-optimization/79529
32025 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32026 ssa_defined_default_def_p to handle cases which are implicitly
32028 * tree-ssa.c (ssa_defined_default_def_p): New function.
32029 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32030 which are implicitly defined.
32031 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32033 2017-02-17 Richard Biener <rguenther@suse.de>
32035 PR middle-end/79576
32036 * params.def (max-ssa-name-query-depth): Limit to 10.
32038 2017-02-17 Richard Biener <rguenther@suse.de>
32040 PR tree-optimization/79552
32041 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32044 2017-02-17 Richard Biener <rguenther@suse.de>
32047 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32049 2017-02-17 Marek Polacek <polacek@redhat.com>
32051 PR middle-end/79536
32052 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32053 (fold_negate_expr): New wrapper.
32055 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32057 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32058 Correct terminology and de-emphasize pre-standard behavior.
32060 2017-02-16 Alan Modra <amodra@gmail.com>
32062 PR rtl-optimization/79286
32063 * ira.c (def_dominates_uses): New function.
32064 (update_equiv_regs): Don't create an equivalence for insns that
32065 may trap where the register def does not dominate the use.
32067 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32069 PR rtl-optimization/78127
32070 * lra.c (lra): Call lra_eliminate before finish the loop after
32073 2017-02-16 Richard Biener <rguenther@suse.de>
32075 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32077 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32078 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32079 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32080 (isl_val_int_from_wi): New function.
32081 (extract_affine_gmp): Rename to ...
32082 (extract_affine_wi): ... this, take a widest_int.
32083 (extract_affine_int): Just wrap extract_affine_wi.
32084 (add_param_constraints): Use isl_val_int_from_wi.
32085 (add_loop_constraints): Likewise, and extract_affine_wi.
32087 2017-02-15 Jeff Law <law@redhat.com>
32089 PR middle-end/79521
32090 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32091 ira_init_register_move_cost_if_necessary.
32093 2017-02-15 Martin Sebor <msebor@redhat.com>
32095 PR middle-end/32003
32096 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32097 removed in a prior commit.
32099 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32101 PR tree-optimization/79347
32102 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32103 counters during peeling.
32105 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32107 * Makefile.in (site.exp): Remove "set ISLVER".
32109 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32112 * real.c (real_from_integer): Call real_convert even for decimal.
32114 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32117 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32119 2017-02-14 Andrew Pinski <apinski@cavium.com>
32121 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32122 cores and change the partno/implementer to be correct.
32123 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32124 the 'B" as the implementer.
32125 * config/aarch64/aarch64-tune.md: Regenerate.
32127 2017-02-14 Carl Love <cel@us.ibm.com>
32129 * config/rs6000/rs6000.c: Add case statement entry to make the
32130 xvcvuxdsp built-in argument unsigned.
32131 * config/rs6000/vsx.md: Fix the source and return operand types so they
32132 match the instruction definitions from the ISA document. Fix typo
32133 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32136 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32139 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32140 member early_clobber_alts.
32141 * lra-lives.c (reg_early_clobber_p): New.
32142 (process_bb_lives): Use it.
32143 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32144 (debug_operand_data): Initialize early_clobber_alts.
32145 (setup_operand_alternative): Set up early_clobber_alts.
32146 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32147 alternatives to new_insn_reg.
32148 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32150 (lra_update_insn_regno_info): Pass the new arg.
32152 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32154 PR middle-end/79505
32155 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32156 (new_oacc_loop_raw): Don't clear already cleared fields.
32159 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32160 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32161 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32163 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32166 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32168 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32171 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32172 the extra instruction to the right place to store 128-bit constant
32175 2017-02-14 Martin Sebor <msebor@redhat.com>
32177 PR middle-end/79448
32178 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32179 warning for strings of unknown length.
32181 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32183 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32185 2017-02-14 Jeff Law <law@redhat.com>
32188 * ira-costs.c (scan_one_insn): Initialize register move costs
32189 for pseudos seen in USE/CLOBBER insns.
32191 PR tree-optimization/79095
32192 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32193 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32194 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32195 if the operands are known to be not equal, then the resulting range
32197 (intersect_ranges): If the new range is ~[0,0] and the old range is
32198 wide, then prefer ~[0,0].
32199 * tree-vrp.c (overflow_comparison_p_1): New function.
32200 (overflow_comparison_p): New function.
32201 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32202 if NAME is used in an overflow test.
32203 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32204 overflow check that can be expressed as an equality test, then adjust
32205 ops to be that equality test.
32207 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32209 * config/s390/s390-builtin-types.def: Remove flags argument.
32210 * config/s390/s390.c (s390_init_builtins): Likewise.
32212 2017-02-14 Martin Liska <mliska@suse.cz>
32214 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32215 vector. Fix trailing white spaces.
32217 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32219 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32222 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32224 PR rtl-optimization/68664
32225 * config/arm/arm.c (arm_sched_can_speculate_insn):
32226 New function. Declare prototype.
32227 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32229 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32231 PR rtl-optimization/68664
32232 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32234 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32236 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32238 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32239 max skip bytes for function, loop and jump.
32241 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32243 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32244 ABS_EXPR for gimple dump.
32246 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32249 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32251 PR tree-optimization/79408
32252 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32253 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32254 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32255 also if rhs1 is INTEGER_CST.
32257 2017-02-14 Richard Biener <rguenther@suse.de>
32259 PR middle-end/79432
32260 * tree-into-ssa.c (insert_phi_nodes): When the function can
32261 have abnormal edges rewrite SSA names with broken use-def
32262 dominance out of SSA and register them for PHI insertion.
32264 2017-02-13 Martin Sebor <msebor@redhat.com>
32266 PR middle-end/79496
32267 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32268 clearing info.nowrite flag when snprintf size argument is a range.
32270 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32272 * cprop.c (cprop_jump): Add missing space in string literal.
32273 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
32274 (get_constraint_for_component_ref): Likewise.
32275 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
32276 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
32277 * lra-constraints.c (process_alt_operands): Likewise.
32278 * ipa-inline.c (inline_small_functions): Likewise.
32279 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
32280 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
32281 * trans-mem.c (diagnose_tm_1_op): Likewise.
32282 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
32283 (grid_parallel_clauses_gridifiable): Likewise.
32285 * config/nvptx/mkoffload.c (process): Add space in between
32288 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
32289 "MOD4_SSE_REGS" and "ALL_REGS".
32291 * spellcheck.c (test_data): Add , in between "foo" and "food".
32293 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32296 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
32297 boundary crossing check and subsequent code generation agree.
32299 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32301 * config/aarch64/aarch64.c (has_memory_op): Delete.
32302 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32305 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32307 PR rtl-optimization/79388
32308 PR rtl-optimization/79450
32309 * combine.c (distribute_notes): When removing TEM_INSN for which
32310 corresponding dest has last value recorded, invalidate that last
32313 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32315 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32316 of explicit '@'. Add missing assembly comment marker on branch costs
32319 2017-02-13 Nathan Sidwell <nathan@acm.org>
32321 * gengtype-lex.l (<in_struct>): Add '/'.
32323 2017-02-13 Martin Liska <mliska@suse.cz>
32326 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32328 2017-02-13 Richard Biener <rguenther@suse.de>
32330 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32332 * configure: Re-generate.
32333 * config.in: Likewise.
32334 * graphite-dependences.c: Simplify as if
32335 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32336 * graphite-isl-ast-to-gimple.c: Likewise.
32337 * graphite-optimize-isl.c: Likewise.
32338 * graphite-poly.c: Likewise.
32339 * graphite-sese-to-poly.c: Likewise.
32340 * graphite.h: Likewise.
32341 * toplev.c: Include isl/version.h and use isl_version () for
32342 printing the ISL version.
32343 * doc/install.texi: Update ISL requirement.
32345 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32347 * doc/standards.texi (Standards): Update reference to
32350 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32352 * doc/extend.texi (Named Address Spaces): sourceware.org now
32354 * doc/install.texi (Binaries): Ditto.
32357 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32359 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32360 standard terminology "stringize"/"stringizing" throughout.
32361 * doc/cppinternals.texi: Likewise.
32363 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32365 * doc/extend.texi: Fix some spelling mistakes and typos.
32366 * doc/invoke.texi: Likewise.
32368 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32371 * params.def (inline-min-speedup) Change from 10 to 8.
32373 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32375 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32378 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32381 * ipa-inline-analysis.c (get_minimal_bb): New function.
32382 (record_modified): Use it.
32383 (remap_edge_change_prob): Handle also ancestor functions.
32385 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32387 * doc/contrib.texi (Contributors): Remove broken link into
32388 the Mauve CVS repository.
32390 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32392 PR middle-end/79454
32393 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32394 result computation whenever lhs doesn't have vector mode, not
32395 just when it has BLKmode.
32397 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32399 * doc/makefile.texi (profiledbootstrap): Refer to the
32400 installation instructions only in textual form.
32402 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32405 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32407 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32409 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32410 (Specific): Update mingw-w64 reference.
32412 (Specific): Remove broken link to Renesas RX processor.
32414 2017-02-10 Richard Biener <rguenther@suse.de>
32416 * toplev.c (process_options): Do not mention obsolete graphite
32417 options when printing sorry message about missing graphite support.
32418 Mention -floop-nest-optimize.
32420 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32422 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32423 (vtst_p16): Likewise.
32424 (vtstq_p8): Likewise.
32425 (vtstq_p16): Likewise.
32427 (vtstq_p64): Likewise.
32428 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32429 (vset_lane_p64): New.
32430 (vsetq_lane_p64): New.
32432 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32434 PR tree-optimization/79411
32435 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32436 stmt operands are SSA_NAMEs used in abnormal phis.
32437 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32440 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32443 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32446 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32448 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32450 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32452 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32456 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32457 not arbitrary TREE_CONSTANT.
32460 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32461 "omp declare target link" attribute unless is_global_var.
32462 * omp-offload.c (find_link_var_op): Likewise.
32464 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32465 Chung-Lin Tang <cltang@codesourcery.com>
32467 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32469 (gimplify_adjust_omp_clauses): Don't delete TILE.
32470 (gimplify_omp_for): Deal with TILE.
32471 * internal-fn.c (expand_GOACC_TILE): New function.
32472 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32474 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32475 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32477 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32478 avoid DIV for outermost collapse var.
32479 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32480 Remove out of date comments, fix whitespace.
32481 * omp-general.c (omp_extract_for_data): Deal with tiling.
32482 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32483 adjust OLF_DIM_BASE value.
32484 (struct omp_for_data): Add tiling field.
32485 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32486 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32487 for auto loops. Remove default auto determining, moved to
32488 oacc_loop_fixed_partitions.
32489 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32490 stmts, add e_mask field.
32491 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32492 (oacc_thread_numbers): Use oacc_dim_call.
32493 (oacc_xform_tile): New.
32494 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32495 (finish_oacc_loop): Adjust for ifns vector.
32496 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32497 add case for GOACC_TILE fns.
32498 (oacc_loop_xform_loop): Delete.
32499 (oacc_loop_process): Iterate over call list directly, and add
32500 handling for GOACC_TILE fns.
32501 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32503 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32504 vector partitioning to outer loops. Assign 2 partitions to loops
32505 when available. Add TILE handling.
32506 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32507 (execite_oacc_device_lower): Process GOACC_TILE fns,
32508 ignore unknown specs.
32509 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32510 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32511 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32513 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32515 * configure.ac (ACX_BUGURL): Update.
32516 * configure: Regenerate.
32518 2017-02-09 Richard Biener <rguenther@suse.de>
32520 PR tree-optimization/69823
32521 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32522 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32524 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32526 * config/arc/arc-c.def: Add __NPS400__ definition.
32527 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32528 (TARGET_NPS400): Define.
32530 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32532 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32534 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32535 pointer, arch_info.
32536 (arc_cpu_types): Fill the arch_info field with a pointer into the
32537 arc_arch_types table.
32538 (arc_selected_cpu): Declare.
32539 * config/arc/arc.c (arc_selected_cpu): Make global.
32540 (arc_selected_arch): Delete.
32541 (arc_base_cpu): Delete.
32542 (arc_override_options): Remove references to deleted variables,
32543 update access to arch information.
32544 (ARC_OPT): Update access to arch information.
32545 (ARC_OPTX): Likewise.
32546 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32547 (TARGET_ARC600): Update access to arch information.
32548 (TARGET_ARC601): Likewise.
32549 (TARGET_ARC700): Likewise.
32550 (TARGET_EM): Likewise.
32551 (TARGET_HS): Likewise.
32552 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32555 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32558 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32559 condition/operands for integer GE/LE/GEU/LEU operations.
32561 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32563 PR translation/79397
32564 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32567 2017-02-08 Martin Jambor <mjambor@suse.cz>
32570 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32571 whether allocation happened.
32572 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32573 nothing was allocated.
32575 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32577 PR tree-optimization/79408
32578 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32579 constant, but SSA_NAME with a known integer range, use the minimum
32580 of that range instead of op1 to determine if modulo can be replaced
32581 with its first operand.
32583 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32585 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32587 2017-02-08 Richard Biener <rguenther@suse.de>
32589 PR tree-optimization/71824
32590 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32591 Check all loops contained in the merged region.
32593 2017-02-07 Andrew Pinski <apinski@cavium.com>
32595 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32597 2017-02-07 Andrew Pinski <apinski@cavium.com>
32599 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32600 (thunderxt88): Likewise.
32601 (thunderxt81): Disable LSE and change v8.1 to v8.
32602 (thunderxt83): Likewise.
32604 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32605 Richard Biener <rguenther@suse.de>
32607 PR middle-end/79399
32608 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32609 type from int to size_t.
32610 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32612 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32614 PR rtl-optimization/79386
32615 * cprop.c (bypass_conditional_jumps): Initialize
32616 bypass_last_basic_block already before splitting bbs after
32617 unconditional traps...
32618 (bypass_conditional_jumps): ... rather than here.
32621 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32622 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32623 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32624 fix -masm=intel patterns.
32626 2017-02-07 Richard Biener <rguenther@suse.de>
32628 PR tree-optimization/79256
32629 PR middle-end/79278
32630 * builtins.c (get_object_alignment_2): Use min_align_of_type
32631 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32632 and ADJUST_FIELD_ALIGN.
32634 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32636 * doc/tm.texi: Regenerate.
32637 * stor-layout.c (layout_decl): Adjust.
32638 (update_alignment_for_field): Likewise.
32639 (place_field): Likewise.
32640 (min_align_of_type): Likewise.
32641 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32642 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32643 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32644 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32645 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32646 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32647 * config/i386/i386.c (x86_field_alignment): Likewise.
32648 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32649 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32650 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32651 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32652 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32653 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32657 2017-01-30 Richard Biener <rguenther@suse.de>
32659 PR tree-optimization/79256
32660 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32661 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32664 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32666 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32667 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32668 builtins to SImode and emit a zero-extend, if necessary.
32670 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32672 * docs/invoke.texi (RISC-V Options): Alphabetize.
32674 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32676 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32679 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32681 * config/riscv/riscv.c: New file.
32682 * common/config/riscv/riscv-common.c: Likewise.
32683 * config.gcc: Likewise.
32684 * config/riscv/constraints.md: Likewise.
32685 * config/riscv/elf.h: Likewise.
32686 * config/riscv/generic.md: Likewise.
32687 * config/riscv/linux.h: Likewise.
32688 * config/riscv/multilib-generator: Likewise.
32689 * config/riscv/peephole.md: Likewise.
32690 * config/riscv/pic.md: Likewise.
32691 * config/riscv/predicates.md: Likewise.
32692 * config/riscv/riscv-builtins.c: Likewise.
32693 * config/riscv/riscv-c.c: Likewise.
32694 * config/riscv/riscv-ftypes.def: Likewise.
32695 * config/riscv/riscv-modes.def: Likewise.
32696 * config/riscv/riscv-opts.h: Likewise.
32697 * config/riscv/riscv-protos.h: Likewise.
32698 * config/riscv/riscv.h: Likewise.
32699 * config/riscv/riscv.md: Likewise.
32700 * config/riscv/riscv.opt: Likewise.
32701 * config/riscv/sync.md: Likewise.
32702 * config/riscv/t-elf-multilib: Likewise.
32703 * config/riscv/t-linux: Likewise.
32704 * config/riscv/t-linux-multilib: Likewise.
32705 * config/riscv/t-riscv: Likewise.
32706 * configure.ac: Likewise.
32707 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32708 Waterman as RISC-V maintainers.
32709 * doc/install.texi: Add RISC-V entries.
32710 * doc/invoke.texi: Add RISC-V options section.
32711 * doc/md.texi: Add RISC-V constraints section.
32712 * configure: Regenerated.
32714 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32717 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32718 false values to be constant vectors with all 0 or all 1 bits set.
32719 (vcondu<mode><mode>): Likewise.
32720 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32722 (fpmask_comparison_operator): Update comment.
32723 (vecint_comparison_operator): New predicate.
32724 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32725 vector conditionals when the true and false values are constant
32726 vectors with all 0 bits or all 1 bits set.
32728 2017-02-06 Martin Sebor <msebor@redhat.com>
32730 PR tree-optimization/79376
32731 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32733 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32735 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32736 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32737 to simplify split condition.
32739 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32741 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32742 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32745 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32747 PR rtl-optimization/68664
32748 * target.def (can_speculate_insn): New hook.
32749 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32750 * doc/tm.texi: Regenerate.
32751 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32752 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32753 (rs6000_sched_can_speculate_insn): New function.
32755 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32757 PR tree-optimization/79284
32758 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32759 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32760 vectorizable_mask_load_store, vectorizable_operation,
32761 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32762 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32763 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32764 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32765 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32766 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32767 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32768 is_gimple_assign (stmt). Replace another such test with
32769 is_gimple_assign (stmt).
32771 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32774 * config/avr/avr.c (rtl-iter.h): Include it.
32775 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32776 (avr_legitimate_combined_insn): ...and implementation.
32778 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32780 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32781 * config/s390/s390.c (s390_const_operand_ok)
32782 (s390_canonicalize_comparison, s390_extract_part)
32783 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32784 (s390_contiguous_bitmask_p, s390_rtx_costs)
32785 (legitimize_pic_address): Likewise.
32786 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32787 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32788 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32789 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32790 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32792 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32794 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32795 REGNO($0) == REGNO($1).
32797 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32799 * config/s390/linux.h(SIZE_TYPE): Add comment.
32801 2017-02-06 Julian Brown <julian@codesourcery.com>
32802 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32803 Virendra Pathak <virendra.pathak@broadcom.com>
32805 * config/aarch64/aarch64-cores.def: Change the scheduler
32807 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32808 * config/aarch64/thunderx2t99.md: New file.
32810 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32812 * doc/standards.texi (Go Language): Update link to language
32815 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32817 * tree-eh.c (lower_resx): Sanitize profile.
32818 (cleanup_empty_eh_move_lp): Likewise.
32820 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32823 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32825 * cfgloopmanip.h (loop_version): Update prototype.
32826 * modulo-sched.c (sms_schedule): Update call of loop_version.
32827 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32828 * tree-parloops.c (gen_parallel_loop): Likewise.
32829 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32830 * tree-ssa-loop-split.c (split_loop): Likewise.
32831 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32832 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32834 2017-02-05 Martin Liska <mliska@suse.cz>
32837 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32839 (print_operand_address): Initialize a struct to zero.
32841 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32843 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32844 garbage collector only in textual form.
32846 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32848 * doc/extend.texi (x86 specific memory model extensions for
32849 transactional memory): Simplify a phrase.
32851 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32854 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32855 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32856 (atomic_storedi_1): Likewise.
32858 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32860 PR tree-optimization/79338
32861 * tree-parloops.c (gather_scalar_reductions): Don't call
32862 vect_analyze_loop_form for loop->inner before destroying loop's
32865 2017-02-03 Martin Sebor <msebor@redhat.com>
32867 PR tree-optimization/79327
32868 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32869 when precision has resulted in leading zeros.
32870 (format_integer): Adjust the likely counter to assume an unknown
32871 argument that may be zero is non-zero.
32873 2017-02-03 Jason Merrill <jason@redhat.com>
32876 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32877 avoid copying non-taken branch.
32879 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32881 PR tree-optimization/79340
32882 * tree-vect-loop.c (vectorizable_reduction): Release
32883 vec_defs elements after safe_splicing them into other vectors.
32886 PR tree-optimization/79327
32887 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32888 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32890 (format_integer): Use wide_int_to_tree instead of build_int_cst
32891 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32892 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32893 of shortest and longest sequence.
32895 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32897 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32898 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32900 2017-02-03 Walter Lee <walt@tilera.com>
32903 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32904 after initial stackframe link reg save.
32905 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32907 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32910 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32911 wu for stxssp alternative.
32913 2017-02-03 Martin Sebor <msebor@redhat.com>
32915 PR tree-optimization/79352
32916 * gimple-fold.c (get_range_strlen): Add argument.
32917 (get_range_strlen): Change return type to bool.
32918 (get_maxval_strlen): Pass in a dummy argument.
32919 * gimple-fold.h (get_range_strlen): Change return type to bool.
32920 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32921 * tree.h (array_at_struct_end_p): Add argument.
32922 * tree.c (array_at_struct_end_p): Handle it.
32924 2017-02-03 Martin Liska <mliska@suse.cz>
32927 * multiple_target.c (create_dispatcher_calls): Redirect edge
32928 from a caller of a dispatcher.
32929 (expand_target_clones): Make the clones local.
32930 (ipa_target_clone): Do both target clones and resolvers.
32931 (ipa_dispatcher_calls): Remove the pass.
32932 (pass_dispatcher_calls::gate): Likewise.
32933 (make_pass_dispatcher_calls): Likewise.
32934 * passes.def (pass_target_clone): Put as very first IPA early
32937 2017-02-03 Martin Liska <mliska@suse.cz>
32939 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32940 in case of a function with ifunc attribute.
32942 2017-02-03 Martin Liska <mliska@suse.cz>
32944 * cgraph.c (cgraph_node::dump): Dump function version info.
32945 * symtab.c (symtab_node::dump_base): Add missing new line.
32947 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32949 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32950 (ifcombine_ifandif): Use it.
32952 2017-02-03 Martin Liska <mliska@suse.cz>
32954 * doc/invoke.texi: Document default value for
32955 use-after-scope-direct-emission-threshold.
32957 2017-02-03 Martin Liska <mliska@suse.cz>
32959 PR tree-optimization/79339
32960 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32961 (format_floating): Likewise.
32963 2017-02-03 Martin Liska <mliska@suse.cz>
32966 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32968 (ipa_node_params_t::remove): Likewise.
32969 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32970 initialization from removed ipa_node_params_t::insert.
32971 (ipa_node_params::~ipa_node_params): Move from removed
32972 ipa_node_params_t::release.
32973 * symbol-summary.h (symbol_summary::m_released): New member.
32974 Do not release a summary twice. Do not allow to call finalizer
32975 for types of a summary that live in GGC memory.
32977 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32979 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32982 2017-02-02 Martin Sebor <msebor@redhat.com>
32984 PR middle-end/79275
32985 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32986 (format_string): Tighten up the range of output for non-constant
32987 strings and correct the expected range for wide non-constant strings.
32989 2017-02-02 Martin Sebor <msebor@redhat.com>
32991 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32993 PR middle-end/32003
32994 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32996 (-fdump-tree-@var): Add to index and document how to come up
32997 with pass-specific option and dump file names.
32998 (-fdump-passes): Clarify where to look for output.
33000 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33002 PR middle-end/77445
33003 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33004 statistics of the analyzed path; allow threading for speed when
33005 any of BBs along the path are optimized for speed.
33007 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33009 PR middle-end/78468
33010 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33011 settings of the virtual registers.
33014 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33016 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33017 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33020 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33022 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33023 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33025 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33027 * config/s390/s390.md: Add missing comments with the expanded
33029 * config/s390/vector.md: Likewise.
33030 * config/s390/vx-builtins.md: Likewise.
33032 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33035 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33036 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33037 conditions on a single line.
33039 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33041 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33042 __S390_VX__ to __VX__.
33044 2017-02-01 Andrew Pinski <apinski@cavium.com>
33046 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33047 stmt_info to record_stmt_cost.
33048 (vect_get_known_peeling_cost): Pass stmt_info if known to
33050 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33051 cpu_vector_cost field into
33052 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33053 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33054 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33055 splitting of scalar_stmt_cost and vec_stmt_cost.
33056 (thunderx_vector_cost): Likewise.
33057 (cortexa57_vector_cost): LIkewise.
33058 (exynosm1_vector_cost): Likewise.
33059 (xgene1_vector_cost): Likewise.
33060 (thunderx2t99_vector_cost): Improve after the splitting of the two
33062 (aarch64_builtin_vectorization_cost): Update for the splitting of
33063 scalar_stmt_cost and vec_stmt_cost.
33065 2017-02-01 Torvald Riegel <triegel@redhat.com>
33066 Richard Henderson <rth@redhat.com>
33068 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33069 conditional on existance of a fast atomic load.
33070 * optabs-query.c (can_atomic_load_p): New function.
33071 * optabs-query.h (can_atomic_load_p): Declare it.
33072 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33073 no fast atomic load is available for the particular size of access.
33074 (expand_atomic_compare_and_swap): Likewise.
33075 (expand_atomic_load): Likewise.
33076 (expand_atomic_store): Likewise.
33077 (expand_atomic_fetch_op): Likewise.
33078 * testsuite/lib/target-supports.exp
33079 (check_effective_target_sync_int_128): Remove x86 because it provides
33080 no fast atomic load.
33081 (check_effective_target_sync_int_128_runtime): Likewise.
33083 2017-02-01 Richard Biener <rguenther@suse.de>
33085 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33086 (graphite_transform_loops): Provide opt-info for optimized nests.
33087 * tree-parloop.c (parallelize_loops): Provide opt-info for
33088 parallelized loops.
33090 2017-02-01 Richard Biener <rguenther@suse.de>
33092 PR middle-end/79315
33093 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33094 was not set before.
33096 2017-02-01 Richard Biener <rguenther@suse.de>
33098 PR tree-optimization/71824
33099 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33100 Verify the loops are valid in the merged SESE region.
33101 (scop_detection::can_represent_loop_1): Check analyzing the
33102 evolution of the number of iterations in the region succeeds.
33104 2017-01-31 Ian Lance Taylor <iant@golang.org>
33106 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33107 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33109 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33111 PR preprocessor/79210
33112 * input.c (get_substring_ranges_for_loc): Replace line_width
33113 assertion with error-handling.
33115 2017-01-31 Richard Biener <rguenther@suse.de>
33117 PR tree-optimization/77318
33118 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33119 (create_pw_aff_from_tree): Take loop parameter.
33120 (add_condition_to_pbb): Pass loop of the condition to
33121 create_pw_aff_from_tree.
33123 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33125 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33126 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33128 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33132 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33134 (convert_int_to_float128): Likewise.
33135 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33136 (convert_int_to_float128): Likewise.
33137 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33138 (UNSPEC_IEEE128_CONVERT): Likewise.
33139 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33140 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33141 Use local variables for IBM extended format.
33142 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33143 (fix_trunc<mode>si2_fprs): Likewise.
33144 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33145 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33146 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33147 to know that we can now have integers of all sizes in vector
33149 (fix<uns>_<mode>di2_hw): Likewise.
33150 (float<uns>_<mode>si2_hw): Likewise.
33151 (fix_<mode>si2_hw): Likewise.
33152 (fixuns_<mode>si2_hw): Likewise.
33153 (float<uns>_<mode>di2_hw): Likewise.
33154 (float_<mode>di2_hw): Likewise.
33155 (float_<mode>si2_hw): Likewise.
33156 (floatuns_<mode>di2_hw): Likewise.
33157 (floatuns_<mode>si2_hw): Likewise.
33158 (xscvqp<su>wz_<mode>): Delete, no longer used.
33159 (xscvqp<su>dz_<mode>): Likewise.
33160 (xscv<su>dqp_<mode>): Likewise.
33161 (ieee128_mfvsrd_64bit): Likewise.
33162 (ieee128_mfvsrd_32bit): Likewise.
33163 (ieee128_mfvsrwz): Likewise.
33164 (ieee128_mtvsrw): Likewise.
33165 (ieee128_mtvsrd_64bit): Likewise.
33166 (ieee128_mtvsrd_32bit): Likewise.
33168 2017-01-31 Martin Liska <mliska@suse.cz>
33171 * ipa-prop.c (ipa_free_all_node_params): Call release method
33172 instead of ~sumbol_summary to not to trigger double times
33175 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33177 PR tree-optimization/71691
33178 * bitmap.h (class auto_bitmap): New.
33179 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33180 is_maybe_undefined instead of ssa_undefined_value_p.
33182 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33184 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33185 __S390_ARCH_LEVEL__ to __ARCH__.
33187 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33189 PR tree-optimization/79267
33190 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33191 if should_remove_lhs_p is true.
33193 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33196 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33197 (add_alignment_attribute): New.
33198 (base_type_die): Add alignment attribute.
33199 (subrange_type_die): Likewise.
33200 (modified_type_die): Likewise.
33201 (gen_array_type_die): Likewise.
33202 (gen_descr_array_type_die: Likewise.
33203 (gen_enumeration_type_die): Likewise.
33204 (gen_subprogram_die): Likewise.
33205 (gen_variable_die): Likewise.
33206 (gen_field_die): Likewise.
33207 (gen_ptr_to_mbr_type_die): Likewise.
33208 (gen_struct_or_union_type_die): Likewise.
33209 (gen_subroutine_type_die): Likewise.
33210 (gen_typedef_die): Likewise.
33211 (base_type_cmp): Compare alignment attribute.
33213 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33216 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33217 (setb_unsigned) New pattern for setb with CCUNS.
33218 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33219 subfc./subfe sequence to avoid overflow problems. Generate a
33220 shorter sequence with cmpld/setb for power9.
33221 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33222 for generating subfc. instruction.
33223 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33224 now uses this instruction.
33226 2017-01-30 Ian Lance Taylor <iant@google.com>
33229 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33230 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33232 2017-01-30 Martin Sebor <msebor@redhat.com>
33234 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33235 Move constant to the right of a relational operator.
33236 (get_mpfr_format_length, format_character, format_string): Ditto.
33237 (should_warn_p, maybe_warn): Same.
33239 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33241 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33244 * asan.c (get_translation_unit_decl): Remove function.
33245 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33247 2017-01-30 Martin Liska <mliska@suse.cz>
33249 PR gcov-profile/79259
33250 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33251 -fprofile-generate.
33253 2017-01-30 Martin Liska <mliska@suse.cz>
33256 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33257 Initialize variables with NULL value.
33259 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33262 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33264 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33266 2017-01-30 Richard Biener <rguenther@suse.de>
33268 PR tree-optimization/79276
33269 * tree-vrp.c (process_assert_insertions): Properly adjust common
33270 when removing a duplicate.
33272 2017-01-30 Richard Biener <rguenther@suse.de>
33274 PR tree-optimization/79256
33275 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33276 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33278 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
33280 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
33283 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
33284 ("*r<noxa>sbg_<mode>_sll_bitmask")
33285 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
33286 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
33287 Use contiguous_bitmask_nowrap_operand.
33289 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33292 * config/rs6000/altivec.h (vec_xl): Revise #define.
33293 (vec_xst): Likewise.
33295 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
33297 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
33299 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
33301 PR rtl-optimization/79194
33302 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33303 traps before call to bypass_conditional_jumps.
33305 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33307 PR tree-optimization/71374
33308 * lra-constraints.c (check_conflict_input_operands): New.
33309 (match_reload): Use it.
33311 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33314 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33315 account to calculate conflict_set.
33317 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33319 PR rtl-optimization/78559
33320 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33321 other_insn in combine.
33323 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33325 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33326 uint16_type_node for BT_UINT16.
33328 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33330 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33331 "RTL Tests" to menu.
33332 (GIMPLE Tests): New node.
33333 (RTL Tests): New node.
33335 2017-01-27 Richard Biener <rguenther@suse.de>
33337 PR tree-optimization/79245
33338 * tree-loop-distribution.c (distribute_loop): Apply cost
33339 modeling also to detected patterns.
33341 2017-01-27 Richard Biener <rguenther@suse.de>
33343 PR tree-optimization/71433
33344 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33345 (compare_assert_loc): New function.
33346 (process_assert_insertions): Sort and optimize assert locations
33347 to remove duplicates and push down identical assertions on
33348 edges to their destination block.
33350 2017-01-27 Richard Biener <rguenther@suse.de>
33352 PR tree-optimization/79244
33353 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33354 out SSA names even if abnormal.
33356 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33358 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33359 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33360 instead of MPFR_RNDN.
33362 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33365 * arm.c (arm_option_override): Don't call build_target_option_node
33366 until after doing all option overrides.
33367 (arm_valid_target_attribute_tree): Likewise.
33369 2017-01-27 Martin Liska <mliska@suse.cz>
33371 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33372 for {cd}tors and C++ {cd}tors.
33374 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33376 * config/s390/s390.md ("*setmem_long_and")
33377 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33379 2017-01-26 Martin Sebor <msebor@redhat.com>
33381 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33384 2017-01-26 Martin Sebor <msebor@redhat.com>
33386 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33387 HAVE_DFmode before using XFmode or DFmode.
33388 (parse_directive): Avoid using the z length modifier to avoid
33389 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33391 PR middle-end/78703
33392 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33393 to accept adjustment as an array.
33394 (get_int_range): New function.
33395 (struct directive): Make width and prec arrays.
33396 (directive::set_width, directive::set_precision): Call get_int_range.
33397 (format_integer, format_floating): Handle width and precision ranges.
33398 (format_string, parse_directive): Same.
33400 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33403 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33404 comdat_type_p set, just clone them, but keep the children in the
33408 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33409 which have direct callers with -fvar-tracking-assignments enabled
33411 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33412 inside of type units.
33414 2017-01-26 Martin Sebor <msebor@redhat.com>
33416 PR middle-end/78703
33417 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33419 (struct format_result): Replace number_chars, number_chars_min,
33420 and number_chars_max with a single member of struct result_range.
33422 (format_result::operator+=): Adjust.
33423 (struct fmtresult): Remove bounded. Handle likely and unlikely
33425 (fmtresult::adjust_for_width_or_precision): New function.
33426 (fmtresult:type_max_digits): New function.
33427 (bytes_remaining): Handle likely and unlikely counters.
33428 (min_bytes_remaining): Remove.
33429 (format_percent): Simplify.
33430 (format_integer, format_floating): Set likely and unlikely counters.
33431 (get_string_length, format_character, format_string): Same.
33432 (format_plain, should_warn_p): New function.
33433 (maybe_warn): Call should_warn_p. Update diagnostic messages
33434 and handle those for all directives, including plain strings.
33435 (format_directive): Handle likely and unlikely counters.
33436 Remove unnecessary quoting from diagnostics. Add an informational
33438 (add_bytes): Remove.
33439 (pass_sprintf_length::compute_format_length): Simplify.
33440 (try_substitute_return_value): Handle likely and unlikely counters.
33442 2017-01-26 Carl Love <cel@us.ibm.com>
33444 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33445 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33447 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33450 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33451 endianess for subregs into account.
33452 * lra-constraints.c (lra_constraints): Do risky transformations
33453 always on the first iteration.
33454 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33455 last_call_used_reg_set.
33456 (process_bb_lives): Define and use last_call_used_reg_set.
33457 * lra.c (lra): Always continue after lra_constraints on the first
33460 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33462 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33464 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33466 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33468 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33469 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33470 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33471 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33472 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33473 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33474 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33475 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33476 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33478 2017-01-26 Marek Polacek <polacek@redhat.com>
33481 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33482 for the third operand.
33484 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33486 PR middle-end/79236
33487 * omp-low.c (struct omp_context): Add simt_stmt field.
33488 (scan_omp_for): Return omp_context *.
33489 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33490 context to the _simt_ SIMD stmt.
33491 (lower_omp_for): For combined SIMD with sibling _simt_
33492 SIMD, make sure to use the same decls in _looptemp_
33493 clauses as in the sibling.
33495 2017-01-26 David Sherwood <david.sherwood@arm.com>
33497 PR middle-end/79212
33498 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33501 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33504 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33505 emit fld b; fld a; if possible.
33507 * brig-builtins.def: Update copyright years.
33508 * config/arm/arm_acle_builtins.def: Update copyright years.
33510 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33513 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33514 constraint instead of o for the stxsd instruction.
33516 2017-01-25 Carl Love <cel@us.ibm.com>
33518 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33519 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33521 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33523 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33525 2017-01-25 Richard Biener <rguenther@suse.de>
33527 PR tree-optimization/69264
33528 * target.def (vector_alignment_reachable): Improve documentation.
33529 * doc/tm.texi: Regenerate.
33530 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33532 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33533 earlier changes with respect to TYPE_USER_ALIGN.
33534 (vector_alignment_reachable_p): Likewise. Improve dumping.
33536 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33539 * config/arm/arm.md (xordi3): Force constant operand into a register
33542 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33544 * doc/invoke.texi (-fstore-merging): Correct default optimization
33545 levels at which it is enabled.
33546 (-O): Move -fstore-merging from list to...
33549 2017-01-25 Richard Biener <rguenther@suse.de>
33552 * omp-expand.c: Include debug.h.
33553 (expand_omp_taskreg): Make sure to generate early debug before
33554 outlining anything from a function.
33555 (expand_omp_target): Likewise.
33556 (grid_expand_target_grid_body): Likewise.
33558 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33561 * asan.c (get_translation_unit_decl): New function.
33562 (asan_add_global): Extract modules file name from globals
33563 TRANSLATION_UNIT_DECL name.
33565 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33568 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33569 for long calls with APCS frame and VFP.
33571 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33573 * cfg.c (original_copy_tables_initialized_p): New function.
33574 * cfg.h (original_copy_tables_initialized_p): New decl.
33575 * cfgrtl.c (relink_block_chain): Guard the call to
33576 free_original_copy_tables with a call to
33577 original_copy_tables_initialized_p.
33578 * cgraph.h (symtab_node::native_rtl_p): New decl.
33579 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33580 (symtab_node::needed_p): Don't assert for early assembly output
33581 for __RTL functions.
33582 (cgraph_node::finalize_function): Set "force_output" for __RTL
33584 (cgraph_node::analyze): Bail out early for __RTL functions.
33585 (analyze_functions): Update assertion to support __RTL functions.
33586 (cgraph_node::expand): Bail out early for __RTL functions.
33587 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33589 * function.h (struct function): Update comment for field
33591 * gimple-expr.c: Include "tree-pass.h".
33592 (gimple_has_body_p): Return false for __RTL functions.
33593 * Makefile.in (OBJS): Add run-rtl-passes.o.
33594 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33596 (gcc::pass_manager::get_clean_slate): New accessor.
33597 * passes.c: Include "insn-addr.h".
33598 (should_skip_pass_p): Add logging. Update logic for running
33599 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33600 property-provider override so it is only done for gimple passes.
33602 (skip_pass): New function.
33603 (execute_one_pass): Call skip_pass when skipping passes.
33604 * read-md.c (md_reader::read_char): Support filtering
33605 the input to a subset of line numbers.
33606 (md_reader::md_reader): Initialize fields
33607 m_first_line and m_last_line.
33608 (md_reader::read_file_fragment): New function.
33609 * read-md.h (md_reader::read_file_fragment): New decl.
33610 (md_reader::m_first_line): New field.
33611 (md_reader::m_last_line): New field.
33612 * read-rtl-function.c (function_reader::create_function): Only
33613 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33614 curr_properties. Set DECL_INITIAL to a dummy block.
33615 (read_rtl_function_body_from_file_range): New function.
33616 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33618 * run-rtl-passes.c: New file.
33619 * run-rtl-passes.h: New file.
33621 2017-01-24 Jeff Law <law@redhat.com>
33623 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33626 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33628 PR tree-optimization/79159
33629 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33630 (record_nonwrapping_iv): Improve boundary using above function if no
33631 value range information.
33633 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33634 Martin Jambor <mjambor@suse.cz>
33636 * brig-builtins.def: New file.
33637 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33638 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33639 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33640 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33641 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33642 * builtin-types.def (BT_INT8): New.
33643 (BT_INT16): Likewise.
33644 (BT_UINT8): Likewise.
33645 (BT_UINT16): Likewise.
33646 (BT_FN_ULONG): Likewise.
33647 (BT_FN_UINT_INT): Likewise.
33648 (BT_FN_UINT_ULONG): Likewise.
33649 (BT_FN_UINT_LONG): Likewise.
33650 (BT_FN_UINT_PTR): Likewise.
33651 (BT_FN_ULONG_PTR): Likewise.
33652 (BT_FN_INT8_FLOAT): Likewise.
33653 (BT_FN_INT16_FLOAT): Likewise.
33654 (BT_FN_UINT32_FLOAT): Likewise.
33655 (BT_FN_UINT16_FLOAT): Likewise.
33656 (BT_FN_UINT8_FLOAT): Likewise.
33657 (BT_FN_UINT64_FLOAT): Likewise.
33658 (BT_FN_UINT16_UINT32): Likewise.
33659 (BT_FN_UINT32_UINT16): Likewise.
33660 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33661 (BT_FN_INT_PTR_INT): Likewise.
33662 (BT_FN_UINT_PTR_UINT): Likewise.
33663 (BT_FN_LONG_PTR_LONG): Likewise.
33664 (BT_FN_ULONG_PTR_ULONG): Likewise.
33665 (BT_FN_VOID_UINT64_UINT64): Likewise.
33666 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33667 (BT_FN_INT8_INT8_INT8): Likewise.
33668 (BT_FN_INT16_INT16_INT16): Likewise.
33669 (BT_FN_INT_INT_INT): Likewise.
33670 (BT_FN_UINT_FLOAT_UINT): Likewise.
33671 (BT_FN_FLOAT_UINT_UINT): Likewise.
33672 (BT_FN_ULONG_UINT_UINT): Likewise.
33673 (BT_FN_ULONG_UINT_PTR): Likewise.
33674 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33675 (BT_FN_UINT_UINT_UINT): Likewise.
33676 (BT_FN_VOID_UINT_PTR): Likewise.
33677 (BT_FN_UINT_UINT_PTR: Likewise.
33678 (BT_FN_UINT32_UINT64_PTR): Likewise.
33679 (BT_FN_INT_INT_UINT_UINT): Likewise.
33680 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33681 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33682 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33683 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33684 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33685 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33686 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33687 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33688 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33689 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33690 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33691 * doc/frontends.texi: List BRIG FE.
33692 * doc/install.texi (Testing): Add BRIG tesring requirements.
33693 * doc/invoke.texi (Overall Options): Mention BRIG.
33694 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33696 2017-01-24 Richard Biener <rguenther@suse.de>
33698 PR translation/79208
33699 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33701 2017-01-24 Martin Jambor <mjambor@suse.cz>
33704 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33705 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33706 and known_contexts.
33708 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33710 PR middle-end/79123
33711 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33712 casts from signed to unsigned really don't have a range.
33714 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33716 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33717 GMP_RNDx for compatiblity.
33719 2017-01-24 Martin Liska <mliska@suse.cz>
33722 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33723 that would prevent us to call alloca with -1 as argument.
33725 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33727 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33728 Avoid -Wformat-security warning.
33730 2017-01-23 Andrew Pinski <apinski@cavium.com>
33732 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33735 2017-01-23 Martin Sebor <msebor@redhat.com>
33737 PR middle-end/78703
33738 * gimple-ssa-sprintf.c (warn_level): New global.
33739 (format_integer): Use it here and throughout the rest of the file.
33740 Use the same switch to compute sign as base.
33741 (maybe_warn): New function.
33742 (format_directive): Factor out warnings into maybe_warn.
33743 Add debugging output. Use warn_level.
33744 (add_bytes): Use warn_level.
33745 (pass_sprintf_length::compute_format_length): Add debugging output.
33746 (try_substitute_return_value): Same.
33747 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33749 PR middle-end/78703
33750 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33751 (struct fmtresult, format_integer, format_floating): Adjust.
33752 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33753 (get_string_length, format_string,format_directive): Same.
33754 (pass_sprintf_length::compute_format_length): Same.
33755 (try_substitute_return_value): Simplify slightly.
33757 PR middle-end/78703
33758 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33759 (fmtresult::operator+=): Outlined.
33760 (struct fmtresult): Add ctors.
33761 (struct conversion_spec): Rename...
33762 (struct directive): ...to this. Add and remove data members.
33763 (directive::set_width, directive::set_precision): New functions.
33764 (format_percent): Use fmtresult ctor.
33765 (get_width_and_precision): Remove.
33766 (format_integer): Make naming changes. Avoid computing width and
33768 (format_floating): Same. Adjust indentation.
33769 (format_character, format_none): New functions.
33770 (format_string): Moved character handling to format_character.
33771 (format_directive): Remove arguments, change return type.
33772 (parse_directive): New function.
33773 (pass_sprintf_length::compute_format_length): Move directive
33774 parsing to parse_directive.
33776 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33778 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33779 (assign_assembler_name_if_needed): ... this.
33780 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33781 (assign_assembler_name_if_needed): ... this.
33782 (free_lang_data_in_cgraph): Adjust callers.
33783 * cgraphunit.c (cgraph_node::analyze): Likewise.
33784 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33787 2017-01-23 Richard Biener <rguenther@suse.de>
33789 PR tree-optimization/79088
33790 PR tree-optimization/79188
33791 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33792 resetting loop bounds after last path deletion. Reset loop
33793 bounds of the target loop, make code match the comments.
33794 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33795 Make sure loops need no fixups.
33797 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33799 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33800 exponent support with double type for first argument.
33801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33802 type returned by __builtin_vec_extract_sig,
33803 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33804 functions from "vector int" to "vector unsigned int" or from
33805 "vector long long int" to "vector unsigned long long int".
33806 Changed type returned by __builtin_vec_extract_exp,
33807 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33808 functions from "vector int" to "vector unsigned int" or from
33809 "vector long long int" to "vector unsigned long long int".
33810 Changed return type of __builtin_vec_test_data_class,
33811 __builtin_vec_test_data_class_sp, and
33812 __builtin_vec_test_data_class_dp from "vector int" to
33813 "vector bool int" or from "vector long long int" to "vector bool
33814 long long int" and changed second argument type from "unsigned
33815 int" to "int". Added new overloaded function forms "vector float
33816 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33817 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33818 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33819 double, vector unsigned long long int)" and "vector double
33820 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33821 long int)". Changed return type of
33822 __builtin_scalar_test_data_class and
33823 __builtin_scalar_test_data_class_sp and
33824 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33825 int" and changed second argument from "unsigned int" to "int".
33826 Changed type returned by __builtin_scalar_test_neg,
33827 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33828 from "int" to "bool int". Added new overloaded function form
33829 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33830 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33831 exponent double-precision with floating point first argument.
33832 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33833 documentation of scalar_test_data_class, scalar_test_neg,
33834 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33835 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33836 vec_test_data_class built-in functions to reflect refinements in
33837 their type signatures.
33839 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33841 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33843 (aarch64_elf_asm_destructor): Likewise.
33845 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33847 PR rtl-optimization/78634
33848 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33849 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33850 * ifcvt.c (noce_try_cmove): Add missing cost check.
33852 PR rtl-optimization/71724
33853 * combine.c (if_then_else_cond): Look for situations where it is
33854 beneficial to undo the work of one of the recursive calls.
33856 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33858 PR tree-optimization/70754
33859 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33860 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33861 combined stmt before it if not NULL.
33862 (combine_chains): Process refs reversely and compute dominance point
33865 2017-01-23 Martin Liska <mliska@suse.cz>
33867 PR tree-optimization/79196
33868 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33869 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33871 (strlen_optimize_stmt): Call the renamed function.
33873 2017-01-23 Michael Matz <matz@suse.de>
33875 PR tree-optimization/78384
33876 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33878 2017-01-23 Richard Biener <rguenther@suse.de>
33880 PR tree-optimization/79186
33881 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33882 both incoming edges before moving an assert.
33884 2017-01-23 Martin Jambor <mjambor@suse.cz>
33886 * ipa-prop.c (load_from_param_1): Removed.
33887 (load_from_unmodified_param): Bits from load_from_param_1 put back
33889 (load_from_param): Removed.
33890 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33891 with stmt. Reverted back to use of load_from_unmodified_param.
33893 2017-01-23 Martin Jambor <mjambor@suse.cz>
33896 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33897 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33898 field a pointer to garbage collected vector, mark lattices and
33899 ipcp_orig_node with GTY((skip)).
33900 (ipa_get_param_count): Adjust to descriptors being a pointer.
33901 (ipa_get_param): Likewise.
33902 (ipa_get_type): Likewise.
33903 (ipa_get_param_move_cost): Likewise.
33904 (ipa_set_param_used): Likewise.
33905 (ipa_get_controlled_uses): Likewise.
33906 (ipa_set_controlled_uses): Likewise.
33907 (ipa_is_param_used): Likewise.
33908 (ipa_node_params_t): Move into garbage collector. New methods insert
33910 (ipa_node_params_sum): Annotate wth GTY(()).
33911 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33913 (ipa_load_from_parm_agg): Adjust declaration.
33914 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33915 * ipa-profile.c (ipa_profile): Likewise.
33916 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33917 (ipa_populate_param_decls): Make descriptors parameter garbage
33919 (ipa_dump_param): Adjust to descriptors being a pointer.
33920 (ipa_alloc_node_params): Likewise.
33921 (ipa_initialize_node_params): Likewise.
33922 (load_from_param_1): Make descriptors parameter garbage collected.
33923 (load_from_unmodified_param): Likewise.
33924 (load_from_param): Likewise.
33925 (ipa_load_from_parm_agg): Likewise.
33926 (ipa_node_params::~ipa_node_params): Removed.
33927 (ipa_free_all_node_params): Remove call to delete operator.
33928 (ipa_node_params_t::insert): New.
33929 (ipa_node_params_t::remove): Likewise.
33930 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33931 copy known_csts and known_contexts vectors.
33932 (ipa_read_node_info): Adjust to descriptors being a pointer.
33933 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33935 (ipcp_transform_function): Make descriptors variable garbage
33938 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33940 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33941 * config/i386/avx512dqintrin.h: Ditto.
33942 * config/i386/avx512fintrin.h: Ditto.
33943 * config/i386/i386.c: Handle new builtins.
33944 * config/i386/i386-builtin.def: Add new builtins.
33945 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33946 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33948 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33949 Martin Liska <mliska@suse.cz>
33951 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33952 * asan.c (asan_expand_poison_ifn): Support stores and use
33953 appropriate ASAN report function.
33954 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33955 * internal-fn.def (ASAN_POISON_USE): Declare.
33956 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33957 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33958 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33959 ASAN_POISON calls w/o LHS.
33960 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33961 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33962 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33963 * gimplify.c (asan_poison_variables): Add attribute
33964 use_after_scope_memory to variables that really needs to live
33966 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33967 having the attribute.
33969 2017-01-23 Martin Liska <mliska@suse.cz>
33971 * asan.c (create_asan_shadow_var): New function.
33972 (asan_expand_poison_ifn): Likewise.
33973 * asan.h (asan_expand_poison_ifn): New declaration.
33974 * internal-fn.c (expand_ASAN_POISON): Likewise.
33975 * internal-fn.def (ASAN_POISON): New builtin.
33976 * sanopt.c (pass_sanopt::execute): Expand
33977 asan_expand_poison_ifn.
33978 * tree-inline.c (copy_decl_for_dup_finish): Make function
33980 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33981 * tree-ssa.c (is_asan_mark_p): New function.
33982 (execute_update_addresses_taken): Rewrite local variables
33983 (identified just by use-after-scope as addressable) into SSA.
33985 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33987 * doc/install.texi (Specific): opensource.apple.com uses https
33988 now. Remove trailing slash.
33990 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33992 * README.Portability: Remove note on an Irix compatibility issue.
33994 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33996 * gcov.c (INCLUDE_ALGORITHM): Define.
33997 (INCLUDE_VECTOR): Define.
33998 No longer include <vector> and <algorithm> directly.
34000 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34002 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34004 * doc/invoke.texi (Code Gen Options): Ditto.
34006 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34009 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34011 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34013 rtl-optimization/79125
34014 * cprop.c (local_cprop_pass): Handle cases where we make an
34015 unconditional trap.
34017 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34021 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34022 read from, for big endian.
34024 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34026 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34027 register pauth builtins for LP64 only.
34029 2017-01-20 Marek Polacek <polacek@redhat.com>
34032 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34035 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34037 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34039 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34040 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34041 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34043 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34046 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34047 in big-endian mode when they are not a single duplicated value.
34049 2017-01-20 Richard Biener <rguenther@suse.de>
34051 * BASE-VER: Bump to 7.0.1.
34053 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34055 * omp-low.c (omplow_simd_context): New struct. Use it...
34056 (lower_rec_simd_input_clauses): ...here and...
34057 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34058 references to idx, lane, max_vf, is_simt.
34060 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34062 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34065 2017-01-20 Martin Jambor <mjambor@suse.cz>
34067 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34068 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34070 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34071 (GTFILES): Rename hsa.c to hsa-common.c.
34072 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34073 * hsa-dump.c: Likewise.
34074 * hsa-gen.c: Likewise.
34075 * hsa-regalloc.c: Likewise.
34076 * ipa-hsa.c: Likewise.
34077 * omp-expand.c: Likewise.
34078 * omp-low.c: Likewise.
34079 * toplev.c: Likewise.
34081 2017-01-20 Marek Polacek <polacek@redhat.com>
34084 * doc/invoke.texi: Document -Wduplicated-branches.
34085 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34086 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34087 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34088 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34089 return 0 only when not OEP_LEXICOGRAPHIC.
34090 (fold_build_cleanup_point_expr): Use the expression
34091 location when building CLEANUP_POINT_EXPR.
34092 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34093 * tree.c (add_expr): Handle error_mark_node.
34095 2017-01-20 Martin Liska <mliska@suse.cz>
34098 * tree-profile.c (init_ic_make_global_vars): Do not call
34100 (gimple_init_gcov_profiler): Likewise.
34102 2017-01-20 Martin Liska <mliska@suse.cz>
34105 * cgraph.h (maybe_create_reference): Remove argument and
34107 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34109 * ipa-cp.c (create_specialized_node): Likewise.
34110 * symtab.c (symtab_node::maybe_create_reference): Handle
34111 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34113 2017-01-20 Martin Liska <mliska@suse.cz>
34115 * read-rtl-function.c (function_reader::create_function): Use
34116 build_decl instread of build_decl_stat.
34118 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34120 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34121 * config/i386/avx512dqintrin.h: Ditto.
34122 * config/i386/avx512fintrin.h: Ditto.
34123 * config/i386/i386-builtin-types.def: Add new types.
34124 * config/i386/i386.c: Handle new types.
34125 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34126 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34127 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34128 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34129 (__builtin_ia32_kshiftridi): New.
34130 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34132 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34136 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34137 define to rs6000_init_stack_protect_guard.
34138 (rs6000_init_stack_protect_guard): New function.
34140 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34141 Yunqiang Su <yunqiang.su@imgtec.com>
34143 * config.gcc (supported_defaults): Add madd4.
34144 (with_madd4): Add validation.
34145 (all_defaults): Add madd4.
34146 * config/mips/mips.opt (mmadd4): New option.
34147 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34149 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34151 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34152 (ISA_HAS_FUSED_MADD4): Likewise.
34153 * doc/invoke.texi (-mmadd4): Document the new option.
34154 * doc/install.texi (--with-madd4): Document the new option.
34156 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34158 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34159 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34160 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34161 (aarch64_init_pauth_hint_builtins): New.
34162 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34163 (aarch64_expand_builtin): Expand new builtins.
34165 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34167 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34168 * combine-stack-adj.c (no_unhandled_cfa): Handle
34169 REG_CFA_TOGGLE_RA_MANGLE.
34170 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34171 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34172 info for return address signing.
34173 (aarch64_expand_epilogue): Likewise.
34175 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34177 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34178 * config/aarch64/aarch64-protos.h
34179 (aarch64_return_address_signing_enabled): New declaration.
34180 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34182 (aarch64_expand_prologue): Sign return address before it's pushed onto
34184 (aarch64_expand_epilogue): Authenticate return address fetched from
34186 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34187 (aarch64_attributes): New function attributes for "sign-return-address".
34188 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34189 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34190 ("*do_return"): Generate combined instructions according to key index.
34191 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34192 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34194 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34195 * config/aarch64/aarch64.opt (msign-return-address=): New.
34196 * doc/extend.texi (AArch64 Function Attributes): Documents
34197 "sign-return-address=".
34198 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34200 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34202 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34203 overall option summary.
34205 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34207 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34208 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34209 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34210 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34212 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34214 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34215 -mpower9-minmax by default for -mcpu=power9.
34216 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34217 128-bit floating point.
34219 2017-01-20 Alan Modra <amodra@gmail.com>
34221 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34222 optimizing for size.
34224 2017-01-20 Alan Modra <amodra@gmail.com>
34227 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34228 for strcmp and strncmp from corresponding builtin decl.
34230 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34232 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34233 instead of i386/rtems-64.h.
34234 * config/i386/rtems-64.h: Remove.
34236 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34240 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34242 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34244 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34246 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34247 Change int to HOST_WIDE_INT.
34248 * config/aarch64/aarch64-protos.h
34249 (aarch64_simd_gen_const_vector_dup): Likewise.
34250 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34252 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34254 * langhooks-def.h (lhd_type_for_size): New decl.
34255 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34256 * langhooks.c (lhd_type_for_size): New function, taken from
34259 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34261 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34262 define_bypass for CR latency.
34263 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34264 (power9-alu2): Add define_bypass for CR latency.
34266 (power9-mul): Update insn latency.
34267 (power9-mul-compare): Update insn latency, bypass latency and remove
34270 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34272 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
34274 * config/aarch64/aarch64.md
34275 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
34276 aarch64_nopcrelative_literal_loads.
34277 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
34279 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
34281 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
34282 TARGET_LOONGSON_3A.
34283 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
34285 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
34288 * config.gcc (supported_defaults): Add lxc1-sxc1.
34289 (with_lxc1_sxc1): Add validation.
34290 (all_defaults): Add lxc1-sxc1.
34291 * config/mips/mips.opt (mlxc1-sxc1): New option.
34292 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34294 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34295 __mips_no_lxc1_sxc1.
34296 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
34297 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
34298 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
34300 2017-01-19 Richard Biener <rguenther@suse.de>
34302 PR tree-optimization/72488
34303 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34304 sure to restore SSA info.
34305 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34307 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34309 PR rtl-optimization/79121
34310 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34311 of the inner type when shifting an extended value.
34313 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34316 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34317 interposable aliases.
34319 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34322 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34323 Use the evmergelohi instruction.
34324 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34325 (mov_sitf_e500_subreg8_2_be): Likewise.
34326 (mov_sitf_e500_subreg12_2_le): Likewise.
34327 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34328 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34329 (mov_sitf_e500_subreg8_2_le): Likewise.
34330 (mov_sitf_e500_subreg12_2_be): Likewise.
34332 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34334 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34335 attribute from vecsimple to vecperm.
34336 (altivec_vbpermq2): Likewise.
34338 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34341 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34343 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34344 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34345 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34346 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34347 case where N arg is SIZE_MAX.
34348 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34349 (cmpstrsi): Add pattern.
34351 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34353 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34354 __builtin_vec_revb builtins.
34355 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34356 built-in functions to support generation of the ISA 3.0 XXBR<x>
34357 vector byte reverse instructions.
34358 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34359 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34360 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34361 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34362 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34363 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34364 (P9V_BUILTIN_VEC_REVB): Likewise.
34365 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34366 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34367 (p9_xxbrq_v16qi): Likewise.
34368 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34369 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34370 (p9_xxbrh_v8hi): Likewise.
34371 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34372 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34373 vec_revb built-in functions.
34375 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34377 PR rtl-optimization/78952
34378 * config/i386/i386.md (any_extract): New code iterator.
34379 (*insvqi_2): Use any_extract for source operand.
34380 (*insvqi_3): Use any_shiftrt for source operand.
34382 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34384 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34386 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34388 2017-01-18 Matthias Klose <doko@ubuntu.com>
34390 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34392 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34394 * config/rs6000/altivec.h (vec_bperm): Change #define.
34395 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34396 (altivec_vbpermq2): New define_insn.
34397 (altivec_vbpermd): Likewise.
34398 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34399 function interface.
34400 (VBPERMD): Likewise.
34401 (VBPERM): New polymorphic function interface.
34402 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34403 Add entries for P9V_BUILTIN_VEC_VBPERM.
34404 * doc/extend.texi: Add interfaces for vec_bperm.
34406 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34408 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34409 first letter of error messages.
34410 (s390_resolve_overloaded_builtin): Likewise.
34411 * config/s390/s390.c (s390_expand_builtin): Likewise.
34412 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34413 (s390_valid_target_attribute_inner_p): Likewise.
34414 * config/s390/s390.md ("tabort"): Likewise.
34416 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34418 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34419 (ISA_AVOID_DIV_HILO): New macro.
34420 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34421 (ISA_HAS_DDIV): Likewise.
34423 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34425 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34427 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34429 * doc/invoke.texi (fabi-version): Spelling fix.
34431 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34434 * doc/invoke.texi (fabi-version): Mention mangling fix for
34437 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34440 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34442 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34445 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34446 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34448 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34450 (stack_protect_test): Ditto.
34451 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34452 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34454 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34455 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34456 -mstack-protector-guard-offset=.
34457 (RS/6000 and PowerPC Options): Ditto.
34459 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34461 * config/i386/i386.h (MASK_CLASS_P): New define.
34462 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34463 there are no registers from different register sets also when
34464 mask registers are used. Update function comment.
34465 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34466 to (*k/*r) and (*k/*km) alternatives.
34468 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34470 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34471 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34472 (EH_RETURN_HANDLER_RTX): New define.
34473 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34474 Force frame pointer in EH return functions.
34475 (aarch64_expand_epilogue): Add barrier for eh_return.
34476 (aarch64_final_eh_return_addr): Remove.
34477 (aarch64_eh_return_handler_rtx): New function.
34478 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34480 (aarch64_eh_return_handler_rtx): New prototype.
34482 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34484 * config/rs6000/altivec.h (vec_rlmi): New #define.
34485 (vec_vrlnm): Likewise.
34486 (vec_rlnm): Likewise.
34487 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34488 (UNSPEC_VRLNM): Likewise.
34489 (VIlong): New mode iterator.
34490 (altivec_vrl<VI_char>mi): New define_insn.
34491 (altivec_vrl<VI_char>nm): Likewise.
34492 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34494 (VRLDNM): Likewise.
34495 (RLNM): New polymorphic function entry.
34496 (VRLWMI): New monomorphic function entry.
34497 (VRLDMI): Likewise.
34498 (RLMI): New polymorphic function entry.
34499 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34500 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34501 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34504 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34507 * dwarf2out.c (field_byte_offset): Restore the
34508 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34509 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34510 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34513 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34516 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34518 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34521 * dwarf2out.c (add_data_member_location_attribute): For constant
34522 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34523 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34524 DW_AT_byte_size attributes.
34526 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34528 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34529 after forcing to constant memory when the code model is medium.
34531 2017-01-17 Julia Koval <julia.koval@intel.com>
34534 * config/i386/avx512fintrin.h
34535 (_mm512_i32gather_ps): Change __addr type to void const*.
34536 (_mm512_mask_i32gather_ps): Ditto.
34537 (_mm512_i32gather_pd): Ditto.
34538 (_mm512_mask_i32gather_pd): Ditto.
34539 (_mm512_i64gather_ps): Ditto.
34540 (_mm512_mask_i64gather_ps): Ditto.
34541 (_mm512_i64gather_pd): Ditto.
34542 (_mm512_mask_i64gather_pd): Ditto.
34543 (_mm512_i32gather_epi32): Ditto.
34544 (_mm512_mask_i32gather_epi32): Ditto.
34545 (_mm512_i32gather_epi64): Ditto.
34546 (_mm512_mask_i32gather_epi64): Ditto.
34547 (_mm512_i64gather_epi32): Ditto.
34548 (_mm512_mask_i64gather_epi32): Ditto.
34549 (_mm512_i64gather_epi64): Ditto.
34550 (_mm512_mask_i64gather_epi64): Ditto.
34551 (_mm512_i32scatter_ps): Change __addr type to void*.
34552 (_mm512_mask_i32scatter_ps): Ditto.
34553 (_mm512_i32scatter_pd): Ditto.
34554 (_mm512_mask_i32scatter_pd): Ditto.
34555 (_mm512_i64scatter_ps): Ditto.
34556 (_mm512_mask_i64scatter_ps): Ditto.
34557 (_mm512_i64scatter_pd): Ditto.
34558 (_mm512_mask_i64scatter_pd): Ditto.
34559 (_mm512_i32scatter_epi32): Ditto.
34560 (_mm512_mask_i32scatter_epi32): Ditto.
34561 (_mm512_i32scatter_epi64): Ditto.
34562 (_mm512_mask_i32scatter_epi64): Ditto.
34563 (_mm512_i64scatter_epi32): Ditto.
34564 (_mm512_mask_i64scatter_epi32): Ditto.
34565 (_mm512_i64scatter_epi64): Ditto.
34566 (_mm512_mask_i64scatter_epi64): Ditto.
34567 * config/i386/avx512pfintrin.h
34568 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34569 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34570 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34571 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34572 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34573 (_mm512_prefetch_i32scatter_ps): Ditto.
34574 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34575 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34576 (_mm512_prefetch_i64scatter_pd): Ditto.
34577 (_mm512_prefetch_i64scatter_ps): Ditto.
34578 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34579 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34580 * config/i386/avx512vlintrin.h
34581 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34582 (_mm_mmask_i32gather_ps): Ditto.
34583 (_mm256_mmask_i32gather_pd): Ditto.
34584 (_mm_mmask_i32gather_pd): Ditto.
34585 (_mm256_mmask_i64gather_ps): Ditto.
34586 (_mm_mmask_i64gather_ps): Ditto.
34587 (_mm256_mmask_i64gather_pd): Ditto.
34588 (_mm_mmask_i64gather_pd): Ditto.
34589 (_mm256_mmask_i32gather_epi32): Ditto.
34590 (_mm_mmask_i32gather_epi32): Ditto.
34591 (_mm256_mmask_i32gather_epi64): Ditto.
34592 (_mm_mmask_i32gather_epi64): Ditto.
34593 (_mm256_mmask_i64gather_epi32): Ditto.
34594 (_mm_mmask_i64gather_epi32): Ditto.
34595 (_mm256_mmask_i64gather_epi64): Ditto.
34596 (_mm_mmask_i64gather_epi64): Ditto.
34597 (_mm256_i32scatter_ps): Change __addr type to void*.
34598 (_mm256_mask_i32scatter_ps): Ditto.
34599 (_mm_i32scatter_ps): Ditto.
34600 (_mm_mask_i32scatter_ps): Ditto.
34601 (_mm256_i32scatter_pd): Ditto.
34602 (_mm256_mask_i32scatter_pd): Ditto.
34603 (_mm_i32scatter_pd): Ditto.
34604 (_mm_mask_i32scatter_pd): Ditto.
34605 (_mm256_i64scatter_ps): Ditto.
34606 (_mm256_mask_i64scatter_ps): Ditto.
34607 (_mm_i64scatter_ps): Ditto.
34608 (_mm_mask_i64scatter_ps): Ditto.
34609 (_mm256_i64scatter_pd): Ditto.
34610 (_mm256_mask_i64scatter_pd): Ditto.
34611 (_mm_i64scatter_pd): Ditto.
34612 (_mm_mask_i64scatter_pd): Ditto.
34613 (_mm256_i32scatter_epi32): Ditto.
34614 (_mm256_mask_i32scatter_epi32): Ditto.
34615 (_mm_i32scatter_epi32): Ditto.
34616 (_mm_mask_i32scatter_epi32): Ditto.
34617 (_mm256_i32scatter_epi64): Ditto.
34618 (_mm256_mask_i32scatter_epi64): Ditto.
34619 (_mm_i32scatter_epi64): Ditto.
34620 (_mm_mask_i32scatter_epi64): Ditto.
34621 (_mm256_i64scatter_epi32): Ditto.
34622 (_mm256_mask_i64scatter_epi32): Ditto.
34623 (_mm_i64scatter_epi32): Ditto.
34624 (_mm_mask_i64scatter_epi32): Ditto.
34625 (_mm256_i64scatter_epi64): Ditto.
34626 (_mm256_mask_i64scatter_epi64): Ditto.
34627 (_mm_i64scatter_epi64): Ditto.
34628 (_mm_mask_i64scatter_epi64): Ditto.
34629 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34630 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34631 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34632 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34633 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34634 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34635 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34636 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34637 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34638 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34639 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34640 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34641 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34642 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34643 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34644 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34645 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34646 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34647 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34648 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34649 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34650 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34651 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34652 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34653 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34654 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34655 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34656 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34657 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34658 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34659 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34660 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34661 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34662 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34663 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34664 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34665 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34666 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34667 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34668 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34669 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34670 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34671 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34672 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34673 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34674 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34675 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34676 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34677 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34678 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34679 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34680 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34681 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34682 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34683 definitions accordingly.
34685 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34686 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34689 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34692 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34695 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34696 hard regs for inner regno.
34698 2017-01-17 Martin Liska <mliska@suse.cz>
34701 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34702 assumption and add comment.
34704 2017-01-17 Nathan Sidwell <nathan@acm.org>
34706 * ipa-visibility.c (localize_node): New function, broken out of ...
34707 (function_and_variable_visibility): ... here. Call it.
34709 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34711 PR middle-end/77445
34712 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34713 correctly set frequency of oudgoing edge.
34714 (duplicate_thread_path): Fix profile updating.
34716 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34719 * configure.ac: Add GCC_BASE_VER.
34720 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34721 version from BASE-VER file.
34722 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34723 (gcc.o): Depend on $(BASEVER).
34724 * common.opt (dumpfullversion): New option.
34725 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34726 * doc/invoke.texi: Document -dumpfullversion.
34727 * doc/install.texi: Document --with-gcc-major-version-only.
34728 * configure: Regenerated.
34730 2017-01-17 Richard Biener <rguenther@suse.de>
34732 PR tree-optimization/71433
34733 * tree-vrp.c (register_new_assert_for): Merge same asserts
34734 on all incoming edges.
34735 (process_assert_insertions_for): Handle insertions at the
34738 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34740 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34741 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34743 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34746 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34749 2017-01-17 Alan Modra <amodra@gmail.com>
34752 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34753 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34754 symbolic stack limit when pic.
34756 2017-01-16 Martin Sebor <msebor@redhat.com>
34758 PR tree-optimization/78608
34759 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34761 2017-01-16 Jeff Law <law@redhat.com>
34764 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34765 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34766 for several include directories that may be relative to sysroot.
34767 * config/i386/x-mingw32 (gplus_includedir): Define.
34768 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34769 (native_system_includedir): Likewise.
34770 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34771 override if TARGET_SYSTEM_ROOT is defined.
34772 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34774 PR tree-optimization/79090
34775 PR tree-optimization/33562
34776 PR tree-optimization/61912
34777 PR tree-optimization/77485
34778 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34779 and computed trims into the dump file.
34781 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34783 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34785 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34788 * gimplify.c (gimplify_init_constructor): If want_value and
34789 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34793 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34794 sequence. Formatting fixes.
34795 (doloop_optimize): Formatting fixes.
34798 * gcc.c (debug_level_greater_than_spec_func): New function.
34799 (static_spec_functions): Add debug-level-gt spec function.
34800 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34802 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34803 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34804 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34805 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34806 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34807 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34809 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34811 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34812 QImode fixups to general and mask registers only.
34814 2017-01-16 Carl Love <cel@us.ibm.com>
34816 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34817 for built-in functions
34818 vector signed char vec_nabs (vector signed char)
34819 vector signed short vec_nabs (vector signed short)
34820 vector signed int vec_nabs (vector signed int)
34821 vector signed long long vec_nabs (vector signed long long)
34822 vector float vec_nabs (vector float)
34823 vector double vec_nabs (vector double)
34824 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34826 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34827 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34828 * doc/extend.texi: Update the documentation file for the new built-in
34831 2017-01-16 Martin Sebor <msebor@redhat.com>
34833 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34836 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34838 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34839 UNSPEC_VSX__XXSPLTD to require special splat handling.
34841 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34844 * system.h: Poison strndup.
34846 2017-01-16 Alan Modra <amodra@gmail.com>
34849 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34852 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34854 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34856 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34858 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34859 call recog here. Assert that INSN_CODE (insn) is non-negative.
34861 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34864 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34866 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34867 in the currently scheduled RTL fragment.
34869 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34871 PR rtl-optimization/78751
34872 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34875 2017-01-14 Jeff Law <law@redhat.com>
34877 PR tree-optimization/79090
34878 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34879 variable length stores.
34880 (compute_trims): Delete dead assignment to *trim_tail.
34881 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34884 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34886 PR rtl-optimization/78626
34887 PR rtl-optimization/78727
34888 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34889 of a block, and split such blocks after everything else is finished.
34891 2017-01-14 Alan Modra <amodra@gmail.com>
34894 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34895 target legitimate_combined_insn.
34896 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34897 (rs6000_legitimate_combined_insn): New function.
34898 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34900 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34901 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34902 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34904 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34906 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34908 2017-01-13 Jeff Law <law@redhat.com>
34910 PR tree-optimization/33562
34911 PR tree-optimization/61912
34912 PR tree-optimization/77485
34913 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34915 (delete_dead_assignment): Likewise.
34916 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34917 statement to delete_dead_call and delete_dead_assignment.
34919 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34922 * substring-locations.c (format_warning_va): Strengthen case 1 so
34923 that both endpoints of the substring must be within the format
34924 range for just the substring to be printed.
34926 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34928 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34929 * config/i386/i386.c (ix86_target_string): Add missing options
34930 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34931 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34932 flags_other and ix86_target_other to flags2_other. Display unknown
34934 (ix86_valid_target_attribute_inner_p): Add missing options and
34935 reorder options by implied ISAs, as in ix86_target_string.
34937 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34939 * hash-table.h (hash_table::too_empty_p): New function.
34940 (hash_table::expand): Use it.
34941 (hash_table::traverse): Likewise.
34942 (hash_table::empty_slot): Use sizeof (value_type) instead of
34943 sizeof (PTR) to convert bytes to elements. Shrink the table
34944 if the current size is excessive for the current number of
34947 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34949 * ira-costs.c (record_reg_classes): Break from the inner loop
34950 early once alt_fail is known to be true. Update outer loop
34951 handling accordingly.
34953 2017-01-13 Jeff Law <law@redhat.com>
34955 * tree-ssa-dse.c (decrement_count): New function.
34956 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34957 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34958 when we know the partially dead statement is a mem* function.
34960 PR tree-optimization/61912
34961 PR tree-optimization/77485
34962 * tree-ssa-dse.c: Include expr.h.
34963 (maybe_trim_constructor_store): New function.
34964 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34966 PR tree-optimization/33562
34967 PR tree-optimization/61912
34968 PR tree-optimization/77485
34969 * doc/invoke.texi: Document new dse-max-object-size param.
34970 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34971 * tree-ssa-dse.c: Include params.h.
34972 (dse_store_status): New enum.
34973 (initialize_ao_ref_for_dse): New, partially extracted from
34975 (valid_ao_ref_for_dse, normalize_ref): New.
34976 (setup_live_bytes_from_ref, compute_trims): Likewise.
34977 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34978 (maybe_trim_partially_dead_store): Likewise.
34979 (maybe_trim_complex_store): Likewise.
34980 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34981 Track what bytes live from the original store. Return tri-state
34982 for dead, partially dead or live.
34983 (dse_dom_walker): Add constructor, destructor and new private members.
34984 (delete_dead_call, delete_dead_assignment): New extracted from
34986 (dse_optimize_stmt): Make a member of dse_dom_walker.
34987 Use initialize_ao_ref_for_dse.
34989 PR tree-optimization/33562
34990 PR tree-optimization/61912
34991 PR tree-optimization/77485
34992 * sbitmap.h (bitmap_count_bits): Prototype.
34993 (bitmap_clear_range, bitmap_set_range): Likewise.
34994 * sbitmap.c (bitmap_clear_range): New function.
34995 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34997 2017-01-13 Martin Liska <mliska@suse.cz>
35000 * function.c (set_cfun): Add new argument force.
35001 * function.h (set_cfun): Likewise.
35002 * ipa-inline-transform.c (inline_call): Use the function when
35003 strict alising from is dropped for function we inline to.
35005 2017-01-13 Richard Biener <rguenther@suse.de>
35007 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35008 for dumping GIMPLE INTEGER_CSTs.
35010 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35012 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35013 to 201112L since C++17.
35015 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35018 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35019 if -fsanitize=kernel-address is present.
35021 2017-01-13 Richard Biener <rguenther@suse.de>
35023 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35024 as _Literal ( type ) number in case usual suffixes do not
35025 preserve all information.
35027 2017-01-13 Richard Biener <rguenther@suse.de>
35029 PR tree-optimization/77283
35030 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35031 and ssa-iterators.h.
35032 (is_feasible_trace): Implement a cost model based on joiner
35035 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35038 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35039 char or short to __float128/_Float128 directly.
35041 2017-01-12 Martin Sebor <msebor@redhat.com>
35043 to -Wformat-overflow.
35044 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35045 (min_bytes_remaining): Same.
35046 (get_string_length): Same.
35047 (format_string): Same.
35048 (format_directive): Same.
35050 (pass_sprintf_length::handle_gimple_call): Same.
35052 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35054 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35055 info.nowrite calls with no lhs that can't throw. Return bool
35056 whether gsi_remove has been called or not.
35057 (pass_sprintf_length::handle_gimple_call): Return bool whether
35058 try_substitute_return_value called gsi_remove. Formatting fix.
35059 (pass_sprintf_length::execute): Don't use gsi_remove if
35060 handle_gimple_call returned true.
35063 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35064 be removed due to side-effects, don't remove following barrier nor
35065 turn the successor edge into fallthru edge.
35067 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35070 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35071 element-reversing loads and stores as not swappable.
35073 2017-01-12 Nathan Sidwell <nathan@acm.org>
35074 Nicolai Stange <nicstange@gmail.com>
35076 * combine.c (try_combine): Don't ignore result of overlap checking
35077 loop. Combine overlap & asm check into single loop.
35079 2017-01-12 Richard Biener <rguenther@suse.de>
35081 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35082 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35084 2017-01-12 Richard Biener <rguenther@suse.de>
35086 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35087 and TS_TARGET_OPTION directly derive from TS_BASE.
35088 * tree-core.h (tree_optimization_option): Derive from tree_base.
35089 (tree_target_option): Likewise.
35091 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35093 * config/i386/i386.c (memory_address_length): Increase len
35094 only when rip_relative_addr_p returns false.
35096 2017-01-11 Julia Koval <julia.koval@intel.com>
35098 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35099 (OPTION_MASK_ISA_SGX_SET): New.
35100 (ix86_handle_option): Handle OPT_msgx.
35101 * config.gcc: Added sgxintrin.h.
35102 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35103 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35104 * config/i386/i386.c (ix86_target_string): Add -msgx.
35106 (ix86_option_override_internal): Handle new options.
35107 (ix86_valid_target_attribute_inner_p): Add sgx.
35108 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35109 * config/i386/i386.opt: Add msgx.
35110 * config/i386/sgxintrin.h: New file.
35111 * config/i386/x86intrin.h: Add sgxintrin.h.
35113 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35116 * fold-const.c (maybe_nonzero_address): Return 1 for function
35118 (tree_single_nonzero_warnv_p): Don't handle function local objects
35122 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35125 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35128 * opts.c: Include "spellcheck.h"
35129 (struct string_fragment): New struct.
35130 (struct edit_distance_traits<const string_fragment &>): New
35132 (get_closest_sanitizer_option): New function.
35133 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35135 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35137 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35139 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35140 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35141 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35142 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35143 for initial die_offset if dwarf_split_debug_info.
35144 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35145 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35147 (output_skeleton_debug_sections): Formatting fix. Use
35148 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35149 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35151 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35153 * config/arm/cortex-a53.md: Add bypasses for
35154 cortex_a53_r2f_cvt.
35155 (cortex_a53_r2f): Only use for transfers.
35156 (cortex_a53_f2r): Likewise.
35157 (cortex_a53_r2f_cvt): Add reservation for conversions.
35158 (cortex_a53_f2r_cvt): Likewise.
35160 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35162 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35163 to all inlined functions, change static to extern.
35165 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35168 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35170 (arm_assemble_integer): Likewise.
35172 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35174 * config.gcc: Use new awk script to check CPU, FPU and architecture
35175 parameters for --with-... options.
35176 * config/arm/parsecpu.awk: New file
35177 * config/arm/arm-cpus.in: New file.
35178 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35180 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35182 * config/arm/t-arm: Update dependency rules.
35183 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35184 of processing .def files.
35185 * config/arm/genopt.sh: Deleted.
35186 * config/arm/gentune.sh: Deleted.
35187 * config/arm/arm-cores.def: Deleted.
35188 * config/arm/arm-arches.def: Deleted.
35189 * config/arm/arm-fpus.def: Deleted.
35190 * config/arm/arm-tune.md: Regenerated.
35191 * config/arm/arm-tables.opt: Regenerated.
35192 * config/arm/arm-cpu.h: New generated file.
35193 * config/arm/arm-cpu-data.h: New generated file.
35194 * config/arm/arm-cpu-cdata.h: New generated file.
35196 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35199 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35201 (input_varpool_node): Unpack dynamically_initialized bit.
35203 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35205 PR rtl-optimization/79032
35206 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35207 the alignment of the adjusted memory reference against that of MODE,
35208 instead of the alignment of the original memory reference.
35210 2017-01-11 Martin Jambor <mjambor@suse.cz>
35212 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35214 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35215 decorated functions.
35217 2017-01-11 Richard Biener <rguenther@suse.de>
35219 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35220 set range/nonnull info for PHI results. Do not set it on
35221 stmts marked for removal.
35223 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35225 * expr.c (store_field): In the bitfield case, fetch the return value
35226 from the registers before applying a single big-endian adjustment.
35227 Always do a final load for a BLKmode value not larger than a word.
35229 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35232 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35233 that we correctly handle column numbers greater than
35234 LINE_MAP_MAX_COLUMN_NUMBER.
35236 2017-01-10 Martin Sebor <msebor@redhat.com>
35238 PR middle-end/78245
35239 * gimple-ssa-sprintf.c (get_destination_size): Call
35240 {init,fini}object_sizes.
35241 * tree-object-size.c (addr_object_size): Adjust.
35242 (pass_through_call): Adjust.
35243 (pass_object_sizes::execute): Adjust.
35244 * tree-object-size.h (fini_object_sizes): Declare.
35246 2017-01-10 Martin Sebor <msebor@redhat.com>
35248 PR tree-optimization/78775
35249 * builtins.c (get_size_range): Move...
35250 * calls.c: ...to here.
35251 (alloc_max_size): Accept zero argument.
35252 (operand_signed_p): Remove.
35253 (maybe_warn_alloc_args_overflow): Call get_size_range.
35254 * calls.h (get_size_range): Declare.
35256 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35258 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35259 from TI's devices.csv file as of September 2016.
35260 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35262 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35264 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35265 * doc/invoke.texi: Likewise.
35266 * doc/md.texi: Likewise.
35267 * doc/objc.texi: Likewise.
35269 2017-01-10 Joshua Conner <joshconner@google.com>
35271 * config/arm/fuchsia-elf.h: New file.
35272 * config/fuchsia.h: New file.
35273 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
35274 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
35276 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
35278 2016-01-10 Richard Biener <rguenther@suse.de>
35280 PR tree-optimization/79034
35281 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35282 Propagate out degenerate PHIs in the joiner.
35284 2017-01-10 Martin Liska <mliska@suse.cz>
35286 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
35287 (sort_congruence_classes_by_decl_uid): Likewise.
35288 (sort_congruence_class_groups_by_decl_uid): Likewise.
35289 (sem_item_optimizer::merge_classes): Sort class, groups in these
35290 classes and members in the groups by DECL_UID of declarations.
35291 This would make merge operations stable.
35293 2017-01-10 Martin Liska <mliska@suse.cz>
35295 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
35296 usage of m_classes_vec.
35297 (sem_item_optimizer::~sem_item_optimizer): Likewise.
35298 (sem_item_optimizer::get_group_by_hash): Likewise.
35299 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
35300 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
35301 (sem_item_optimizer::verify_classes): Likewise.
35302 (sem_item_optimizer::process_cong_reduction): Likewise.
35303 (sem_item_optimizer::dump_cong_classes): Likewise.
35304 (sem_item_optimizer::merge_classes): Likewise.
35305 * ipa-icf.h (congruence_class_hash): Rename from
35306 congruence_class_group_hash. Remove declaration of m_classes_vec.
35308 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35310 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35311 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35312 * config.gcc: Add avx512vpopcntdqintrin.h.
35313 * config/i386/avx512vpopcntdqintrin.h: New.
35314 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35315 * config/i386/i386-builtin-types.def: Add new types.
35316 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35317 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35318 __builtin_ia32_vpopcountq_v8di_mask): New.
35319 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35320 __AVX512VPOPCNTDQ__.
35321 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35322 (PTA_AVX512VPOPCNTDQ): Define.
35323 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35324 TARGET_AVX512VPOPCNTDQ_P): Define.
35325 * config/i386/i386.opt: Add mavx512vpopcntdq.
35326 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35327 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35329 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35331 PR middle-end/77484
35332 * predict.def (PRED_CALL): Set to 67.
35334 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35336 * expr.c (store_field): In the bitfield case, if the value comes from
35337 a function call and is of an aggregate type returned in registers, do
35338 not modify the field mode; extract the value in all cases if the mode
35339 is BLKmode and the size is not larger than a word.
35341 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35344 * config/i386/cpuid.h: Fix undefined behavior.
35346 2017-01-04 Jeff Law <law@redhat.com>
35348 PR tree-optimization/79007
35349 PR tree-optimization/67955
35350 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35351 conservative for pt.null when flag_non_call_exceptions is on.
35353 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35355 PR translation/79019
35356 PR translation/79020
35357 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35358 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35359 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35361 * config/avr/avr.opt (maccumulate-args): Likewise.
35362 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35363 * common.opt (freport-bug): Likewise.
35364 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35365 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35366 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35367 translatable string.
35368 * config/i386/i386.c (function_value_32): Likewise.
35369 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35370 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35372 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35373 * common/config/msp430/msp430-common.c (msp430_handle_option):
35375 * symtab.c (symtab_node::verify_base): Likewise.
35376 * opts.c (set_debug_level): Likewise.
35377 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35378 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35379 missing whitespace to translatable strings.
35380 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35381 * config/sh/superh.h: Likewise.
35382 * config/i386/xopintrin.h: Likewise.
35383 * config/i386/znver1.md: Likewise.
35384 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35385 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35386 * double-int.h (struct double_int): Likewise.
35387 * double-int.c (div_and_round_double): Likewise.
35388 * wide-int.cc: Likewise.
35389 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35390 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35391 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35392 (crossjumps_occurred): ... this.
35393 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35396 PR tree-optimization/78899
35397 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35398 returning bool return struct loop *, NULL for failure and the new
35400 (versionable_outer_loop_p): Don't version outer loop if it has
35401 dont_vectorized bit set.
35402 (tree_if_conversion): When versioning outer loop, ensure
35403 tree_if_conversion is performed also on the inner loop of the
35404 non-vectorizable outer loop copy.
35405 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35406 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35407 prevent vectorization of it.
35408 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35409 the outer loop vectorization of the non-scalar version is attempted
35410 before vectorization of the inner loop in scalar version. If
35411 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35412 vectorization of its inner loop.
35413 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35414 has 2 inner loops, rename also on edges from bb whose single pred
35415 is outer_loop->header. Fix typo in function comment.
35417 2017-01-09 Martin Sebor <msebor@redhat.com>
35420 * asan.c (asan_emit_stack_protection): Increase local buffer size
35421 to avoid snprintf truncation warning.
35423 2017-01-09 Andrew Pinski <apinski@cavium.com>
35425 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35426 to reference thunderx2t99 for the tuning structure
35427 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35429 (thunderx2t99_extra_costs): This.
35430 * config/aarch64/aarch64-tune.md: Regenerate.
35431 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35432 (vulcan_addrcost_table): This.
35433 (vulcan_regmove_cost): Rename to ...
35434 (thunderx2t99_regmove_cost): This.
35435 (vulcan_vector_cost): Rename to ...
35436 (thunderx2t99_vector_cost): this.
35437 (vulcan_branch_cost): Rename to ...
35438 (thunderx2t99_branch_cost): This.
35439 (vulcan_tunings): Rename to ...
35440 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35441 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35443 2017-01-09 Martin Jambor <mjambor@suse.cz>
35447 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35448 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35449 (propagate_vr_accross_jump_function): Use the above function for all
35450 value range computations for pass-through jump functions and type
35451 converasion from explicit value range values.
35452 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35453 parameters from TYPE_ARG_TYPES.
35454 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35455 (ipa_write_node_info): Stream type of the actual argument.
35456 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35458 2017-01-09 Martin Liska <mliska@suse.cz>
35461 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35462 (lookup_compiler): Do not show error message with have_E.
35464 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35466 PR tree-optimization/78938
35467 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35468 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35469 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35470 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35473 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35475 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35478 2017-01-09 Richard Biener <rguenther@suse.de>
35480 PR tree-optimization/78997
35481 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35482 name condition properly.
35484 2017-01-09 Richard Biener <rguenther@suse.de>
35487 * dwarf2out.c (is_cxx): New overload with context.
35488 (is_naming_typedef_decl): Use it.
35490 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35492 * invoke.texi (Option Summary): Correct spacing in option lists
35493 and add line breaks to fix over-long lines.
35495 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35497 PR middle-end/17660
35499 * extend.texi (Common Variable Attributes): Add xref to GCC
35500 Internals manual to explain mode attribute keywords.
35502 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35505 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35506 and Preprocessor Options.
35507 (Options for Linking): Document -pthread here....
35508 (RS/6000 and PowerPC Options): ...not here.
35509 (Solaris 2 Options): ...or here.
35510 * doc/cppopts.texi: Document -pthread.
35512 2017-01-08 Martin Sebor <msebor@redhat.com>
35514 PR middle-end/77708
35515 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35516 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35517 New member functions.
35518 (format_directive): Used them.
35520 (pass_sprintf_length::handle_gimple_call): Same.
35521 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35522 to avoid truncation for any argument.
35523 (extract_affine_mul): Same.
35524 * tree.c (get_file_function_name): Same.
35526 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35528 PR middle-end/77484
35529 * predict.def (PRED_INDIR_CALL): Set to 86.
35531 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35533 PR preprocessor/54124
35534 * doc/cppopts.texi: Reformat -d subtable to list the full name
35535 of the options. Add cross-reference to the docs for the general
35536 compiler -d options.
35537 * doc/invoke.texi (Developer Options): Add cross-reference to the
35538 preprocessor-specific -d option documentation.
35540 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35542 PR preprocessor/13498
35543 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35544 redudant material, and reflect new command-line options.
35545 (System Headers): Likewise.
35547 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35549 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35550 -isystem, and -idirafter. Copy-edit.
35551 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35552 default for -ftrack-macro-expansion. Delete obsolete and
35553 badly-formatted implementation details about -fdebug-cpp output.
35554 * doc/cppwarnopts.texi: Copy-edit.
35556 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35559 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35560 that the transition from a max line width >= 1<<10 to narrower
35561 lines works correctly.
35563 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35565 * doc/options.texi (PerFunction): New.
35566 * opt-functions.awk (switch_flags): Map both Optimization and
35567 PerFunction to CL_OPTIMIZATION.
35568 * opth-gen.awk: Test for PerFunction flag along with
35570 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35571 it only when the latter is present. Skip those that don't in
35572 the hash function generator.
35573 * common.opt (fvar-tracking): Mark as PerFunction instead of
35575 (fvar-tracking-assignments): Likewise.
35576 (fvar-tracking-assignments-toggle): Likewise.
35577 (fvar-tracking-uninit): Likewise.
35579 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35581 PR translation/79018
35582 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35585 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35588 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35589 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35590 TARGET_LONG_JUMP_TABLE_OFFSETS.
35591 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35592 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35593 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35594 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35595 * config/m68k/m68k.md (tablejump expander): Likewise.
35596 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35597 TARGET_LONG_JUMP_TABLE_OFFSETS.
35598 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35599 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35601 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35602 David Holsgrove <david.holsgrove@xilinx.com>
35604 * common/config/microblaze/microblaze-common.c
35605 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35606 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35608 * config/microblaze/microblaze.c (microblaze_must_save_register)
35609 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35611 (microblaze_eh_return): New function.
35612 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35613 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35614 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35615 * config/microblaze/microblaze.md (eh_return): New pattern.
35617 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35619 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35620 GCC_DIAGNOSTIC_STRINGIFY): Define.
35622 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35624 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35626 * config/arm/arm.md (<mcrr>): New.
35628 * config/arm/arm.c (arm_arch5te): New.
35629 (arm_option_override): Set arm_arch5te.
35630 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35632 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35633 (arm_mcrr_qualifiers): ... this. New.
35634 (MRRC_QUALIFIERS): Define to...
35635 (arm_mrrc_qualifiers): ... this. New.
35636 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35638 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35639 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35640 (MRRCI, mrrc, MRRC): New.
35641 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35642 VUNSPEC_MRRC2): New.
35644 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35646 * config/arm/arm.md (<mcr>): New.
35648 * config/arm/arm.c (arm_coproc_builtin_available): Add
35649 support for mcr, mrc, mcr2 and mrc2.
35650 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35651 (arm_mcr_qualifiers): ... this. New.
35652 (MRC_QUALIFIERS): Define to ...
35653 (arm_mrc_qualifiers): ... this. New.
35654 (MCR_QUALIFIERS): Define to ...
35655 (arm_mcr_qualifiers): ... this. New.
35656 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35658 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35659 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35660 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35661 VUNSPEC_MRC2): New.
35663 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35665 * config/arm/arm.md (*ldc): New.
35669 * config/arm/arm.c (arm_coproc_builtin_available): Add
35670 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35671 (arm_coproc_ldc_stc_legitimate_address): New.
35672 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35673 'qualifier_const_pointer'.
35674 (LDC_QUALIFIERS): Define to...
35675 (arm_ldc_qualifiers): ... this. New.
35676 (STC_QUALIFIERS): Define to...
35677 (arm_stc_qualifiers): ... this. New.
35678 * config/arm/arm-protos.h
35679 (arm_coproc_ldc_stc_legitimate_address): New.
35680 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35681 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35682 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35683 stc2, stcl, stc2l): New.
35684 * config/arm/constraints.md (Uz): New.
35685 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35686 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35687 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35688 VUNSPEC_STC2L): New.
35690 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35692 * config/arm/arm.md (<cdp>): New.
35693 * config/arm/arm.c (neon_const_bounds): Rename this ...
35694 (arm_const_bounds): ... this.
35695 (arm_coproc_builtin_available): New.
35696 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35697 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35698 (CDP_QUALIFIERS): Define to...
35699 (arm_cdp_qualifiers): ... this. New.
35701 (arm_expand_builtin_args): Add case for 6 arguments.
35702 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35703 (arm_const_bounds): ... this.
35704 (arm_coproc_builtin_available): New.
35705 * config/arm/arm_acle.h (__arm_cdp): New.
35707 * config/arm/arm_acle_builtins.def (cdp): New.
35709 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35710 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35711 'arm_const_bounds'.
35712 * config/arm/types.md (coproc): New.
35713 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35714 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35715 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35716 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35718 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35720 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35721 (UBINOP_QUALIFIERS): New.
35723 (acle_builtin_data): New. Change comment.
35724 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35725 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35726 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35727 arm_acle_builtins.def.
35728 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35729 (arm_init_acle_builtins): New.
35730 (CRC32_BUILTIN): Remove.
35731 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35732 crc32cb, crc32ch and crc32cw.
35733 (arm_init_crc32_builtins): Remove.
35734 (arm_init_builtins): Use arm_init_acle_builtins rather
35735 than arm_init_crc32_builtins.
35736 (arm_expand_acle_builtin): New.
35737 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35738 * config/arm/arm_acle_builtins.def: New.
35740 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35742 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35743 (arm_builtin_datum): ... this.
35744 (arm_init_neon_builtin): Rename to ...
35745 (arm_init_builtin): ... this. Add a new parameters PREFIX
35746 and USE_SIG_IN_NAME.
35747 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35748 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35749 'arm_builtin_datum'.
35750 (arm_init_vfp_builtins): Likewise.
35751 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35752 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35753 (arm_expand_neon_args): Rename to ...
35754 (arm_expand_builtin_args): ... this. Rename builtin_arg
35755 enum values and differentiate between ARG_BUILTIN_MEMORY
35756 and ARG_BUILTIN_NEON_MEMORY.
35757 (arm_expand_neon_builtin_1): Rename to ...
35758 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35759 values, arm_expand_builtin_args and add bool parameter NEON.
35760 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35761 (arm_expand_vfp_builtin): Likewise.
35762 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35764 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35766 PR middle-end/77484
35767 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35768 * predict.c (tree_estimate_probability_bb): Reverse direction of
35769 polymorphic call predictor.
35771 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35773 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35774 (determine_pass_name_match): ...this new function and...
35775 (should_skip_pass_p): ...this new function.
35777 2017-01-06 Nathan Sidwell <nathan@acm.org>
35779 * ipa-visibility.c (function_and_variable_visibility): Reformat
35780 comments and long lines. Remove extrneous if.
35781 * symtab.c (symtab_node::make_decl_local): Fix code format.
35782 (symtab_node::set_section_for_node): Fix comment typo.
35784 2017-01-06 Martin Liska <mliska@suse.cz>
35787 * lra-constraints.c: Rename invariant to lra_invariant.
35788 * predict.c (set_even_probabilities): Initialize e to NULL.
35790 2017-01-05 Martin Sebor <msebor@redhat.com>
35792 PR tree-optimization/78910
35793 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35794 (format_integer): Correct off-by-one error in the handling
35795 of precision with negative numbers in signed conversions..
35797 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35799 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35801 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35803 PR tree-optimization/71016
35804 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35805 factor_out_conditional_conversion. Formatting fix.
35806 (factor_out_conditional_conversion): Add cond_stmt argument.
35807 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35808 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35811 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35813 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35814 read-rtl-function.o, and selftest-rtl.o.
35815 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35816 (selftest::aarch64_test_loading_full_dump): New function.
35817 (selftest::aarch64_run_selftests): New function.
35818 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35819 selftest::aarch64_run_selftests.
35820 * config/i386/i386.c
35821 (selftest::ix86_test_loading_dump_fragment_1): New function.
35822 (selftest::ix86_test_loading_call_insn): New function.
35823 (selftest::ix86_test_loading_full_dump): New function.
35824 (selftest::ix86_test_loading_unspec): New function.
35825 (selftest::ix86_run_selftests): Call the new functions.
35826 * emit-rtl.c (maybe_set_max_label_num): New function.
35827 * emit-rtl.h (maybe_set_max_label_num): New decl.
35828 * function.c (instantiate_decls): Guard call to
35829 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35830 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35832 * gensupport.c (gen_reader::gen_reader): Pass "false"
35833 for new "compact" param of rtx_reader.
35834 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35835 rather than an empty string for NULL strings.
35836 * read-md.c: Potentially include config.h rather than bconfig.h.
35837 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35838 (have_error): New global, copied from errors.c.
35839 (md_reader::read_name): Rename to...
35840 (md_reader::read_name_1): ...this, adding "out_loc" param,
35841 and converting "missing name or number" to returning false, rather
35843 (md_reader::read_name): Reimplement in terms of read_name_1.
35844 (md_reader::read_name_or_nil): New function.
35845 (md_reader::read_string): Handle "(nil)" by returning NULL.
35846 (md_reader::md_reader): Add new param "compact".
35847 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35848 (md_reader::read_file): New method.
35849 * read-md.h (md_reader::md_reader): Add new param "compact".
35850 (md_reader::read_file): New method.
35851 (md_reader::is_compact): New accessor.
35852 (md_reader::read_name): Convert return type from void to file_location.
35853 (md_reader::read_name_or_nil): New decl.
35854 (md_reader::read_name_1): New decl.
35855 (md_reader::m_compact): New field.
35856 (noop_reader::noop_reader): Pass "false" for new "compact" param
35858 (rtx_reader::rtx_reader): Add new "compact" param.
35859 (rtx_reader::read_rtx_operand): Make virtual and convert return
35860 type from void to rtx.
35861 (rtx_reader::read_until): New decl.
35862 (rtx_reader::handle_any_trailing_information): New virtual function.
35863 (rtx_reader::postprocess): New virtual function.
35864 (rtx_reader::finalize_string): New virtual function.
35865 (rtx_reader::m_in_call_function_usage): New field.
35866 (rtx_reader::m_reuse_rtx_by_id): New field.
35867 * read-rtl-function.c: New file.
35868 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35869 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35870 (selftest::verify_three_block_rtl_cfg): New decl.
35871 * read-rtl-function.h: New file.
35872 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35873 For host, include function.h, memmodel.h, and emit-rtl.h.
35874 (one_time_initialization): New function.
35875 (struct compact_insn_name): New struct.
35876 (compact_insn_names): New array.
35877 (find_code): Handle insn codes in compact dumps.
35878 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35879 (bind_subst_iter_and_attr): Likewise.
35880 (add_condition_to_string): Likewise.
35881 (add_condition_to_rtx): Likewise.
35882 (apply_attribute_uses): Likewise.
35883 (add_current_iterators): Likewise.
35884 (apply_iterators): Likewise.
35885 (initialize_iterators): Guard usage of apply_subst_iterator with
35886 #ifdef GENERATOR_FILE.
35887 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35888 (md_reader::read_mapping): Likewise.
35889 (add_define_attr_for_define_subst): Likewise.
35890 (add_define_subst_attr): Likewise.
35891 (read_subst_mapping): Likewise.
35892 (check_code_iterator): Likewise.
35893 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35895 (one_time_initialization): New function.
35896 (rtx_reader::read_until): New method.
35897 (read_flags): New function.
35898 (parse_reg_note_name): New function.
35899 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35900 Handle reuse_rtx ids.
35901 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35902 Add parsing support for RTL dumps, mirroring the special-cases in
35903 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35904 values, and calling handle_any_trailing_information.
35905 (rtx_reader::read_rtx_operand): Convert return type from void
35906 to rtx, returning return_rtx. Handle case 'e'. Call
35907 finalize_string on XSTR and XTMPL fields.
35908 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35909 "(nil)" values were omitted. Call the postprocess vfunc on the
35911 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35912 class ctor. Initialize m_in_call_function_usage. Call
35913 one_time_initialization.
35914 * rtl-tests.c (selftest::test_uncond_jump): Call
35915 set_new_first_and_last_insn.
35916 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35917 * selftest-rtl.c: New file.
35918 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35919 (selftest::get_insn_by_uid): New decl.
35920 * selftest-run-tests.c (selftest::run_tests): Call
35921 read_rtl_function_c_tests.
35922 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35923 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35926 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35928 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35929 operands in a special way. Assert that pos+len <= mode precision.
35931 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35933 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35934 3 argument Alias with unlimited for the negative form.
35935 (fno-vect-cost-model): Removed.
35937 2017-01-05 Martin Liska <mliska@suse.cz>
35939 * hsa-gen.c (gen_hsa_divmod): New function.
35940 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35942 2017-01-05 Martin Liska <mliska@suse.cz>
35945 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35948 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35950 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35951 small constant length operands.
35953 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35955 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35956 between loop iterations.
35958 2017-01-05 Martin Liska <mliska@suse.cz>
35961 * gimplify.c (gimplify_decl_expr): Compare to
35962 asan_poisoned_variables instread of checking flags.
35963 (gimplify_target_expr): Likewise.
35964 (gimplify_expr): Likewise.
35965 (gimplify_function_tree): Conditionally initialize
35966 asan_poisoned_variables.
35968 2017-01-04 Jeff Law <law@redhat.com>
35970 PR tree-optimizatin/78812
35971 * rtl.h (contains_mem_rtx_p): Prototype.
35972 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35973 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35974 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35975 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35976 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35978 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35980 * input.c (assert_char_at_range): Default-initialize actual_range.
35982 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35984 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35985 to match the caller.
35987 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35989 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35990 insns after final jump in test to emit dummy move.
35992 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35994 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35995 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35997 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35999 * multiple_target.c (create_dispatcher_calls): Init e_next.
36000 * tree-ssa-loop-split.c (split_loop): Init border.
36001 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36004 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36009 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36010 (altivec_register_operand): Do not return true if the operand
36011 contains a SUBREG mixing SImode and SFmode.
36012 (vsx_register_operand): Likewise.
36013 (vsx_reg_sfsubreg_ok): New predicate.
36014 (vfloat_operand): Do not return true if the operand contains a
36015 SUBREG mixing SImode and SFmode.
36016 (vint_operand): Likewise.
36017 (vlogical_operand): Likewise.
36018 (gpc_reg_operand): Likewise.
36019 (int_reg_operand): Likewise.
36020 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36021 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36022 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36024 (rs6000_emit_move_si_sf_subreg): New helper function.
36025 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36026 fixup SUBREGs involving SImode and SFmode.
36027 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36028 numbers for the new peephole2 optimization.
36029 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36030 the GLIBC math library that do AND/IOR/XOR operations on single
36031 precision floating point.
36032 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36033 target macros to say whether we need to avoid SUBREGs mixing
36035 (TARGET_ALLOW_SF_SUBREG): Likewise.
36036 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36037 (UNSPEC_SI_FROM_SF): Likewise.
36038 (iorxor): Change spacing.
36039 (and_ior_xor): New iterator for AND, IOR, and XOR.
36040 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36041 (movdi_from_sf_zero_ext): Likewise.
36042 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36043 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36044 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36045 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36046 (fms<mode>4): Likewise.
36047 (fnma<mode>4): Likewise.
36048 (fnms<mode>4): Likewise.
36049 (nfma<mode>4): Likewise.
36050 (nfms<mode>4): Likewise.
36052 2017-01-04 Marek Polacek <polacek@redhat.com>
36055 * doc/invoke.texi: Document -Wpointer-compare.
36057 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36059 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36062 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36063 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36066 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36069 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36070 documentation of the powerpc_popcntb_ok attribute.
36071 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36072 code to issue warning messages if a requested CPU configuration is
36073 not supported by the binary (assembler and loader) toolchain.
36074 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36075 made to define a built-in function that has been disabled.
36076 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36077 made to define a built-in function that has been disabled.
36078 (altivec_init_builtins): Add comment explaining why definition
36079 of the DST built-in functions is not preceded by an assertion
36080 check. Add assertions to prevent ICE if attempts are made to
36081 define an altivec predicate or an abs* built-in function that has
36083 (htm_init_builtins): Add comment explaining why definition of the
36084 htm built-in functions is not preceded by an assertion check.
36086 2017-01-04 Jeff Law <law@redhat.com>
36088 PR tree-optimizatin/67955
36089 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36090 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36091 the points-to solution does not include pt_null. Use DECL_PT_UID
36094 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36096 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36097 Use gen_int_mode instead of gen_lopwart for const_int operands.
36099 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36101 PR tree-optimization/71563
36102 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36103 out of range value - has low bits known to be zero.
36105 2017-01-04 Alan Modra <amodra@gmail.com>
36107 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36108 * configure: Regenerate.
36109 * config.in: Regenerate.
36111 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36114 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36115 a substring of the message, but strcmp with the whole message. Ifdef
36116 ENABLE_NLS, translate the message first using dgettext.
36118 2017-01-03 Jeff Law <law@redhat.com>
36120 PR tree-optimizatin/78856
36121 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36122 (mark_threaded_blocks): Remove code to truncate thread paths that
36123 cross multiple loop headers. Instead invalidate the cached loop
36124 iteration information and handle case of a thread path walking
36125 into an irreducible region.
36127 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36130 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36131 assertions. Add support for doing the signbit if the IEEE 128-bit
36132 floating point value is in a GPR.
36133 * config/rs6000/rs6000.md (Fsignbit): Delete.
36134 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36135 Update the length attribute if the value is in a GPR.
36136 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36137 the sign or zero extension instruction, since the value is always 0/1.
36138 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36141 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36142 extracting SImode to a GPR register so that we can generate a
36143 store, limit the vector to be in a traditional Altivec register
36144 for the vextuwrx instruction.
36146 2017-01-03 Ian Lance Taylor <iant@google.com>
36148 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36150 2017-01-03 Martin Sebor <msebor@redhat.com>
36152 PR tree-optimization/78696
36153 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36154 precision. Use MPFR for %f for greater fidelity. Correct handling
36156 (pass_sprintf_length::compute_format_length): Set width and precision
36157 specified by asrerisk to void_node for vararg functions.
36158 (try_substitute_return_value): Adjust dump output.
36160 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36162 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36164 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36166 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36167 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36168 -mlra/-mno-lra was passed to the compiler.
36170 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36172 PR rtl-optimization/65618
36173 * emit-rtl.c (try_split): Move initialization of "before" and
36174 "after" to just before the call to emit_insn_after_setloc.
36176 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36178 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36180 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36182 * dwarf2out.c (gen_enumeration_type_die): When
36183 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36185 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36187 PR tree-optimization/78965
36188 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36189 Change first argument from const call_info & to call_info &. For %n
36190 set info.nowrite to false.
36192 PR middle-end/78901
36193 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36194 possibly throwing calls.
36196 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36197 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36198 and fns handling, rather than in a separate case SSA_NAME.
36200 2017-01-02 Jeff Law <law@redhat.com>
36202 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36203 fixes for first_period and second_period variables.
36205 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36208 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36209 (*insvqi_1): New insn pattern.
36210 (*insvqi_1_mem_rex64): Ditto.
36211 (*insvqi_2): Ditto.
36212 (*insvqi_3): Rename from *insvqi.
36214 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36216 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36218 * doc/cfg.texi (Edges): Remove reference to Java.
36219 (Maintaining the CFG): Ditto.
36221 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36223 PR middle-end/77674
36224 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36225 transparent aliases.
36227 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36229 PR middle-end/77484
36230 * predict.def (PRED_CALL): Update hitrate.
36231 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36232 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36233 into direct/indirect/polymorphic variants.
36235 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36237 Update copyright years.
36239 * gcc.c (process_command): Update copyright notice dates.
36240 * gcov-dump.c (print_version): Ditto.
36241 * gcov.c (print_version): Ditto.
36242 * gcov-tool.c (print_version): Ditto.
36243 * gengtype.c (create_file): Ditto.
36244 * doc/cpp.texi: Bump @copying's copyright year.
36245 * doc/cppinternals.texi: Ditto.
36246 * doc/gcc.texi: Ditto.
36247 * doc/gccint.texi: Ditto.
36248 * doc/gcov.texi: Ditto.
36249 * doc/install.texi: Ditto.
36250 * doc/invoke.texi: Ditto.
36252 Copyright (C) 2017 Free Software Foundation, Inc.
36254 Copying and distribution of this file, with or without modification,
36255 are permitted in any medium without royalty provided the copyright
36256 notice and this notice are preserved.